35775 A Country Environmental Analysis Publication ENVIRONMENT STRATEGY PAPERS NO. 9 Does the Treatment of Environmental Liability during Privatization Really Matter? An Empirical Evaluation in Central Europe Randall Bluffstone Theodore Panayotou December 2003 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Manufactured in the United States of America First printing December 2003 In 2001, the World Bank completed the comprehensive two-year process of preparing its Environment Strategy, Making Sustainable Commitments: An Environment Strategy for the World Bank. It was endorsed by the Bank's Board of Directors and published in October 2001. The Environment Strategy Paper series includes reports prepared to facilitate implementation of the Strategy. The Environment Strategy emphasizes the need to strengthen the analytical foundation of environmental work at the country level. Country Environmental Analysis (CEA) has been identified as one of the key environmental diagnostic tools for systematically evaluating the environmental priorities of development and poverty reduction strategies in client countries, the environmental implications of key policies, and countries' institutional capacity and performance to address their priorities. Executive Editor: Ede Ijjasz Managing Editor: Poonam Pillai Editor: Nancy Levine Designer / Production Manager: Jim Cantrell Contents vii Acknowledgments 1 Executive Summary 9 Chapter 1: Introduction 13 Chapter 2: Summary of Presumed Best Practices and Hypotheses to Be Tested 15 Chapter 3: Privatization and Environmental Liability Policies in the Countries Studied 17 Chapter 4: Description of the Sample and Data Collection Methodology 21 Chapter 5: Descriptive Statistics 25 Chapter 6: Econometric Results 27 Privatization Prices 31 Total Revenues from Privatization 33 Foreign Ownership 34 Efficiency of Privatization Processes 46 Site Remediation 50 Postprivatization Environmental Conflicts 54 Postprivatization Environmental Performance 63 Chapter 7: Conclusions Appendices 71 Appendix A: Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic 71 Hungary 73 Lithuania 74 Poland iii Does the Treatment of Environmental Liability during Privatization Really Matter? 76 Romania 78 Slovak Republic 81 Appendix B: Industries Included in the Sample 87 Appendix C: Survey Instrument 87 Definition 87 Section 1: Basic Firm Information 88 Section 2: Information about the Privatization Process 89 Section 3: Handling of Environmental Liabilities before Privatization 89 Section 4: Handling of Environmental Liabilities after Privatization 91 Appendix D: Regression Results Not Included in Chapter 6 91 Outcome 1 -- Privatization Prices 97 Outcome 2 -- Privatization Revenues 100 Outcome 4 -- Efficiency of Privatization Process 112 Outcome 5 -- Site Remediation 112 Outcome 7 -- Postprivatization Environmental Performance 117 Notes 121 Bibliography Boxes 71 Box A.1 Hungary: A low price for a river-contaminating plant 73 Box A.2 Lithuania: Sale of a fertilizer plant 75 Box A.3 Poland: Privatization of a paper producer 77 Box A.4 Romania: Economic and environmental improvements at a chemical plant with foreign investors Tables 4 Table ES.1 Summary of key policy findings 13 Table 2.1 Presumed best practices for environmental liability policies 17 Table 4.1 Summary of sample characteristics 17 Table 4.2 Sample size by country 18 Table 4.3 Industrial sectors included in sample 21 Table 5.1 Year of privatization of firms in the sample 23 Table 5.2 Government provision of environmental information in privatization packets, by country (percentage of cases) 23 Table 5.3 Environmental audits, by country (percentage of cases) 23 Table 5.4 Degree to which site remediation has been conducted, by country 27 Table 6.1 Conditioning variables used in regression equations to address outcomes 1 through 6 28 Table 6.2 Conditioning variables used in regression equations to address outcome 7 29 Table 6.3 Tobit regression. Dependent variable: Share price as a proportion of firm revenues in the year prior to privatization 32 Table 6.4 Tobit regression. Dependent variable: Privatization revenues in 1997 U.S. iv Environment Strategy Papers Contents dollars as a percentage of firm revenues in the year prior to privatization 35 Table 6.5 Tobit regression. Dependent variable: Time required to privatize, in months 37 Table 6.6 Tobit regression. Dependent variable: Number of tender rounds required to privatize firms 38 Table 6.7 Tobit regression. Dependent variable: Number of auction rounds required to sell at least 30 percent of firms' shares 39 Table 6.8 Tobit regression. Dependent variable: Number of tenders received in the round in which firms were privatized 41 Table 6.9 Tobit regression. Dependent variable: Percentage of all shares purchased from the government at the time of privatization 43 Table 6.10 Tobit regression. Dependent variable: Time required to privatize, in months 45 Table 6.11 Tobit regression. Dependent variable: Number of tender rounds required to privatize 47 Table 6.12 Binomial logit regression. Dependent variable: Whether any site remediation had been conducted since privatization (dummy variable) 49 Table 6.13 Binomial logit regression. Dependent variable: Whether any site remediation had been conducted since privatization (dummy variable) 51 Table 6.14 Binomial logit regression. Dependent variable: Whether administrative enforcement actions had been brought by the government against the firm (dummy) 53 Table 6.15 OLS regression. Dependent variable: Number of environment-related legal actions brought by the government against firms since privatization 55 Table 6.16 OLS regression. Dependent variable: Change in wastewater emissions, 1994­97, as a proportion of firm revenues in 1997 57 Table 6.17 Binomial logit regression. Dependent variable: Whether firms had intro- duced internal environmental audits, waste minimization, or pollution prevention programs (dummy) 58 Table 6.18 Binomial logit regression. Dependent variable: Whether firm had an environmental plan in place (dummy) 60 Table 6.19 Binomial logit regression. Dependent variable: Whether firm management reported seeking iso 14000 certification (dummy) 78 Table A.1 Privatization methods used during the second wave of privatization in the Slovak Republic 91 Table D.1 Tobit Regression. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 92 Table D.2 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 92 Table D.3 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 93 Table D.4 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 94 Table D.5 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 94 Table D.6 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Bluffstone and Panayotou v Does the Treatment of Environmental Liability during Privatization Really Matter? 95 Table D.7 Tobit Regression. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 96 Table D.8 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization 96 Table D.9 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization 97 Table D.10 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization 97 Table D.11 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization 98 Table D.12 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of 1990 Firm Revenues 99 Table D.13 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of 1990 Firm Revenues 100 Table D.14 Ordinary Least Squares. Dependent Variable: Time Required to Privitize, in Months 100 Table D.15 Tobit Regression with Heteroskedasticity (as a Function of Firm Revenues in 1997) Correction. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm 101 Table D.16 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm 102 Table D.17 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Round in Which Firms Were Privatized 103 Table D.18 Ordinary Least Squares. Dependent Variable: Percentage of All Shares Purchased from the Government at the Time of Privatization 103 Table D.19 Tobit Regression. Dependent Variable: Time Required to Privatize, in Months 104 Table D.20 Ordinary Least Squares. Dependent Variable: Time Required to Privatize, in Months 105 Table D.21 Ordinary Least Squares. Dependent Variable: Time Required to Privatize, in Months 105 Table D.22 Tobit Regression with Heteroskedasticity Correction. Dependent Variable: Number of Tender Rounds Required to Privatize 106 Table D.23 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize 107 Table D.24 Tobit Regression. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized 108 Table D.25 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized 109 Table D.26 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Round in Which Firms Were Privatized 110 Table D.27 Ordinary Least Squares. Dependent Variable: Number of Auction Rounds Required to Sell at Least 30 Percent of the Firm's Shares 111 Table D.28 Ordinary Least Squares. Dependent Variable: Number of Auction Rounds Required to Sell at Least 30 Percent of the Firm's Shares 112 Table D.29 Binomial Logit. Dependent Variable: Whether Any Site Remediation Has Been Conducted since Privatization (Dummy Variable) 112 Table D.30 Ordinary Least Squares. Dependent Variable: Change in Particulate Emis sions, 1994­97, as a Proportion of Firm Revenues in 1997 113 Table D.31 Ordinary Least Squares. Dependent Variable: Change in Particulate Emis sions, 1994­97, as a Proportion of Firm Revenues in 1997 114 Table D.32 Ordinary Least Squares. Dependent Variable: Change in Tons of Particulate Emissions, 1994­97, per Dollar of Revenue Generated by Firms in 1997 114 Table D.33 Binomial Logit. Dependent Variable: Whether Firm Had an Environmental Plan in Place (Dummy) 115 Table D.34 Binomial Logit. Dependent Variable: Whether Firms Were Seeking ISO 14000 Series Certification (Dummy) vi Environment Strategy Papers Acknowledgments T his study was prepared as part of a well as the contributions of the organizations larger effort at the Environment that implemented the data collection effort Department of the World Bank to underlying this paper: TARKI (Hungary); the learn from good practices and provide policy Center for Environmental Policy and the advice to developing and transition econo- research firm SIC (Lithuania); the Institute of mies for improving their environmental Economics, Polish Academy of Science management frameworks and practices. The (Poland); the Center for Environmentally study was managed by Magda Lovei, lead Sustainable Economics Policy (Romania); and environmental economist in the Environment Incoma (Slovak Republic). Comments by Department and team leader of the Environ- Roberto Pedace and other participants in ment Strategy team. University of Redlands forums are gratefully acknowledged. The report was prepared by Randall Bluffstone and Theodore Panayotou. Randall Bluffstone The authors are grateful to Susmita Dasgupta is with the Department of Economics, Port- for extensive peer review, as well as to Milen land State University. Theodore Panayotou is Dyoulgerov, Stefan Schwager and many others director of the Environment and Natural for constructive comments. The initial results Resources Program, Center for International of the study were discussed with World Bank Development at Harvard University, John F. staff at an Environment Strategy Roundtable Kennedy School of Government. on July 24, 2002. The authors acknowledge the support of the Thanks are due to the government of Norway, U.S. Agency for Environmental Development which provided financial assistance for (USAID) in funding the data collection, as completing the study. vii Executive Summary S ince the late 1980s, the countries of method of transferring ownership to the private Central Europe have been working to sector. In Poland early industrial privatizations reform their economies and reorient relied heavily on liquidation, but vouchers, them to compete in 21st century markets. combined with 15 investment funds, were also Privatization is a particularly important part of used. Hungary, by contrast, focused on capital this process, and countries have used a variety privatization to investors, selling firms outright of privatization methods. Simultaneously, through such methods as tenders, auctions, and Central European countries are having to public stock offerings. struggle with severe environmental problems left over from previous regimes. Site contami- An important aspect of the relationship be- nation is an important challenge, particularly in tween privatization and site contamination is industries such as chemicals, oil refining, that when investors purchase industrial firms, smelting, and waste oil processing. they face the risk of acquiring major environ- mental liabilities. The concern that environ- This report examines whether Central Euro- mental liability poses for investors, and particu- pean countries have applied the generally larly foreigners, has long been discussed. For accepted best practices for handling potential example, in a study of 1,000 large North environmental liabilities in the course of American and Western European firms con- privatization and, if so, whether these practices ducted in 1992, half of the firms that actually have yielded the expected results. Five coun- evaluated sites in Central European countries tries are surveyed: Hungary, Lithuania, Poland, reported rejecting the sites partly on environ- Romania, and the Slovak Republic. These mental grounds. Liability for past practices and countries' privatization programs have some for inherited contamination risks ranked as the features in common--most importantly, the most important environmental concern among goal of transferring whole economies to private the surveyed firms. hands as quickly as possible--but the ap- proaches have been very different. In Avoiding complications in privatization Lithuania, Romania, and the Slovak Republic processes is, of course, not the only goal. voucher mass privatization was the main Cleanup and improvements in ongoing envi- 1 Does the Treatment of Environmental Liability during Privatization Really Matter? ronmental performance may be necessary to 2. Enterprises that were sold using mass avert serious current and future risks to human privatization techniques were at least 30 health and the environment. As ownership of percent owned by one person, firm, or enterprises and sites is transferred to private consortium by July 1998. hands, most would agree that it is important to promote site remediation when appropriate From the descriptive statistics, certain observa- and to take steps to reduce ongoing air and tions can be made. First, firms in so-called water pollution. The literature suggests that "dirty" industries tend to be large, with high certain environmental liability policies, such levels of employment. Most investors in these as provision of information, analysis of firms purchased majority shares at the time of environmental issues before privatization, and privatization. Few were foreigners; only 25 incentives to investors in the shape of dis- percent of firms indicated that foreign inves- counts, indemnification, and sharing of tors had any stake at all. In general, govern- cleanup responsibilities by the state, will ments have maintained significant stakes in encourage better economic and environmen- these firms, with an average of 35 percent tal outcomes than others. government ownership remaining after privatization. The analysis in this report relies on firm-level data from 216 industrial firms in the five Privatization of firms in the selected industries countries, distributed as follows: Hungary, 29 generally took less than a year. Only rarely firms; Lithuania, 50; Poland, 64; Romania, 54; did privatization agency officials report that and Slovak Republic, 19. The industries environmental issues delayed privatization. It examined were ones that could reasonably be expected to have site contamination problems. is also unlikely that preprivatization manage- The primary data sources were privatized ment teams erected roadblocks to companies, but information was also gathered privatization, since current management from privatization agencies, ministries of usually did not believe that these teams were environment, and statistical agencies. hostile to the process. Despite this lack of opposition, management changes were Links between environmental liability policies, common in the first year after privatization. privatization outcomes, and postprivatization performance are likely to exist when owners Information on firms was generally provided are truly able to control their firms and make to potential investors, and in one-third of those necessary changes. The literature suggests that cases environmental information was in- such control occurs when ownership is cluded. Information specifically focusing on relatively concentrated. For firms to be site contamination was offered less frequently. included in the sample, postprivatization Instead of relying on government-provided ownership had to meet one of the following information, almost half of the investors conditions: financed preliminary audits, and 14 percent paid for secondary audits. Although indemnifi- 1. Majority interests were sold outright to cation is sometimes advocated as a useful risk individuals, firms, or consortia at the time reduction tool, in only a few cases did of privatization. privatization agencies do anything but pass 2 Environment Strategy Papers Executive Summary 100 percent of environmental liabilities to Privatization prices. The inclusion of an new owners. Explicit price discounts were environmental section in the information rarely given, implying that only limited overt packets given to investors was associated with consideration was given to the risk of inves- increased privatization prices, measured as tors' buying into industries with potentially share price as a proportion of revenues in the serious environmental problems. Investors year prior to privatization. Compared with the seem to have been left on their own to find mean, the provision of environmental informa- out about any site contamination problems tion by governments was associated with an and to negotiate with privatization agencies as increase in privatization prices of 173 percent. best they could. There also appears to be an autonomous trend toward lower prices, with a year's delay in A significant, but not overwhelming, degree of privatization estimated to reduce prices by 50 site remediation appears to have occurred in to 75 percent. This finding supports the study sample firms. Over 20 percent of firms by Bluffstone and Panayotou (2000) suggesting reported being engaged in some type of site that the quality of privatized firms should remediation. Firms also exhibit a variety of decline over time. Firms in the metal manufac- other positive environmental behaviors such turing industry were heavily discounted, as having internal environmental management yielding lower prices. systems and environmental plans and seeking ISO 14000 series certification. On average in Privatization revenues. The most dramatic the sample, emissions of wastewater and key results occurred when environmental informa- pollutants declined significantly during the tion was provided to investors and was period 1994­97. followed up with site remediation planning. When both were done, privatization revenues The regression analysis offers results that were estimated to triple. Metallurgical compa- generally validate the presumed best practices nies were heavily discounted. Firms in (see table ES.1). Seven outcomes were of Hungary and Poland brought four times lower interest in the analysis: revenues, compared with the mean, suggest- ing that privatization agencies in those 1. Privatization prices countries have not received top prices or sold 2. Privatization revenues received by large blocks of shares of industrial firms. privatization agencies 3. Ownership by foreigners Foreign ownership. Foreign investors did not 4. Efficiency of privatization processes appear to respond to provision of environmen- 5. Degree of postprivatization site remediation tal information by governments. There was no 6. Extent of postprivatization conflicts, such as relation between the provision of environmen- lawsuits and administrative actions tal information and the percentage of foreign 7. Ongoing postprivatization environmental ownership in firms, nor was any relation performance, measured using various between the provision of environmental pollution and environmental indexes. information and site remediation observed. Bluffstone and Panayotou 3 Does the Treatment of Environmental Liability during Privatization Really Matter? Table ES.1 Summary of key policy findings Estimated relationship with outcome (comparisons are Outcome of interest Policy or other variable with means) Issues of importance to ministries of economy and privatization agencies Prices for privatized firms Provision of environmental information Increase (173 percent) Privatization of metallurgical firms Decrease (70 percent) Privatization revenues Provision of environmental information Increase (200 percent) and development of site remediation plans Location of company in Poland or Decrease (300 percent) Hungary Foreign ownership in Provision of environmental information Neutral privatized firms and development of site remediation plans Time required to privatize Provision of environmental information Increase (4 months; 50 percent) firms (months) Promotion of foreign ownership Marginal increase Development of site remediation plans Decrease Provision of environmental information Decrease (6.5 months; 83 and development of site remediation percent) plans Location of company in Poland or Increase (4 months; 50 percent) Hungary Number of tender rounds Provision of environmental information Increase (1.5 rounds; 45 needed to privatize percent) Provision of environmental information Decrease (2.73 rounds; 170 and development of site remediation percent) plans Issues of importance to ministries of environment Site remediation after Preliminary audits Increase in likelihood (67 privatization percent) Development of site remediation plans Increase in likelihood (120 percent) Management turnover after Double likelihood privatization Location of company in Poland or Increase in likelihood (76 Hungary percent) Likelihood of environmental Provision of environmental information Increase in likelihood (33 administrative actions against percent) firms Number of environmental Provision of environmental information Increase (200 percent) legal actions against firms Pollution emissions Provision of environmental information Neutral Development of site remediation plans Neutral 4 Environment Strategy Papers Executive Summary Table ES.1 Summary of key policy findings (continued) Estimated relationship with outcome (comparisons are Outcome of interest Policy or other variable with means) Positive environmental Provision of environmental information Increase in likelihood (20­40 behaviors (e.g., internal percent) environmental management systems, environmental plans, ISO 14000 certification) Development of site remediation plans Neutral Efficiency of privatization processes. It is Privatization is estimated to take 83 percent sometimes said that privatization agencies less time, and fewer tenders are required (170 may be reluctant to give investors environ- percent lower than the mean). Looking at the mental information or to encourage them to full context, then, it appears that providing investigate sites because of concern that environmental information at the time of privatization processes will be slowed. The privatization does indeed speed privatization, literature, however, suggests that providing but for firms with contamination problems, more information will facilitate privatizations this information must be accompanied by by reducing risk. In our study a number of concrete plans for addressing the problems. measures indicate that when environmental Without such plans, the information tends to information was included in privatization delay privatization. packets, privatization took systematically longer, and smaller share blocks were sold. Site remediation. A particularly important reason for providing site contamination and One interpretation of these results is that when other environmental information to investors site contamination issues are raised, time is and for developing site remediation plans is to required for due diligence. Alternatively, the facilitate site remediation after privatization. In causality may be reversed if environmental practice, concerted efforts are required to spur information tends to be provided for firms site remediation. Whereas the provision of with more environmental problems that would environmental information in privatization in any case take longer to sell. Finally, it is packets is unrelated to site remediation, possible that nonprovision of environmental preliminary and secondary audits are corre- information is really a proxy for insider lated with increases in the probability of privatization. It seems likely that insider deals postprivatization remediation of about 67 would require no environmental information percent, compared with the mean probability. and would be quicker than other modes of The effect of site remediation planning property transfer. appears to be even more important. For firms that had developed site remediation plans The combination of provision of environmen- before or shortly after privatization, the tal information and a site remediation plan is probability that site remediation occurred was associated with more rapid privatization. 120 percent greater than for the average firm. Bluffstone and Panayotou 5 Does the Treatment of Environmental Liability during Privatization Really Matter? The results suggest that audits and remediation lems. Heightened enforcement may therefore planning dramatically increase the chance that not be a problem. remediation will occur. Investors financed most audits, and firms probably prepared most Postprivatization environmental performance. remediation plans. We do not know what Our data on pollution emissions are rather factors motivated the decision to follow up limited, and we therefore hesitate to draw information gathering with remediation, but it strong conclusions, but there does not appear seems reasonable to suppose that firms were to be a very strong relationship between acting in their own interests, motivated by a pollution after privatization and provision of desire to improve the work environment or to environmental information before protect themselves from potential future privatization. Environmental information was, liabilities. Many firms, of course, did not however, positively associated with some remediate. The apparently strong link between environmental behaviors that in the future environmental audits, site remediation plan- could yield results. For example, in compari- ning, and site cleanup suggests that it would son with the mean, provision of environmen- probably be in the interests of Central Euro- tal information in privatization packets was pean societies to finance environmental audits associated with a 29 percent increase in the and preprivatization development of site probability that firms had environmental audit remediation plans for firms where cleanup and other internal environmental management programs in place after privatization. It was seems desirable. also associated with a 43 percent increase in the chance that firms had environmental plans Postprivatization conflicts. There is some in place. evidence that governments have utilized the environmental information provided to Despite some caveats, our empirical results in investors for enforcement purposes. Compared general strongly support the theoretical with the mean probability, the provision of literature on best practices for addressing environmental information in privatization environmental liabilities at the time of packets was associated with an increase of privatization. The provision of environmental about 33 percent in the probability of environ- information at the time of privatization mental administrative actions such as notices appears to be associated with dramatic of violations, fines, and closure. Similarly, increases in privatization prices and, when when written information on potential site combined with site remediation planning, contamination was provided to most or all with very large increases in privatization investors, the number of environment-related revenues and quicker privatizations. Provision legal actions against firms after privatization of environmental information alone is corre- roughly tripled. Although it may be that lated with positive environmental behaviors regulators focused more on firms for which after privatization but seems to be associated they had information, the firms in the sample with longer privatization times. Greater use of are all in generally highly polluting industries, investigative instruments, such as audits and and information was more likely to have been site remediation planning, appears to be provided for firms with more serious prob- associated with the remediation of contami- 6 Environment Strategy Papers Executive Summary nated sites. The process of generating informa- to investigate environmental issues, bring tion seems to assist governments in their them to light, and work with investors to environmental enforcement efforts. Given address them vigorously. The evidence these impressive benefits, for firms that are suggests that doing so will yield economic as likely to have or are known to have environ- well as environmental benefits. mental problems, governments would do well Bluffstone and Panayotou 7 Chapter 1 Introduction S ince the late 1980s, the countries of examples include chemicals, oil refining, Central Europe have been reforming smelting, and waste oil processing. their economies and reorienting them Privatization and site contamination are to compete in 21st century markets. related because when investors buy into such Privatization is a particularly important industries, they simultaneously face the risk of instrument in this process. Several purchasing major liability problems. As Boyd privatization vehicles have been utilized in (1996) correctly states, the policy problem is these countries, including direct sales to to identify instruments that allow the identifi- strategic investors, auctions, voucher sales, cation and allocation of site remediation costs and management or employee buyouts. After in ways that minimize distortions in a decade of experience, it is becoming clear privatization, in site remediation, and in that issues of corporate governance are future safety decisions. crucial. Unless steps are taken to create incentives for efficiency and to ensure that That environmental liability is a concern of infusions of expertise and finance accompany investors, and particularly foreigners, has transfers of ownership, the true goal of been long discussed and continues to be an privatization--improved enterprise perfor- issue (Lovei and Gentry 2002). In a study of mance--may remain elusive (Frydman and 1,000 large North American and Western Rapaczynski 1994; Gray 1996; World Bank European firms conducted in 1992, of those 2002). that actually evaluated sites in Central Euro- pean countries, half reported rejecting them Simultaneously with privatization, countries partly on environmental grounds. Liability for have to deal with often severe environmental past practices and for inherited contamination problems left over from previous regimes. Site risks ranked as the most important environ- contamination is one important challenge. As mental concern among surveyed firms Goldenman (1993) has pointed out, these (Klavens and Zamparutti 1995). contamination problems are likely to be concentrated in particular industries that are It is not known whether foreign investment generally thought to be highly polluting; has indeed been lost due to environmental 9 Does the Treatment of Environmental Liability during Privatization Really Matter? factors, or how much has been lost, but tion, is simply a risk tax on privatization anecdotal evidence from the region suggests markets, and information is believed to be that poor handling of possible environmental perhaps the most powerful instrument for liabilities certainly complicates foreign reducing risk (Bluffstone and Panayotou investment and may have deterred it (Auer, 2000; Lovei and Gentry 2002). Such informa- Reuveny, and Adler 2001). Dodds and tion can take many forms but presumably Wächter (1993), at the beginning of their begins with the inclusion of basic environ- discussion of environmental issues, observe, mental information in packets prepared for "Prospective purchasers looking at their first potential investors. Preliminary audits may acquisition on the other side of the Wall may be conducted if they seem warranted, and be inclined to panic when the issue of secondary audits that include sampling and environmental damage first comes up . . . analysis may also be used. Current thinking clearly there are concerns to address when on best practices seems to suggest that buying an industrial facility that meets 1950s provision of information by privatization standards when 1990s rules apply." agencies is important, as is offering investors the opportunity to investigate for themselves. Avoiding complications in privatization is, of course, not the only goal. Cleanup and Although investing in information is believed improvement in ongoing environmental to reduce risks to investors and to allow performance may be necessary if serious privatization markets to work more effi- current and future risks to human health and ciently, environmental analysis is necessary the environment are to be mitigated or if privatization agencies are to take control of avoided. Most would agree that as ownership cleanup decisions. The German privatization of enterprises and sites is transferred to agency, the Treuhandanstalt, and its succes- private hands, it is important that policies sor agency, the Bundesanstalt für promote site remediation when appropriate. vereinigungsbedingte Sonderaufgaben, used In reality, however, continuing air and water environmental audits with particular effec- pollution is likely to be much more impor- tiveness to keep costs down. In the early tant than site contamination for human 1990s cleanup costs in eastern Germany health. The EAP Task Force (1998) noted that were expected to run to hundreds of billions nonferrous metallurgy, iron and steel, pulp of dollars. By 1996, with privatization and paper, and chemicals have the potential virtually complete, only US$6.4 billion had to be especially problematic. The infusion of been spent, largely because remediation financial and human capital that activities were prioritized exclusively on the privatization provides offers an opportunity basis of health risks.1 to address both site contamination and ongoing pollution issues (Goldenman 1997; Once an estimate of environmental liabilities Auer, Reuveny, and Adler 2001; Lovei and is made, responsibility for settling them can Gentry 2002). be either transferred to investors or retained by privatization agencies. If the former Uncertainty about future liabilities, whether strategy is chosen, investors might be ex- for site contamination or for ongoing pollu- pected to include in their price bids a risk 10 Environment Strategy Papers Introduction premium for the uncertainty involved ing environmental liabilities is a divergence (Balaban, Bluffstone, and Panayotou 1994; of interests and an asymmetry of information Boyd 1996; Goldenman 1997).2 Alterna- between enterprise managers and the two tively, governments might retain the respon- transaction parties--privatization agencies sibility for past polluting activities. As Lovei and investors. Indeed, as argued by Frydman and Gentry (2002) and Goldenman (1993) and Rapaczynski (1994), Central European have discussed in detail, contractual instru- governments cannot even be considered sole ments for doing this include indemnifications owners of state enterprises because managers and releases. Indemnifications allow and workers also claim de facto and even de privatization agencies to reimburse investors jure rights to firms. Managers may want to for specific cleanup activities necessary to keep enterprises in state hands to preserve remediate sites. In the Czech Republic and their own positions, or they may hope to Hungary, for example, individual escrow purchase enterprises themselves. They may accounts and earmarked funds were used to also have concerns about investor quality finance cleanup (Lovei and Gentry 2002).3 A and intentions. release is functionally similar except that the privatization agency is responsible for This paper examines the relationship be- arranging site remediation. tween environmental liability policies at the time of privatization and the economic and As Bluffstone and Panayotou (2000) show in environmental outcomes observed in Central a theoretical model, where there is symmet- European countries. Chapter 2, next, reviews ric information between privatization agen- the environmental liability policies that are cies and investors, price discounts and believed to be preferred. Chapter 3 surveys indemnifications are equivalent instruments. privatization and environmental liability Privatization agencies should prefer price policies in the five countries examined: discounts, which shift the risk to investors. Hungary, Lithuania, Poland, Romania, and Only when governments have better informa- the Slovak Republic. The data collected from tion about firms than investors, or the quality firms in these countries are presented in of the privatization pool is systematically chapter 4, and chapter 5 gives descriptive deteriorating (as the authors suggest should statistics from the data. Chapter 6, the heart occur), do indemnifications become a useful of the paper, contains a series of economet- policy tool. Under such circumstances, ric analyses of the relationship between indemnifications may help overcome infor- various environmental liability policies and mation problems and keep investors from environmental and economic outcomes. excessively discounting firms. Chapter 7 presents conclusions. It is believed that an important factor compli- cating privatization and policies for address- Bluffstone and Panayotou 11 Chapter 2 Summary of Presumed Best Practices and Hypotheses to Be Tested A s was noted in chapter 1, the litera- economic and environmental outcomes than ture suggests that certain environmen- others. The main categories of superior policies and tal liability policies offer better expected outcomes are summarized in table 2.1. Table 2.1 Presumed best practices for environmental liability policies Environmental liability policy Expected outcomes State invests in information State has information that can then be provided to regarding potential environmental investors, reducing their risks; state is better liabilities and provides that informed for its negotiations with investors, information to investors. Investors resulting in higher privatization prices; higher- also have the opportunity to quality investors are attracted; state and firms can investigate environmental issues better plan for environmental remediation; more themselves. site cleanup actually occurs. Preprivatization environmental Postprivatization conflicts regarding remediation problems are inventoried and are reduced; more remediation takes place. differentiated from problems occurring after privatization. On-site and off-site remediation Postprivatization conflicts are reduced; more Increasing requirements and responsibilities remediation actually takes place; levels of sophistication are clearly defined and allocated remediation are well known, and both firms and between the state and investors. the state are able to budget for these costs. States accept financial Risks for investors are reduced if investors know responsibility for a share of they are not financially responsible for remedying remediation. (Some authors problems they did not cause; investors are better suggest that clearly defined able to evaluate firms; higher-quality investors are indemnifications rather than price attracted; privatization prices are higher than when reductions be used.) there is more risk; firms hold the state responsible for financing cleanup, and therefore more remediation occurs. Future environmental Higher-quality investors are attracted; risks for performance is improved through investors are reduced if investors know what their better handling of environmental ongoing environmental responsibilities are; liabilities in privatization contracts. investors are better able to evaluate firms; ongoing Ongoing pollution is included in environmental performance is improved. privatization contracts. 13 Does the Treatment of Environmental Liability during Privatization Really Matter? In this paper we examine empirically whether mental performance and is negatively the presumed best practices have been used in correlated with postprivatization conflicts Central Europe and, if so, whether they and time required for privatization. yielded the expected results. The specific ! Clear delineation of responsibilities and hypotheses to be tested are as follows: planning for environmental remediation are positively correlated with privatization ! Access to more information on the environ- prices, privatization revenues, degree of mental status of firms by the state and site remediation, and better ongoing investors is positively correlated with environmental performance and are privatization prices, privatization revenues, negatively correlated with postprivatization degree of foreign ownership, degree of site conflicts and time required for remediation, and better ongoing environ- privatization. 14 Environment Strategy Papers Chapter 3 Privatization and Environmental Liability Policies in the Countries Studied T he privatization programs in Hungary, analyze what worked and which policies were Lithuania, Poland, Romania, and the less successful, and indeed they provide the Slovak Republic share some features, variance necessary for the empirical work the most important of which is the goal of presented in subsequent chapters. Appendix A transferring whole economies to private hands presents details about the policies and ap- as quickly as possible. All the programs also proaches in place at various times in the have in common the tendency for larger state- countries under study. owned enterprises to be broken up into component parts for privatization. As a result, It is important to point out that in 1990­91 the the number of entities created by privatization countries analyzed in this paper had limited has always been higher than the number of laws and policies for addressing potential firms slated for privatization. environmental liabilities during the process of property transfer. It is probably also reasonable Very different approaches to privatization are to say that they did not necessarily appreciate evident. In Lithuania, Romania and the Slovak the importance of linking environmental and Republic, voucher mass privatization was the privatization issues. In the absence of specific main way ownership was transferred to the policies, sometimes environmental liabilities private sector. In Poland early industrial were only implicitly passed to investors. Often, privatizations made heavy use of liquidation, privatization was carried out with the tacit but vouchers, combined with 15 investment understanding on both sides that contingent funds, were also used. Hungary, by contrast, liabilities were unlikely ever to be manifested. focused mainly on a capital privatization program (Goldenman 1997). Mass privatization programs, in particular, ignored environmental issues. Environmental Views regarding privatization and the treatment assessments were rarely done before transfer of of environmental liabilities differed dramati- ownership (which was generally accomplished cally across countries and evolved over time. via vouchers), and the private citizens who at These differences offer an opportunity to least initially owned the companies had little 15 Does the Treatment of Environmental Liability during Privatization Really Matter? reason to invest in information themselves or unknowingly taken on. Requirements to settle to incorporate environmental liabilities into environmental liabilities are likely to come their voucher investment decisions. Further- from two sources: Central European govern- more, at the time of privatization no require- ments will probably demand site cleanup as ments concerning cleanup of contaminated part of better environmental management in sites or other settling of environmental liabili- the future, and governments and third parties ties were imposed or included in agreements may seek settlement of liabilities under civil with investors. This is perhaps not surprising: law. Civil law codes almost universally no funds were generated during the include provisions for successor liability, privatization process, and so additional which means that private owners could find resources were not made available to often themselves responsible for environmental cash-strapped firms. liabilities even if they did not investigate such problems and were not informed about them. It is difficult to argue that environmental Since there is little reason to expect that liabilities hampered mass privatization, but it environmental liabilities were in any way is also probable that environmental problems affected or addressed by mass privatization, were simply put off to the future, when the focus in this paper is on capital existing or successor owners would be forced privatization, in which firms are sold outright to face the liabilities that firms had perhaps via tenders, auctions, or public stock offerings. 16 Environment Strategy Papers Chapter 4 Description of the Sample and Data Collection Methodology T he sample was selected specifically to Table 4.1 Summary of sample characteristics answer the research questions posed in Sample size 216 (number of firms) this report and to test the two classes of hypotheses. A wide cross-section of countries Data collection period July­September 1998 and industries was included in the analysis to Countries Hungary, Lithuania, Poland, Romania, Slovak increase the generality of the findings. Republic Privatization characteristics were chosen in such a way that firm owners were likely to Industrial sectors Animal raising, mining, electrical power, have the incentives and the opportunity to manufacturing respond to environmental liability policies. We therefore believe that the sample is representa- Privatization (1) Majority interest was characteristics sold outright, or (2) tive of the population of firms for which enterprises sold by mass environmental liability policies are likely to be privatization were at most important. least 30 percent owned by one investor or consortium The analysis relies on firm-level survey data from over 200 firms in Hungary, Lithuania, Poland, Romania, and the Slovak Republic. Table 4.2 Sample size by country Sample characteristics are presented in table Sample size 4.1 and sample sizes by country in table 4.2. Country (number of firms) Data were collected during July­September Hungary 29 1998 by professional research organizations Lithuania 50 Poland 64 that were headquartered in the countries of Romania 54 study and either had substantial experience in Slovak Republic 19 environmental economics research or special- Total 216 ized in survey implementation.4 The sample was made up of firms operating in Those considering investing in firms included industries that are generally highly polluting. in the sample could therefore reasonably be 17 Does the Treatment of Environmental Liability during Privatization Really Matter? concerned about potential environmental ownership also had to be relatively concen- liabilities due to past off-site and on-site trated. This stratification was made because contamination.5 The four sectors examined we wanted to include only cases in which were agriculture (animal raising only), mining, owners were truly able to control firms and to electrical power, and manufacturing. A list of respond to environmental liability policies industry sectors is provided in appendix B. (see, among others, Gray 1996). "Real Firms in the sample were chosen by in- privatization" is defined in our sample as country research firms. The firms had to be in occurring if, as of July 1998, firms had the the selected industries and in operation in July following characteristics: 1998, when the survey was undertaken. (Sometimes, countries had only one or two ! Majority interests were sold outright to firms in the targeted industries.) For a firm to individuals, firms, or consortia at the time be included in the sample, postprivatization of privatization Table 4.3 Industrial sectors included in sample Percentage of all Industrial sector NAICS code Number of firms firms in the sample Animal raising 112 7 3.24 Oil and gas extraction 211 1 0.46 Mining 212 2 0.93 Electrical power 221 2 0.93 Food manufacturing 311 20 9.26 Beverage and tobacco 312 9 4.17 Textile mills 313 11 5.09 Textile products 314 6 2.78 Apparel manufacturing 315 15 6.94 Leather products 316 5 2.31 Wood products 321 3 1.39 Paper manufacturing 322 5 2.31 Printing 323 1 0.46 Petroleum and coal product manufacturing 324 3 1.39 Chemical manufacturing 325 21 9.72 Plastics and rubber products 326 2 0.93 Nonmetallic mineral products 327 12 5.56 Primary metal manufacturing 331 7 3.24 Fabricated metal manufacturing 332 9 4.17 Machinery manufacturing 333 25 11.57 Computer and electronic products 334 9 4.17 Electrical equipment, appliances, and components 335 4 1.85 Transportation equipment manufacturing 336 12 5.56 Furniture and related product manufacturing 337 6 2.78 Miscellaneous manufacturing 339 13 6.02 No industry reported 5 2.31 Note: NAICS, North American Industrial Classification System. Boldface type indicates industries that are particularly likely to exhibit site contamination problems. 18 Environment Strategy Papers Description of the Sample and Data Collection Methodology ! Enterprises that were sold using mass agencies, were utilized. Data were collected privatization techniques were at least 30 on the following topics: percent owned by one person, firm, or consortium by July 1998. ! Basic firm characteristics (e.g., year of privatization, export orientation, revenues, As shown in table 4.3, a variety of industries foreign ownership) that could potentially have environmental ! Firm environmental management (existence liability problems are represented in the of an environment department, environ- sample. Sectors that are especially likely to mental permits for operation, monitoring of have problems are shown in boldface type. emissions) ! The privatization process (type of The questionnaire that was used is attached as privatization, number of auction or tender appendix C. The primary data sources for the rounds needed to privatize, information study were the privatized companies them- provided to investors) selves. Survey enumerators set up appoint- ! Handling of environmental liabilities during ments and went in person to conduct the privatization (provision of information survey. Often, they were forced to return a about site contamination, conduct of audits, second time to collect additional data. In development of remediation plans, price many cases not all data could be obtained reductions) from firm management, and other data ! Treatment of environmental liabilities after sources, such as privatization agencies, privatization (site remediation, enforcement ministries of environment, and statistical actions, legal actions against the firm). Bluffstone and Panayotou 19 Chapter 5 Descriptive Statistics T he median enterprise in the sample was year prior to privatization, with a mean value privatized in 1995; the earliest of 18 percent higher. privatization was in 1990 (table 5.1). Virtually all Romanian firms in the sample Foreign participation in the sampled firms was were privatized in 1998, substantially raising the exception in 1997, with 76 percent of firms the average year. Over 79 percent of firms in having no foreign ownership and an average the sample (170 firms) were privatized by foreign ownership share of 12 percent. Thir- outright purchase, and 21 percent (46 firms) teen percent of the firms had majority foreign were privatized through accumulation of over ownership. Average government ownership in 30 percent of shares by July 1998. Of the firms the sample in 1997 was 35 percent; 28 percent privatized by outright purchase, 63 percent of the firms had no government ownership. In were privatized using tenders, 29 percent by 75 percent of the cases, over 49 percent of the auction, and 8 percent by other means. shares owned by the government were sold. Table 5.1 Year of priva- On average, 70 percent of the government tization of firms in the Mean employment shares were sold at the time of privatization. sample in the sample was Twenty percent of the firms in the sample Decile Year 1,264 in the year Minimum 1990 prior to privati- transferred 100 percent of shares to investors at 10 1992 zation and 1,027 in the time of privatization. 20 1992 1997. Firms in the 30 1993 sample were For the 99 cases in which tenders were used, 40 1994 therefore generally the mean number of rounds required to Median 1995 large but exhibited privatize the firm was 1.6, with a maximum of 60 1996 70 1997 significant attrition 5 rounds reported. Half of the cases took only 75 1998 over time. Output one round. The mean number of bidders in the 80 1998 of firms' main tender in which firms were privatized was 2.6, 90 1998 products was but 30 percent of tenders had only 1 bidder, Maximum 1998 generally higher in and only 20 percent of firms had more than 4 Note: Number of firms in sample, 216. 1997 than in the bidders. 21 Does the Treatment of Environmental Liability during Privatization Really Matter? Among firms sold outright to investors, Typically, however, written information privatization took, on average, about 9 focusing explicitly on possible site contamina- months. Fifty percent of sales took 7 months tion was not provided to investors (see table or less, and 30 percent took over 10 months. 5.2). Indeed, only 21 percent of 187 firms said The maximum time was 72 months. Environ- that such information was provided to most or mental issues were in general not reported to all investors. Perhaps because of this lack of be an important determinant of the length of written information, 47 percent of firms time needed to privatize. In only 7.3 percent reported that preliminary environmental audits of cases did privatization officials report that were conducted prior to privatization (see environmental issues delayed privatization, table 5.3 for country data). Of the firms suggesting that environmental issues were reporting sources of finance for preliminary handled either efficiently or not at all. audits (the choices given on the questionnaire were investors, privatization agency, both According to firm management, the previous investors and privatization agency, and other management was in general not hostile to organization), 70 percent said that investors privatization. Only 3 out of 202 firms said that financed audits, while only two firms said that the former team was "probably" or "definitely" privatization agencies financed their audits. hostile to privatization. By contrast, 190 Fourteen percent of firms conducted second- respondents reported that preprivatization ary audits prior to privatization, virtually all of management was "probably not" or "definitely them financed by investors. It therefore seems not" hostile to the process. This finding that potential buyers themselves invested in strongly contradicts the qualitative findings of information. Frydman and Rapaczynski (1994) and other observers. It is, of course, possible that The information from audits does not seem to preprivatization management was the same as have been used for planning site remediation. that in place after privatization, but manage- In only 9.1 percent of the cases were site ment turnover was rather high; 44 percent out remediation plans developed before or shortly of the 201 firms answering this question said after privatization that included clearly that more than one member of the senior defined cleanup measures. Our data do not management team had been replaced within tell us who created remediation plans, but it one year after privatization. seems reasonable to suppose that firms prepared them, either before or after Prior to privatization, governments prepared privatization. These documents were then information packets for 76 percent of the firms probably reviewed and refined in conjunction in the sample. Among the firms that received with government agencies, including minis- information packets, 33 percent reported that tries of environment and privatization agen- sections dealing with environmental and site cies. In the rare cases in which site contamination issues were included. Sixty remediation plans were developed, cleanup percent of the information packets were costs were wholly the responsibility of inves- prepared by firms, 30 percent by consultants, tors. For cases where remediation plans were and 10 percent by privatization agencies. in place, an average of 31.4 percent of plans 22 Environment Strategy Papers Descriptive Statistics Table 5.2 Government provision of environmental information in privatization packets, by country (percentage of cases) Environmental Specific information on site Country information provided contamination provided Hungary 64.7 42.9 Lithuania 15.4 19.1 Poland 50.0 30.6 Romania 13.0 7.4 Slovak Republic 50.0 0 Table 5.3 Environmental audits, by country (percentage of cases) Country Preliminary audit Secondary audit Hungary 32.1 21.4 Lithuania 10.4 2.0 Poland 48.0 9.0 Romania 98.1 26.4 Slovak Republic 0 0 had been completed as of July 1998, but only 54.8 percent said that it was complete. Of 9 firms reported these data. those that had not completed site remediation, 41.2 percent said they planned to continue the Whether or not a plan was developed, in only work in the future. 5 cases (of 80 respondents reporting) did privatization agencies exempt investors from Environmental enforcement actions, such as cleanup costs. Liability seems to have virtually notification of violations, fines, or closures, always passed to investors. Only one firm were rare after privatization but were not reported having paid a reduced price to unheard of. Indeed, 9.3 percent of firms account explicitly for the existence of site reported having been subject to some kind of contamination. Thus, it appears that a policy enforcement action. Environment-related legal of "let the buyer beware"--and let buyers actions brought by the public against firms gather their own environmental and site were quite rare; 97 percent of firms reported contamination information--was generally in no such issues. Similar results were reported effect for sample enterprises at the time of for lawsuits brought by governments against privatization. firms. Following privatization, only 2 firms out As for whether site remediation actually Table 5.4 Degree to which site remediation has occurred, 21.4 percent of 182 firms been conducted, by country reporting said that some degree of site Percentage of cases in which some remediation had taken place since Country site remediation was undertaken privatization (see table 5.4 for country Hungary 40.0 Lithuania 21.3 data). We do not have information on the Poland 20.0 quality of remediation, but of those firms Romania 12.0 reporting that cleanup had taken place, Slovak Republic 7.1 Bluffstone and Panayotou 23 Does the Treatment of Environmental Liability during Privatization Really Matter? of 194 responding had brought environment- approximately 75 tons in 1997, 18 tons less related lawsuits against their governments. than in 1994. Turning to ongoing environmental behavior, To sum up, the data suggest that most firms in 68 percent of the firms in the sample reported the sample were privatized by tender and that having internal audit, pollution prevention, or over 30 percent of shares were typically waste minimization programs. Twenty-three purchased outright, rather than controlling percent of firms said that they were seeking interests being accumulated over time. Foreign ISO 14000 series certification, and 46 percent investment was the exception rather than the of the sample reported having an environmen- rule. General information was typically pro- tal management plan in place. In no cases do vided to investors, but provision of environ- we have information on how advanced these mental information of any kind was relatively planning processes were. uncommon. Potential purchasers therefore often invested in audits, but planning for site Although firms in the sample are large pollut- remediation was rare, suggesting that the audits ers, pollution emissions have, in general, were done for the purpose of due diligence. declined over time. Average wastewater Despite the relative lack of planning, site emissions in 1997 were over 296,000 cubic remediation at some level was reported by meters, but this amount was more than 45,000 about 20 percent of firms. Investors received cubic meters lower than in 1994. Biological few indemnifications or price discounts, but oxygen demand (BOD) emissions in 1997 environmental enforcement actions and averaged 114 tons, about 11 tons less than in lawsuits were also rare. These results suggest 1994. The picture for air pollutants was that until 1998, at least, pollution and site mixed. Sulfur dioxide emissions averaged 106 contamination issues may have been handled tons in 1997--an increase of about 20 tons informally, which might make indemnifications from 1994. Particulate emissions averaged and price discounts less necessary. 24 Environment Strategy Papers Chapter 6 Econometric Results T his chapter presents the econometric We make use of data on information provided analysis of the relationship between to investors, which constitute our independent environmental liability policies and "policy" variables. The key independent privatization and postprivatization outcomes. (dummy) variables used are as follows: The outcomes (that is, the dependent variables) examined include: ! Prior to privatization, was an information packet prepared for investors? 1. Privatization prices ! If an information packet was prepared, did it 2. Total privatization revenues received by include an environmental section that dealt privatization agencies with site contamination? 3. Foreign ownership ! Was any written information about possible 4. Efficiency of privatization processes site contamination provided to all or most 5. Site remediation investors? 6. Postprivatization conflicts, such as lawsuits ! Was a preliminary audit (one that utilized a and administrative actions questionnaire to learn the history of the 7. Postprivatization environmental perfor- facility) conducted before privatization? mance, measured with the use of various ! Was a secondary audit (one that included pollution and environmental indexes. sampling and analysis of soils and other materials) conducted before privatization? The environmental liability policies analyzed focus on (a) the availability and provision of a In addition, the seven outcomes were regressed variety of types of information regarding on a dummy variable indicating whether a site potential liabilities and (b) the development of remediation plan was developed and in place site remediation plans that include clearly at the time of or shortly after privatization. This defined cleanup measures to address site variable was used to attempt to capture the contamination. These policies are considered degree to which environmental issues were important methods for improving privatization clarified and were truly incorporated into and postprivatization outcomes. privatization processes. We do not have 25 Does the Treatment of Environmental Liability during Privatization Really Matter? information on who prepared the remediation magnitudes of the effects that changes in plans or how detailed they were. independent variables can be expected to have on dependent variables. For this reason, Finally, because conventional wisdom would for all final Tobit and logit models, the so- suggest that best practice in privatization called marginal effects of one-unit changes in should include both provision of information all independent variables (starting from the and the development of site remediation mean values) are included in parentheses next plans, a variable indicating whether both these to the coefficient estimates.6 These values are variables were used was included. In several the ones quoted in the text in discussions of cases this variable proved to be important. In magnitudes of effects. what follows, each environmental information After full models were run, insignificant variable is examined alone. The variable that variables were evaluated on the basis of the p- performed best is then combined with the use values and were then dropped from equations of site remediation plans into one model, and based on F or likelihood ratio tests. In some their interaction effects are examined. cases strong priors suggested that variables should remain in equations, and they were Because so few firms received explicit price therefore retained. In all cases environmental discounts or indemnification for potential information and site remediation planning liabilities, it was not possible to evaluate those variables were retained.7 policies using econometric methods. Only five firms received indemnifications, and only one Two different full-model specifications were paid a discounted price to account explicitly used. For the first six outcomes, which deal for potential liabilities. It is, of course, possible with issues other than ongoing environmental that investors implicitly discounted firms for performance, the basic firm information the uncertainties associated with contingent presented in table 6.1 was included as condi- liabilities. This possibility is considered in tioning variables in initial regression equa- subsequent sections. tions. Lower-limit censured Tobit regressions were To examine issues of ongoing pollution, it was used for all models in which the dependent necessary to include environmental monitor- ing and enforcement variables on the right- variable was continuous and a lower bound of hand side of the regressions. Degrees of zero for the dependent variable existed in the freedom were much more limited in the data. In such cases the Tobit is the most models of environmental performance, and appropriate estimation technique. The ordi- therefore we needed to limit our conditioning nary least squares (OLS) results are also variables to ones for which we had strong presented, but these results are provided only priors of a relationship with our environmental in appendix D. performance measures. Table 6.2 shows the conditioning variables used in the full model Where the dependent variable was a dummy of environmental performance. (0 or 1) variable, binomial logit or probit models were used. For Tobit and logit models, The next sections include only what are the coefficient estimates tell us little about the believed to be the most interesting and salient 26 Environment Strategy Papers Econometric Results Table 6.1 Conditioning variables used in regression equations to address outcomes 1 through 6 Reason to expect an effect on Variable privatization or postprivatization outcomes } Year of privatization Investors and governments learn over time, affecting outcomes. Earlier privatizations were generally considered more risky than later sales. Chemical industry (e.g., basic chemicals, resins, paints, pesticides) dummy variable (NAICS code 325) Primary metal manufacturing industry (e.g., iron and steel, aluminum, nonferrous metals, foundries) dummy variable (NAICS code 331) These economic sectors are often particularly Mining and oil and gas extraction prone to site contamination problems. dummy variable (NAICS codes 211 and 212) Petroleum and coal products industry (e.g., refineries, asphalt, paving and roofing products) dummy variable (NAICS code 324) Percentage of total shares held by Government typically has different goals from private investors, government in 1997 and retention of stakes by government may imply ongoing control. Percentage of total shares held by Larger shares held by single investors or group could imply lower largest private investor or group decision transactions costs. Percentage of total shares held by Foreign investors may bring experience and norms with them. foreign investors Revenues (in 1997 U.S. dollars) in year Firms may experience scale economies or diseconomies. prior to privatization Dummy variable indicating whether Proxy for management hostility at the time of privatization. more than one member of senior Hostile management may make privatization more difficult. management changed within one year of privatization Dummy variable indicating whether Advanced economic reform countries may have better the firm is in the more advanced privatization outcomes than countries slower to reform. reform countries of Hungary and Poland Revenue (in 1997 U.S. dollars) per More efficient companies may behave differently from less employee efficient ones. Note: NAICS, North American Industrial Classification System. results. In the interest of brevity and clarity, Relationship with provision of environmental most OLS results and certain other results are information presented only in appendix D. We know from the previous section that PRIVATIZATION PRICES information packets were generally offered to investors and that sometimes these packets Dependent variable: Share price as contained subsections dealing with site a proportion of firm revenues in the contamination. How did the provision of year before privatization environmental information to potential Bluffstone and Panayotou 27 Does the Treatment of Environmental Liability during Privatization Really Matter? Table 6.2 Conditioning variables used in regression equations to address outcome 7 Variable Reason to expect an effect on outcomes Year of privatization There may be autonomous trends in pollution emissions. Primary metal manufacturing This industry is known to have serious environmental problems. industry Mining and oil and gas extraction This industry is known to have serious environmental problems. Percentage of total shares held Government typically has different goals from private investors, and by government in 1997 retention of stakes by government may have implications for environmental performance. Percentage of total shares held Foreign investors may bring experience and norms with them that by foreign investors affect pollution emissions. Revenue (in 1997 U.S. dollars) More labor efficient companies may also be more environmentally per employee efficient. Hungary or Poland dummy These countries are known to be advanced reformers and therefore may have unobserved superiority related to environmental protection. Revenue (in 1997 U.S. dollars) More economically efficient companies are likely also to be more per employee environmentally efficient. Frequency of internal emissions Firms that monitor more will be more aware of pollution problems monitoring by firm employees and therefore are likely to have better environmental performance. Frequency with which air and Better control by regulators should reduce pollution emissions. water pollution are monitored by regulators Whether inspections by A proxy for regulatory stringency; unannounced visits are likely to be a environmental regulators are better enforcement measure than announced inspections. announced (dummy) investors affect privatization prices? Here, price meaningful. Approximately 80 percent privatization prices are measured as the ratio of firms in the data set fall into this category. of the price of one share sold to the annual revenues of firms in the year before As shown in table 6.3, for cases where privatization, both in 1997 U.S. dollars.8 information was provided, share prices tended to be higher when an environmental section Two types of privatization are considered in was included in the information packet.9 At this report: (a) when at least a 30 percent share the mean of all independent variables, the is purchased at the time of privatization, and (b) when one investor or investor group provision of environmental information by the accumulates a controlling interest over a government was associated with an increase period of time. This section focuses on cases in privatization prices of $0.182 per dollar of in which at least 30 percent of all shares was revenue in 1997--an increase, compared with sold outright at the time of privatization. Only the mean price, of 173 percent. This estimate in this case is the concept of a single share was significant at much more than the 1 28 Environment Strategy Papers Econometric Results Table 6.3 Tobit regression. Dependent variable: Share price as a proportion of firm revenues in the year prior to privatization Full model Final model Coefficient p-value Coefficient p-value Constant 226.358 1.56E­7 225.092 2.85E­5 (144.1) Whether environmental information is included 0.261721 0.0005139 0.283608 0.0002978 in general information packets provided to (0.182) investors (dummy) Year of privatization ­0.113604 1.58E­7 ­0.113054 2.83E­5 (­0.072) Chemical industry dummy variable ­0.0436403 0.531566 ­0.114187 0.207023 Primary metal manufacturing industry ­0.35957 0.0042632 ­0.446406 0.0099808 dummy variable (­0.073) Mining and oil and gas extraction dummy 0.0773701 0.697609 variable Petroleum and coal products industry ­0.152022 0.208985 ­0.207859 0.218364 dummy variable (­0.29) Percentage of total shares held by 0.0057472 0.0012426 0.0078605 0.0002568 government in 1997 (0.005) Percentage of total shares held by largest 0.0026084 0.0760148 0.0054607 0.0043593 private investor or group (0.0035) Percentage of total shares held by foreign ­0.0010074 0.505686 investors Revenues (in 1997 U.S. dollars) in year prior ­1.08E­10 0.701492 to privatization More than one member of senior 0.11297 0.0139216 0.0675924 0.262981 management changed within one year of (0.043) privatization (dummy) Dummy variable for Hungary and Poland ­0.0780233 0.269895 Revenue (in 1997 U.S. dollars) per employee ­1.43E­7 0.62813 Log likelihood ­16.21 ­12.72 Mean of dependent variable 0.054 0.105 Number of observations 77 87 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of the independent variables. percent level, suggesting that we can be determines the provision of environmental extremely confident in the estimate. information. The interpretation of this result that is consis- Note, however, that we have tried to adjust tent with the literature would be that all else for a variety of firm characteristics. If indeed being equal, providing environmental infor- we were able to adjust fully for differences in mation to potential investors makes firms firm quality, we would conclude that provid- more desirable and results in higher prices. In ing environmental information to investors reality, though, privatization agencies may causes desirability, presumably by helping to have provided more environmental informa- clarify quality and reduce uncertainty. What- tion for inherently desirable firms. The causal- ever the origin, the results strongly suggest ity would then be reversed: desirability that provision of environmental information Bluffstone and Panayotou 29 Does the Treatment of Environmental Liability during Privatization Really Matter? does not make industrial firms that could It is probably not correct to say that the reasonably be supposed to have environmen- provision of information reduced privatization tal problems less desirable to investors. prices. Rather, this result suggests that, all else Indeed, if anything, the opposite seems to be being equal, general information was more true. likely to be provided for inferior industrial firms. The most desirable firms were probably Both year of privatization and the metal investigated by potential investors or were industry dummy were negatively and signifi- sold to insiders without providing such cantly related to privatization prices. Being in information. The result possibly does suggest the metal industry was correlated with lower that the provision of general information per privatization prices (equivalent to about 70 se does not in itself bring higher prices for percent of the mean), suggesting that the metal industrial firms. industry is one of the least desirable industries. An one-year increase in the date of Relationship with having a well defined privatization reduced prices by slightly over cleanup plan, and the interaction of environ- 70 percent of the mean privatization price. mental information and site remediation This result confirms the prediction of an plans autonomous price decline due to adverse selection presented in Bluffstone and How is the existence of a well defined site Panayotou (2000). remediation plan linked with privatization prices? The spirit of the literature is that the If, instead of the variable for whether an clearer and more explicit is the description of environmental section was included in the privatization packet, we use a variable site contamination problems, the lower is focusing on whether written information on uncertainty and the better will be the possible site contamination was provided, no privatization outcomes. Our hypothesis is that relationship with privatization prices is the existence of a cleanup plan should detected. The same is true for the existence of increase privatization prices.11 primary and secondary audits. In this case the Tobit and OLS results conflict. Appendix D presents the results of a Tobit The OLS results suggest that the existence of a model of the relationship between general cleanup plan was associated with a reduction firm information and privatization prices. The in price, but when we adjust for the censuring results suggest that the provision of general with a Tobit model, this relationship disap- information was associated with reduced pears. We also included environmental privatization prices (p-value ~ 0).10 With a information, the existence of a site mean value of the dependent variable of 0.13, remediation plan, and the interaction between the magnitude of the effect was extremely the two variables in one model without strong. At the mean of the dependent variable, materially changing our conclusions. These provision of a general information packet was results suggest that environmental information correlated with a reduction in price of over may be the important policy variable with 100 percent. regard to privatization prices. 30 Environment Strategy Papers Econometric Results TOTAL REVENUES FROM PRIVATIZATION with the mean (p~0). This result is again very much in concert with the predictions of Dependent variable: Total revenues Bluffstone and Panayotou (2000). The other from privatization (price per share highly significant variable in the Tobit model sold multiplied by the number of was the Hungary and Poland dummy variable. shares sold) as a percentage of firm This variable came in strongly negative (about revenues in the year before four times the mean value!) and highly privatization significant, with a p-value of virtually zero, suggesting that all else being equal, firms in Relationship with provision of environmental Hungary and Poland were generating much information lower privatization revenues for national treasuries. We now analyze the determinants of privatization revenues as a percentage of Relationship with having a cleanup plan that revenues in the year before privatization. This clearly defines cleanup measures and respon- analysis is again restricted to cases in which at sibilities, and the interaction of environmen- least 30 percent ownership stakes were sold tal information and site remediation plans outright at the time of privatization. Privatization revenues are calculated as the In neither the full model nor the final model price per share multiplied by the number of was any relationship detected between the shares sold at the time of privatization. existence of a remediation plan alone and privatization revenues. All results for this subsection are presented in appendix D. The Tobit models do not allow us Table 6.4 presents a Tobit model that includes to say with confidence that providing either the provision of environmental information, general firm information or environmental site remediation planning, and the interaction information is correlated with increased between the two variables. We see that privatization revenues. The most significant including all variables together changes the result was in the final Tobit model of environ- results dramatically. Of most interest is the mental information, which was positively case when environmental information, associated with privatization revenues at the including site contamination, is provided and 14 percent level. The estimated marginal is followed up with a site remediation plan; effect of providing environmental information privatization revenues are estimated to was to increase revenues by 75 percent of the increase by 200 percent compared with mean mean value of US$2.27. This result corre- privatization revenues. This estimate is sponds very well with the results for significant at approximately the p = 0.08 level privatization prices. and is robust when we move from the full model to the final model. We therefore can be Year of privatization was negatively related to reasonably sure of this conclusion. revenues (as it was to prices). Privatization one year later was estimated to be correlated The results suggest that in order to sell more with revenues of about 70 percent compared shares at higher prices, and so actually bring Bluffstone and Panayotou 31 Does the Treatment of Environmental Liability during Privatization Really Matter? Table 6.4. Tobit regression. Dependent variable: Privatization revenues in 1997 U.S. dollars as a percentage of firm revenues in the year prior to privatization Full model Final model Coefficient p-value Coefficient p-value Constant 5,919.3 0.000182 4,924.92 1.24E­9 (3,252.8) Whether environmental 2.67529 0.291645 1.2916 0.467348 information, including information (0.853) on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan ­0.49671 0.827477 ­1.01102 0.623705 was in place (dummy) (­0.668) Interaction term between the 5.29678 0.207506 7.00833 0.077907 existence of a site remediation plan (4.63) and the provision of environmental information at the time of privatization Year of privatization ­2.96614 0.000188 ­2.4644 1.27E­9 (­1.63) Chemical industry dummy 2.8518 0.145914 2.37673 0.214695 variable (1.57) Primary metal manufacturing ­9.39773 0.113379 ­6.28941 0.238977 industry dummy variable (­4.15) Mining and oil and gas extraction 0.0151 0.99783 dummy variable Petroleum and coal products ­1.93029 0.559923 industry dummy variable Percentage of total shares held by 0.078921 0.217194 government in 1997 Percentage of total shares held by 0.087387 0.085981 largest private investor or group Percentage of total shares held by ­0.07789 0.081272 ­0.09483 0.02545 foreign investors (­0.063) Revenues (in 1997 U.S. dollars) in ­1.46E­9 0.847151 year prior to privatization More than one member of senior ­0.01199 0.99305 management changed within one year of privatization (dummy) Dummy variable for Hungary and ­9.10406 3.58E­5 ­9.49893 4.59E­7 Poland (­6.27) Revenue (in 1997 U.S. dollars) per 8.61E­8 0.991328 employee Log likelihood ­193.68 ­197.99 Mean of dependent variable 2.33 2.27 Number of observations 66 68 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of the independent variables. more revenues into treasuries through dressed. This means both providing informa- privatization of industrial enterprises, site tion and developing site remediation plans. contamination issues need to be fully ad- The pursuit of either policy by itself seems to 32 Environment Strategy Papers Econometric Results have limited effects on revenues. We cannot environmental audits, despite the conven- say with certainty that this combination of tional wisdom that foreign owners are more policies leads to increased revenues--it is likely to implement due diligence procedures. certainly possible that site contamination Similar results were obtained from a logit issues were more likely to have been compre- model that attempted to explain the existence hensively addressed for inherently more of any foreign ownership in firms. We there- desirable firms, but it is difficult to envision fore must be very doubtful that a relationship why this would be the case. Because compet- existed between foreign ownership and ing explanations are limited, we can be rather environmental information. sure that fully addressing site contamination problems through information and Variables positively correlated with percent- remediation planning yields higher revenues. age of foreign ownership and the foreign ownership dummy were the chemical industry Results for conditioning variables were similar dummy and firm revenues in 1997. The to those presented in the analysis of environ- percentage of shares held by the government mental information alone. The Poland- in 1997 was negatively related to foreign Hungary dummy variable is again negatively ownership.12 and significantly related to privatization revenues, as is the year of privatization. Relationship with having a cleanup plan that clearly defines cleanup measures and respon- FOREIGN OWNERSHIP sibilities, and the interaction of environmen- tal information and site remediation plans Dependent variable: Percentage of total firm ownership held by The rationale for the existence of a relation- foreigners in 1997 ship between cleanup plans and foreign ownership is similar to that for privatization Relationship with provision of environmental prices. Foreign owners are in general consid- information ered to be high-quality investors and are hypothesized to be sensitive to the financial In Central Europe, foreign investors have been risks associated with site contamination. We actively sought for the finance and expertise therefore expect privatization programs that they are often able to provide. If provision of include organized systems for inventorying environmental information improves and addressing contamination problems to be privatization outcomes, we might expect to attractive. Our hypothesis is that the existence see higher foreign participation when more of a cleanup plan should increase foreign information is provided. ownership. In reality, though, in no specifications were The results suggest that there is no relationship the environmental information variables in the between foreign ownership share (or foreign data set significantly associated with percent- ownership in firms) and the existence of a site age of ownership. This conclusion also remediation plan. An autonomous decline in applied to both secondary and primary foreign participation was noted in the data, Bluffstone and Panayotou 33 Does the Treatment of Environmental Liability during Privatization Really Matter? however, with the year-on-year foreign ! Number of tenders received ownership share declining by 2.5 percent per during privatization year. If foreign ownership is a positive ! Number of auction rounds privatization outcome, this result corresponds required to privatize firms well with the strong year-to-year decline in ! Number of auction bids received privatization prices and revenues that was during privatization estimated to exist. This result again supports ! Percentage of total ownership the results of Bluffstone and Panayotou (2000). sold at the time of privatization ! Dummy variable indicating Alternative models that included environmen- whether, according to tal information, the existence of a site privatization officials, environ- remediation plan in one model, and an mental issues delayed interaction term yielded no additional insights. privatization of firms At least at the stage of transition covered by the data, there did not seem to be a relation- Relationship with provision of environmental ship between environmental information, site information remediation plans, or a sequence including both these instruments. Another hypothesis put forth in the literature is that providing more information about firms' As was mentioned earlier, the amount of environmental status will increase the effi- foreign ownership in the data set is limited. ciency of privatization processes. For ex- The average foreign ownership share was only ample, by resolving uncertainties regarding 12 percent, and over three-quarters of all firms environmental liabilities, demand for firm had no foreign ownership. With foreign shares will be enhanced, and competition in participation the exception rather than the rule auctions will increase. The literature also in these industrial firms, it perhaps makes suggests (although it perhaps does not specifi- sense that factors other than environmental cally assert) that resolving uncertainty will information would drive foreign ownership. help privatization move more quickly. We may also have a problem of omitted variables that is causing environmental Anecdotal evidence, however, indicates that information to be insignificant. the opposite view is often held by privatization agencies. These organizations EFFICIENCY have sometimes been reluctant to incorporate OF PRIVATIZATION PRO- environmental issues into privatization CESSES processes for fear of slowing privatization or Dependent variables: reducing its quality. Indeed, this may be the mainstream view in the region. As is shown in ! Time in months required to the overview of country policies (and in more privatize firms detail in appendix A), for much of the 1990s ! Number of tender rounds environmental liability problems were required to privatize firms ignored. 34 Environment Strategy Papers Econometric Results Because we are interested in the quality of the Table 6.5 presents a Tobit model of the time process, and because it is impossible to required to privatize firms. The results indicate analyze the process when privatization takes that provision of environmental information to place over a period of years, the analysis is investors appears to be related to a slowing of restricted to firms for which at least a 30 the privatization process. Where environmen- percent share was purchased at the time of tal information was provided, firms are privatization. We first test whether the process estimated to have taken approximately 4.2 was accelerated or slowed by the provision of months longer to privatize, which is equiva- information. We then test whether the effi- lent to increasing the mean privatization time ciency of auctions and tenders was improved by 50 percent. This coefficient estimate is also by the inclusion of environmental information. highly significant and appears to be very Table 6.5. Tobit regression. Dependent variable: Time required to privatize, in months Full model Final model Coefficient p-value Coefficient p-value Constant 1,603.87 0.002425 1,484.07 0.000602 (1.411.5) Whether environmental information, 4.78153 0.007693 4.44164 0.008093 including information on site (4.22) remediation, was provided in privatization packets (dummy) Year of privatization ­0.79984 0.002606 ­0.73928 0.000642 (­0.703) Chemical industry dummy variable ­0.04569 0.981752 Primary metal manufacturing ­4.5573 0.150209 ­4.51723 0.153143 industry dummy variable (­4.30) Mining and oil and gas extraction 6.59153 0.085287 6.74159 0.074499 dummy variable (6.41) Petroleum and coal products ­1.27114 0.690724 industry dummy variable Percentage of total shares held by 0.013538 0.678054 government in 1997 Percentage of total shares held by ­0.05952 0.046083 ­0.067 0.003047 largest private investor or group (­0.064) Percentage of total shares held by 0.044013 0.111566 0.042536 0.102699 foreign investors (0.0404) Revenues (in 1997 U.S. dollars) in ­1.19E­8 0.094041 ­1.17E­8 0.093837 year prior to privatization (­1.11 E­8) More than one member of senior ­1.69405 0.143611 ­1.66864 0.147555 management changed within one (­1.59) year of privatization (dummy) Dummy variable for Hungary and 4.18E+00 0.03032 4.00764 0.023719 Poland (3.811) Revenue (in 1997 U.S. dollars) per ­1.02E­5 0.178161 ­1.02E­5 0.177015 employee (­0.97 E­5) Log likelihood ­255.09 ­256.01 Mean of dependent variable 8.37 8.37 Number of observations 85 85 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of the independent variables. Bluffstone and Panayotou 35 Does the Treatment of Environmental Liability during Privatization Really Matter? robust to changes in specification. If we drop under these circumstances, no information any or all of the conditioning variables, results would be required. Such phenomena would change very little. It therefore appears that show up in the data in the form of less time when the government provided information being required to sell firms for which no on the environmental status of firms, including information was provided. site contamination, the time required to privatize was increased. In terms of conditioning variables, privatization processes appear to have be- Some may see these results as evidence that come more efficient over time. Year-on-year providing environmental information to privatization time was estimated to decline by investors has the downside of causing 21 days, and this estimate was significant at privatization delays. It is possible that this is the 1 percent level. This finding perhaps correct and that government provision of suggests that, all else being equal, information on potential environmental privatization agencies have become more problems, including site contamination, efficient in their operations. Firms where stimulates further due diligence measures that ownership ended up more concentrated took require more time. Alternatively, firms for less time to privatize than other firms, as did which environmental information was pro- firms with postprivatization management vided may have been less desirable for changes (p = 0.15) and more efficient firms reasons other than those included in the (that is, those with higher revenues per regression--year of privatization, revenues, employee in 1997).13 Firms with higher efficiency, industry, and country of location. revenues took less time to privatize, but the In other words, some key unobservable aspect effect was estimated to be small (p = 0.09). of firm quality that is correlated with the provision of environmental information could Attracting foreign participation appears to take have been omitted, biasing our coefficient longer. Increasing foreign participation by 10 estimate. percent is estimated to increase privatization time by about 12 days, and this estimate was For example, provision of environmental significant at the 10 percent level. An interest- information may have been more likely when ing result was that mining firms were esti- environmental issues, particularly site con- mated to take more than six months longer to tamination, were important. We do not have a privatize than other firms--an increase of true proxy for environmental problems, other more than two-thirds compared with the mean than industry, and it is likely that provision of time to privatize of eight months (significant at environmental information variables is at least the 7 percent level). This result suggests that a partial proxy for the environmental quality mining firms are systematically more difficult of firms prior to privatization. to privatize. Another possible explanation is insider deals. Firms in Hungary and Poland are predicted to For example, companies may have been take longer to privatize than firms in other quickly purchased by managers or employees countries. Simply being located in those (e.g., through liquidation, as in Poland), but advanced reform countries is estimated to 36 Environment Strategy Papers Econometric Results increase privatization time by almost four ship was detected between primary and months, or 45 percent. This result was signifi- secondary audits and time required to priva- cant at the 2 percent level and perhaps tize, nor was there any relationship between corresponds well with the conventional the provision of general firm information and wisdom that privatization in these countries time required to privatize. has generally taken longer than elsewhere. Table 6.6 analyzes the determinants of the Similar results were found for a model utiliz- number of tender rounds required for firms ing the variable indicating whether informa- privatized by that method. The mean number tion about possible site contamination was of rounds required was 1.5, with a standard provided to most or all investors. No relation- deviation of 0.76 rounds. Consistent with the Table 6.6. Tobit regression. Dependent variable: Number of tender rounds required to privatize firms Full model Final model Coefficient p-value Coefficient p-value Constant 44.1216 6.23E­1 0.18001 0.658085 (0.177) Whether environmental information, 0.513855 0.083962 0.69565 0.004762 including information on site (0.684) remediation, was provided in privatization packets (dummy) Year of privatization ­0.02224 0.622594 Chemical industry dummy variable 0.398678 0.153283 Primary metal manufacturing ­0.02645 9.60E­1 industry dummy variable Mining and oil and gas extraction 0.237452 0.65848 dummy variable Petroleum and coal products ­0.06632 0.881737 industry dummy variable Percentage of total shares held by 0.015471 0.01525 0.011507 0.003806 government in 1997 (0.0113) Percentage of total shares held by 0.013486 6.80E­3 0.014066 0.004707 largest private investor or group (0.0138) Percentage of total shares held by ­0.00922 6.43E­2 ­0.00742 0.097668 foreign investors (­0.0073) Revenues (in 1997 U.S. dollars) in 9.49E­10 0.347289 year prior to privatization More than one member of senior 0.268064 0.141251 0.256346 0.154891 management changed within one (0.25) year of privatization (dummy) Dummy variable for Hungary and 4.88E­1 0.245821 Poland Revenue (in 1997 U.S. dollars) per ­1.29E­6 0.220065 ­1.09E­6 0.311949 employee (­1.07 E­6) Log likelihood ­70.20 ­73.82 Mean of dependent variable 1.47 1.49 Number of observations 68 69 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. Bluffstone and Panayotou 37 Does the Treatment of Environmental Liability during Privatization Really Matter? preceding Tobit model, the provision of the percentage of government ownership in environmental information was associated 1997 and the percentage owned by the largest with an increase in the number of tender investor. These results suggest that selling at rounds required to privatize firms. This least a 30 percent interest in firms while also estimate was significant at the 1 percent level. retaining higher shares in government hands The effect was large; provision of environmen- takes longer and that selling larger blocks of tal information was associated with an in- shares at one time requires more tender crease in number of rounds of about 46 rounds. The estimated magnitudes of the percent at the mean. effects, however, were modest. For example, an increase in the share held by government Also highly significant and positively associ- of 10 percent (an 18 percent increase com- ated with the number of tender rounds were pared with the mean value of 60 percent in Table 6.7. Tobit regression. Dependent variable: Number of auction rounds required to sell at least 30 percent of firms' shares Full model Final model Coefficient p-value Coefficient p-value Constant 394.998 0.28608 1.85689 6.28E­11 (1.788) Whether environmental information, 1.87822 0.025705 1.27097 0.0169197 including information on site (1.22) remediation, was provided in privatization packets (dummy) Year of privatization ­0.41427 0.763863 Chemical industry dummy variable ­0.19742 0.288154 Primary metal manufacturing ­12.6805 0.137568 industry dummy variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by 0.003062 0.812851 government in 1997 Percentage of total shares held by 0.010872 0.410448 largest private investor or group Percentage of total shares held by ­0.0271 0.089168 foreign investors Revenues (in 1997 U.S. dollars) in 5.13E­8 0.154965 ­3.54E­9 0.356265 year prior to privatization (­3.4 E­9) More than one member of senior ­2.92E­1 0.618678 ­0.290654 0.466286 management changed within one (­0.28) year of privatization (dummy) Dummy variable for Hungary and ­2.42344 0.048549 ­1.11101 0.0443712 Poland (­1.07) Revenue (in 1997 U.S. dollars) per ­1.69E­6 0.912624 employee Log likelihood ­26.5 ­37.63 Mean of dependent variable 1.55 1.74 Number of observations 20 27 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. 38 Environment Strategy Papers Econometric Results this subsample) increased the number of privatization were sold via tenders. Observa- tender rounds by 0.11, or about 8 percent. tions on auctions are therefore much fewer Foreign ownership and firm efficiency were than for tenders. Because observations are negatively and significantly associated with limited and the sample changes somewhat in number of tender rounds, but these effects the move from the full to the final model, the were small. results should be interpreted with care. We now turn our attention to auction mecha- Table 6.7 presents results on auction rounds nisms. As noted, most firms for which at least that are broadly consistent with those related 30 percent of shares were sold at the time of to time and number of tender rounds required Table 6.8. Tobit regression. Dependent variable: Number of tenders received in the round in which firms were privatized Full model Final model Coefficient p-value Coefficient p-value Constant ­51.0005 0.796129 ­0.14694 0.873694 (­0.142) Whether environmental information, 0.20525 0.746074 0.114599 0.834485 including information on site (0.11) remediation, was provided in privatization packets (dummy) Year of privatization 0.025179 0.79977 Chemical industry dummy variable ­0.80176 0.207703 ­0.74033 0.215433 (­0.716) Primary metal manufacturing ­0.98848 0.396015 ­0.97087 0.400878 industry dummy variable (­0.939) Mining and oil and gas extraction ­1.12785 0.486321 ­1.19447 0.460499 dummy variable (­1.16) Petroleum and coal products ­0.54386 0.585688 industry dummy variable Percentage of total shares held by 0.047153 0.003488 0.04235 1.20E­5 government in 1997 (0.041) Percentage of total shares held by 0.02135 0.060919 0.021602 0.056102 largest private investor or group (0.021) Whether there was any foreign 1.21973 0.10148 1.05062 0.128885 ownership (dummy) (1.017) Revenues (in 1997 U.S. dollars) in ­4.40E­9 0.06143 ­4.54E­9 0.055274 year prior to privatization (­4.39 E­9) More than one member of senior 0.284017 0.476509 management changed within one year of privatization (dummy) Dummy variable for Hungary and 0.434655 0.666208 Poland Revenue (in 1997 U.S. dollars) per ­5.34E­7 0.824562 employee Log likelihood ­123.32 ­123.9 Mean of dependent variable 2.88 2.88 Number of observations 68 68 Note: Firms included in this regression are those for which at least 30 percent stakes were sold outright and firms were priva- tized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. Bluffstone and Panayotou 39 Does the Treatment of Environmental Liability during Privatization Really Matter? to privatize. Firms for which environmental the results suggest that, all else being equal, information was provided in privatization competition for larger firms was less than for packets are estimated to require 1.22 more smaller firms.14 auction rounds to privatize. This marginal effect is about 70 percent of the mean. It is We ran the regression with both percentage of also of note that coefficient estimates are foreign ownership and a dummy variable highly significant in both the full and final indicating whether there was any foreign models (p ~ 0.02) and do not change dramati- involvement in firms. The estimates were cally between the two models. With the similar in both equations, but the dummy caveat that the sample is small, we can variable offered more precision. Firms that therefore be reasonably confident of the had some foreign participation were priva- results. tized in significantly more competitive markets than those without foreign ownership. Of the conditioning variables, only the Indeed, in the final model, firms with foreign Hungary-Poland dummy was significant in the participation had an estimated 1.0 more final model. The model suggests that firms tenders (about 40 percent more) than other privatized by auction in those countries took firms (p = 0.13). We interpret this result as 1.07 fewer rounds to privatize (p = 0.04). This indicating that some quality of firms not result somewhat contradicts the findings with observable by us attracted more competition regard to privatization time, which suggested and that foreigners seemed to seek out those that Hungarian and Polish firms took longer to firms even if competition was perhaps keener. privatize. Somewhat surprisingly, firms with a higher Table 6.8 presents results related to the retained government share in 1997 had more number of tenders received in the rounds in competitive privatization markets. For ex- which firms were privatized. This variable ample, a 10 percent increase in government measures the attractiveness of firms to poten- ownership (18 percent compared with the tial investors and also the degree of competi- mean level) was correlated with an increase in tion that existed in privatization markets. We number of tenders of 0.41, or 14 percent more see that in both the full and final models than the mean. Probably this reflects govern- environmental information provided at the ment retention of shares in higher-quality time of privatization was positively, but not firms--a result that corresponds well with the significantly, related to the competitiveness of predictions of Akerloff-type adverse selection privatization markets (p = 0.78 in the final models such as those presented in Balaban, model). Bluffstone, and Panayotou (1994) and Bluffstone and Panayotou (2000). Our conditioning variables offer significant insights into the operation of privatization Another way to look at the efficiency of markets in Central Europe. Firm revenues in privatization is whether few or many shares 1997 were negatively and highly significantly were sold at the time of privatization. We correlated with number of tenders received. again define privatization as the selling of Although the magnitude of the effect is small, more than a 30 percent stake outright. All else 40 Environment Strategy Papers Econometric Results Table 6.9. Tobit regression. Dependent variable: Percentage of all shares purchased from the government at the time of privatization Full model Final model Coefficient p-value Coefficient p-value Constant ­3,257.56 0.124605 ­3,509.81 0.066531 (­3,507.82) Whether environmental information, ­12.096 0.060619 ­10.1546 0.045635 including information on site (­10.15) remediation, was provided in privatization packets (dummy) Year of privatization 1.67656 0.115586 1.80299 0.060441 (1.802) Chemical industry dummy variable ­3.51619 0.633329 Primary metal manufacturing ­21.2484 0.084155 ­22.4396 0.061947 industry dummy variable (­22.42) Mining and oil and gas extraction ­2.30552 0.88098 dummy variable Petroleum and coal products 12.7186 0.321271 industry dummy variable Percentage of total shares held by ­0.43203 0.000458 ­0.43327 1.33E­7 government in 1997 (­0.433) Percentage of total shares held by ­0.02991 0.789199 largest private investor or group Percentage of total shares held by ­0.02991 0.775629 foreign investors Revenues (in 1997 U.S. dollars) in ­7.42E­9 0.794146 year prior to privatization More than one member of senior ­0.51371 0.908955 management changed within one year of privatization (dummy) Dummy variable for Hungary and 8.61411 0.224253 6.51 0.283869 Poland (6.51) Revenue (in 1997 U.S. dollars) per 1.38E­5 0.652013 ­8.17E­6 0.460691 employee (­8.17E­6) Log likelihood ­398.92 ­455.37 Mean of dependent variable 63.1 63.1 Number of observations 90 103 Note: Firms included in this regression are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. being equal, privatization agencies are likely time of privatization of 10 percent (p = 0.05), to prefer to sell a higher percentage of shares or approximately 15 percent fewer shares at one time. In the subsample the mean (compared with the mean) than when no percentage of shares sold at the time of information was offered. As was true for the privatization was approximately 63 percent. analysis of privatization time, this finding could be attributable to a number of factors, As shown in table 6.9, the provision of which have already been discussed. environmental information prior to privatization was associated with a reduction Not surprisingly, a higher percentage of in the percentage of total shares sold at the government ownership retained in 1997 was Bluffstone and Panayotou 41 Does the Treatment of Environmental Liability during Privatization Really Matter? associated with a lower percentage of shares mental due diligence of the type one would sold at the time of privatization. The model hope to see in complex industrial firms may estimates that a 1 percent increase in govern- take time. ment ownership in 1997 was associated with a 0.43 percent reduction in total shares sold at Simply providing environmental information the time of privatization (p ~ 0). Firms in the to potential investors at the time of metal manufacturing industry sold much privatization does not seem to resolve the smaller share blocks at the time of problem that firms with more complicated and privatization. Indeed, firms in this industry difficult environmental problems take longer sold ownership blocks that were 22 percent to privatize. Those who claim the opposite are smaller than firms in other industries, which is probably being overly optimistic; providing about one-third of the mean share block sold environmental information alone does not of 65 percent (p = 0.06). No other condition- seem to be a tool for speeding privatization. ing variables were associated with the depen- dent variable at any reasonable level of Relationship with having a well defined significance. cleanup plan, and the interaction of environ- mental information and site remediation No relationship was found between the other plans environmental information variables and the percentage of shares sold at the time of This section examines the relationship be- privatization. In other specifications, results tween having a site remediation plan before or were similar to those presented in table 6.9.15 shortly after privatization that includes clearly defined cleanup measures and our measures The results in this section suggest that realiza- of the efficiency of the privatization process. tion of the benefits associated with better We again emphasize that we know which management of environmental liabilities (e.g., firms developed remediation plans prior to or higher prices, higher privatization revenues, shortly after privatization but not who pre- and more competitive privatization markets) pared the plans. requires extra time. The evidence indicates that when environmental information was The existence of a site remediation plan alone provided to investors, all else being equal, is negatively correlated with privatization privatization took longer, and smaller blocks time, but the estimate is only significant at of shares were sold at one time. the 30 percent level in the Tobit model. We therefore only report the results in A number of reasons for these results have appendix D. been discussed, but even if environmental information indeed delayed privatization, this The results from models of privatization time outcome would not necessarily be a negative so far suggest that the time taken to privatize is one. Like a catchy headline in a newspaper positively related to the provision of environ- article, environmental information provided in mental information and is potentially reduced privatization packets may cause investors to by the existence of site remediation plans. It is investigate further. In other words, environ- likely, however, that these two policies are 42 Environment Strategy Papers Econometric Results linked and, when used together, may yield ment of site remediation plans have indepen- different outcomes. Solely considering each dent effects on privatization time but that measure independently may therefore not when used together, their effects are particu- offer a full picture. larly powerful. In the final model the provision of environmental information was associated Table 6.10 suggests that the provision of with an increase in privatization time of about environmental information and the develop- three months, which is consistent with previ- Table 6.10. Tobit regression. Dependent variable: Time required to privatize, in months Full model Final model Coefficient p-value Coefficient p-value Constant 1,155.6 0.246531 7.07152 2.89E­15 (6.631) Whether environmental information, 3.74481 0.070223 3.36958 0.073991 including information on site (3.16) remediation, was provided in privatization packets (dummy) Whether a site remediation plan was ­1.63109 0.440016 ­1.81168 0.386973 in place prior to or shortly after (­1.70) privatization (dummy) Interaction term between the ­6.3076 0.156296 ­6.93485 0.118144 existence of a site remediation plan (­6.50) and the provision of environmental information in privatization packets Year of privatization ­0.57503 0.250559 Chemical industry dummy variable 0.078187 9.69E­1 Primary metal manufacturing ­5.23419 1.16E­1 industry dummy variable Mining and oil and gas extraction 3.74259 0.498453 3.85699 0.487979 dummy variable (3.62) Petroleum and coal products ­0.60999 0.850963 industry dummy variable Percentage of total shares held by 0.003474 0.933063 government in 1997 Percentage of total shares held by ­6.76E­2 0.068588 ­0.05525 0.017564 largest private investor or group (­0.0518) Percentage of total shares held by 0.037645 0.208243 0.04523 0.109079 foreign investors (0.0424) Revenues (in 1997 U.S. dollars) in ­7.55E­9 0.304701 ­7.47E­9 0.312314 year prior to privatization (­7.01E­9) More than one member of senior ­6.35E­1 0.60718 management changed within one year of privatization (dummy) Dummy variable for Hungary and 6.3174 0.004368 6.53184 0.001241 Poland (6.125) Revenue (in 1997 U.S. dollars) per ­1.30E­5 0.09078 ­1.25E­5 0.099724 employee (­1.18E­5) Log likelihood ­224.42 ­226.36 Mean of dependent variable 7.8 7.8 Number of observations 75 75 Note: Firms included in this regression are those for which at least 30 percent stakes were sold outright and firms were priva- tized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. Bluffstone and Panayotou 43 Does the Treatment of Environmental Liability during Privatization Really Matter? ous results (p = 0.07). Site remediation Table 6.11 presents a model of the number of planning was associated with a reduction in tender rounds required to privatize firms. The privatization time of approximately 1.7 model includes the interaction between months, but the reduction was insignificant (p environmental information and the develop- = 0.39). When information was provided and ment of site remediation plans.16 We see that was then followed up with a site remediation in the final model there is no independent plan, however, privatization time was esti- relationship between number of tender rounds mated to be reduced by 6.5 months, or required to privatize and either environmental information or the development of site approximately 83 percent of the mean. This remediation plans. decline is significant at the 11 percent level in the Tobit model. In the OLS model the The interaction term indicating that environ- coefficient estimate was very similar but was mental information was provided and then a significant at the 1 percent level. remediation plan was developed, however, is negatively related to the number of tender Taken in full context, it therefore appears that rounds required to privatize. The estimated the generally accepted best practice of marginal effect is ­2.73, which means that providing environmental information at the when these instruments are used together, the time of privatization does indeed make decline in the number of tender rounds privatization markets work more quickly. The needed to privatize was 174 percent of the main condition is that for firms with serious mean number of rounds. The estimate is contamination problems (that is, the firms significant at about the p ~0 level of confi- most likely to benefit from provision of dence. We can therefore be quite confident in information), information should be accompa- this result. nied by concrete plans for addressing contami- nation problems; otherwise the benefit of We conclude that while environmental faster privatization is not realized. information alone is positively correlated with the number of tender rounds required to This result is intuitive and not unexpected. privatize, and site remediation plans are Those firms in our sample that had site unrelated, the combination of these two policies is most important. As was true for remediation plans developed them prior to or privatization time, combining these policies shortly after privatization, so the information offers an opportunity to speed significantly associated with those remediation plans was privatizations of firms that are likely to have largely collected before the date of site contamination problems. privatization. Governments using these two policies together therefore signal a transparent In this model the chemical industry dummy is and straightforward approach toward site positively related to the number of tender contamination problems that is likely to be rounds needed to privatize. Firm revenues attractive to investors. This is the approach appear to be correlated with increases in that Hungary used in its highly successful tender rounds, although the magnitude of the privatization of the petroleum firm MOL Rt. coefficient estimate is small. Changes in firm 44 Environment Strategy Papers Econometric Results Table 6.11. Tobit regression. Dependent variable: Number of tender rounds required to privatize Full model Final model Coefficient p-value Coefficient p-value Constant 116.364 0.666752 1.12981 4.67E­13 (1.10) Whether environmental 0.596909 0.083472 0.203248 0.385735 information, including information (0.199) on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan 0.374645 0.208863 0.168977 0.622825 was in place prior to or shortly after (0.165) privatization (dummy) Interaction term between the ­2.30271 0.00285 ­2.79161 0.0004031 existence of a site remediation plan (­2.73) and the provision of environmental information in privatization packets Year of privatization ­0.05807 0.668707 Chemical industry dummy variable 0.670979 0.019018 0.629723 0.0258082 (0.616) Primary metal manufacturing ­0.08152 0.894175 industry dummy variable Mining and oil and gas extraction ­0.77284 0.289528 ­0.808738 0.353084 dummy variable (­0.79) Petroleum and coal products ­0.21248 0.631077 industry dummy variable Percentage of total shares held by 0.00762 0.523972 government in 1997 Percentage of total shares held by 5.17E­3 0.390713 largest private investor or group Percentage of total shares held by ­0.00721 0.189373 foreign investors Revenues (in 1997 U.S. dollars) in 1.90E­9 0.066909 2.59E­9 0.0186412 year prior to privatization (2.53E­9) More than one member of senior 3.74E­1 0.04511 0.52046 0.0074321 management changed within one (0.51) year of privatization (dummy) Dummy variable for Hungary and 0.122486 0.84882 Poland Revenue (in 1997 U.S. dollars) per ­1.41E­6 0.175697 employee Log likelihood ­57.07 ­78.81 Mean of dependent variable 1.46 1.57 Number of observations 59 68 Note: Firms included in this regression are those for which at least 30 percent stakes were sold outright and firms were priva- tized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. management also appear to be positively with environmental information variable were related to number of tender rounds. not significant at any reasonable level. In the OLS model, however, site remediation For auction mechanisms, in the Tobit model planning was negatively and significantly the site remediation plan and the interaction related to the number of auction rounds Bluffstone and Panayotou 45 Does the Treatment of Environmental Liability during Privatization Really Matter? required to sell at least 30 percent of firms' industries had to be dropped from the regres- shares. These results are presented in appen- sions. dix D. Using binomial logit and probit models, we Site remediation plans were not associated first examined the relationship between with percentage of all shares purchased from whether any site remediation was reported the government at the time of privatization. and the provision of environmental informa- Only government ownership was significant, tion, including site contamination, in and it was negatively related to total percent- privatization packets. The provision of envi- age of shares purchased. The model that ronmental information is correlated with an included interactions between environmental increase in the probability that site information and the existence of a site remediation was reported, but this estimate remediation plan did not yield any additional was not significantly different from zero at any insights. Environmental information remained reasonable level. The same was true when a negatively and significantly correlated with dummy variable indicating that site contami- nation information had been provided to most the percentage of shares purchased at the time or all investors was used. of privatization. Table 6.12 presents the results of an analysis We were unable to run either logit or probit of whether preliminary audits were conducted models on the variable indicating whether (either by governments or by firms) and the privatization officials believed that environ- probability of site cleanup. The results, which mental issues delayed privatizations because are significant at the 4 percent level, suggest the site remediation plan dummy predicted that conducting a preliminary audit was the dependent variable perfectly. correlated with a roughly 15 percent increase in the probability that site remediation was SITE REMEDIATION reported. With only 21 percent of firms reporting cleanup, this effect is estimated to be Dependent variable: dummy quite large. Similar results were derived using variable indicating whether, accord- a probit model. This result confirms what ing to firm management, any site might be considered a reasonable presump- remediation had occurred since tion that site remediation is more likely to privatization occur if basic site investigation takes place.17 Relationship with provision of environmental For the conditioning variables, an increase in information government ownership of 10 percent was associated with a roughly 2 percent decline in In the overall sample, approximately 21 the probability of site remediation (p ~0). This percent of firms reported having engaged in result is highly significant and confirms the some site remediation. Because of perfect conventional wisdom that government-owned collinearity in all subsamples, the dummy firms in the region are less likely to be envi- variables for metal, mining, and petroleum ronmentally responsible. Firm efficiency 46 Environment Strategy Papers Econometric Results Table 6.12. Binomial logit regression. Dependent variable: Whether any site remediation had been conducted since privatization (dummy variable) Full model Final model Coefficient p-value Coefficient p-value Constant 38.1657 0.912125 ­1.98528 2.59E­5 (­2.72) Whether a preliminary audit was conducted 0.977829 0.115955 1.05949 0.043525 (dummy) (0.145) Year of privatization ­0.02008 0.907924 Chemical industry dummy variable ­0.43502 0.650452 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by government in ­0.01552 0.270687 ­0.01806 0.008008 1997 (­0.00248) Percentage of total shares held by largest private ­0.00822 0.437227 investor or group Percentage of total shares held by foreign 0.006112 0.649733 investors Revenues (in 1997 U.S. dollars) in year prior to 1.15E­8 0.060375 1.39E­8 0.01084 privatization (1.9E­9) More than one member of senior management 0.6138 0.243384 0.980728 0.039536 changed within one year of privatization (dummy) (0.135) Dummy variable for Hungary and Poland 0.940735 0.162192 Revenue (in 1997 U.S. dollars) per employee ­5.70E­6 0.274153 ­5.62E­6 0.145229 (­7.7E­7) Log likelihood ­49.0 ­60.1 Restricted log likelihood ­60.7 ­70.4 X2 23.3 22.7 Number of observations 117 140 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. measured as revenue per employee was also (larger firms are more likely to have site negatively correlated with site remediation, contamination problems because their opera- but this estimate was significant only at the 14 tions are larger) and of availability of resources percent level. for remediation. Firm revenues in 1997 were positively and Management turnover after privatization was significantly correlated with the probability of positively related to site remediation. Indeed, remediation at the 1 percent level. In this for firms that experienced management specification an increase in annual firm turnover in the year after privatization, the revenues in 1997 of US$10 million was probability that site remediation occurred predicted to increase the probability of site increased about 13.5 percent (p = 0.04), or remediation by 1.9 percent (p = 0.01). This about 61 percent of the mean. This result finding probably reflects the effects of size suggests that organizational change may be an Bluffstone and Panayotou 47 Does the Treatment of Environmental Liability during Privatization Really Matter? important precursor to site remediation and conclusion appears to be particularly justified perhaps other environmentally friendly for preliminary audits, which typically were behaviors. Indeed, rather than being a proxy done by investors. for management hostility and intransigence at the time of privatization (which would be Relationship with having a well defined expected to have a variety of negative results), cleanup plan, and the interaction of environ- management turnover may really capture the mental information and site remediation ability of new owners to make necessary plans changes in the firm and to innovate as needed. This section examines how the existence of The results for the relationship between the site remediation plans may be associated with performance of a secondary audit and the site remediation in the five countries ana- probability of site remediation being reported lyzed. Our hypothesis is that the existence of a are consistent with the previous regression. cleanup plan with clearly defined remediation The presence of a secondary audit is associ- measures at the time of privatization or shortly ated with an increased probability of site afterward should be positively associated with remediation of 13 percent in the final model. whether remediation was reported to have This estimate was only significant at the 14 occurred after privatization. The reasoning is percent level, but in the full model the that if planning with clearly defined measures estimated coefficient was much larger and was is done, firms and governments will have significant at the 6 percent level. In alternative invested in site remediation information and specifications the secondary audit variable is will also have agreed on the key measures that significant at better than the 10 percent level. need to be undertaken. Under such circum- Because these findings mirror closely those for stances, ambiguities are resolved, and some- preliminary audits, they are reported only in thing akin to a contract is put in place with appendix D. regard to site remediation. Firms in Hungary and Poland were estimated Table 6.13 presents results that confirm our to undertake more site remediation than firms hypothesis and are consistent with the previ- in other countries. Indeed, being located in ous findings in this section. As was generally these countries where reforms are advanced is found for environmental information, the estimated to increase the probability of existence of a site remediation plan at the time remediation by 11 percent, and this result was of privatization or shortly afterward was significant at the 8 percent level. strongly and positively associated with whether site remediation was reported to have A number of specifications were tried, includ- actually occurred. Logit model results sug- ing a variety of environmental information gested that the existence of a plan was associ- variables and firm characteristics. In general, ated with an increase of approximately 25 information variables exhibited a positive percent in the probability that site remediation relationship with reported site cleanup, was reported. This estimate was significant at suggesting that environmental information has less than the 1 percent level, implying that the a positive relationship with cleanup. This estimate is quite precise. The effect is also 48 Environment Strategy Papers Econometric Results Table 6.13. Binomial logit regression. Dependent variable: Whether any site remediation had been conducted since privatization (dummy variable) Full model Final model Coefficient p-value Coefficient p-value Constant 60.9511 0.858876 ­2.75607 7.68E­9 (­0.35) Whether a site remediation plan was developed prior 2.15422 0.006348 1.9537 0.007971 to or shortly after privatization (dummy) (0.255) Year of privatization ­0.03171 0.853791 Chemical industry dummy variable ­0.60251 0.566602 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by government in ­0.00563 0.697039 1997 Percentage of total shares held by largest private ­0.00656 0.541607 investor or group Percentage of total shares held by foreign investors 0.004385 0.722817 Revenues (in 1997 U.S. dollars) in year prior to 7.26E­9 0.044274 7.17E­9 0.031386 privatization (9.10E­10) More than one member of senior management 0.502243 0.356337 0.776173 0.104949 changed within one year of privatization (dummy) (0.0975) Dummy variable for Hungary and Poland 1.47552 0.048515 1.28894 0.008795 (0.162) Revenue (in 1997 U.S. dollars) per employee ­3.62E­6 0.335575 ­3.36E­6 0.249817 (­4.2E­7) Log likelihood ­47.4 ­59.7 Restricted log likelihood ­59.6 ­71.6 X2 24.5 23.7 Number of observations 118 147 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. large because in the sample only 21 percent remediation. Indeed, being in those countries of firms reported having engaged in increased the probability of remediation by a postprivatization remediation. Similar results very large 16.2 percent. This estimate was were derived using a probit model (coefficient significant at less than the 1 percent level. estimate 1.13). Change in management was once again positively associated with site remediation; Also positively associated with remediation management changes immediately after were firm revenues, but the magnitude of the privatization were correlated with a 10 relationship was rather small. Consistent with percent increase in the probability of site the regression with secondary audits as the remediation (p = 0.10). key independent variable, if a firm's home country was Hungary or Poland, the firm was We also estimated a model that included both more likely to have engaged in site the provision of environmental information Bluffstone and Panayotou 49 Does the Treatment of Environmental Liability during Privatization Really Matter? and the development of a site remediation ! Dummy variable indicating plan, as well as the interaction between the whether environment-related two variables. Including these variables legal actions had been brought together did not change the results. Environ- by governments against firms mental information provided at the time of since privatization privatization was again unrelated to reported site remediation. The existence of a site Relationship with provision of environmental remediation plan was again positively corre- information lated, and the interaction term was not significant at any reasonable level of confi- The literature suggests that providing informa- dence. tion on the environmental status of firms at the time of privatization should allow investors Substituting the preliminary audit variable, and government officials to agree on environ- which was significant in previous regressions, mental issues and address potential conflicts for the environmental information variable did before ownership is transferred. That should not change these results. Indeed, including reduce postprivatization conflicts. We first both audits and the site remediation plan examine whether reduced levels of adminis- variable in the same equation caused the audit trative enforcement actions such as fines, variable to become insignificant. We therefore notices of violation, and closures were conclude that in this case it is really the associated with the provision of information. existence of site remediation plans that has the We then examine the relationship with most influence on whether any site environment-related legal actions taken by the remediation actually occurs. government or the public against firms. We also asked firms about legal actions they may POSTPRIVATIZATION ENVIRONMENTAL have taken against the government, but no CONFLICTS firms had challenged the government in that way. Dependent variables: Table 6.14 presents the results of a logit model ! Dummy variable indicating that examined the relationship between the whether administrative enforce- provision of environmental information in ment actions had been taken privatization packets and the occurrence of against the firm for environmen- administrative actions after privatization. The tal violations dummy variables for mining and petroleum ! Number of environment-related sectors were dropped from the model because legal actions brought by the of perfect collinearity with the dependent public against firms since variable. The estimated relationship between privatization the administrative actions by the government ! Number of environment-related against the firm and the provision of environ- legal actions brought by govern- mental information in privatization packets is ments against firms since positive. Indeed, such provision is correlated privatization with a 3.4 percent increase in the probability 50 Environment Strategy Papers Econometric Results Table 6.14. Binomial logit regression. Dependent variable: Whether administrative enforce- ment actions had been brought by the government against the firm (dummy) Full model Final model Coefficient p-value Coefficient p-value Constant 1,604.89 0.01403 706.411 0.038189 (12.55) Whether environmental information, including 3.06518 0.016932 1.93778 0.02069 information on site contamination, was provided in (0.034) privatization packets (dummy) Year of privatization ­0.80487 0.014052 ­0.35469 0.038015 (­0.0063) Chemical industry dummy variable 0.391136 0.771072 Primary metal manufacturing industry dummy 0.87808 0.863497 variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by government in ­0.12529 0.043687 ­0.05071 0.039111 1997 (­0.0009) Percentage of total shares held by largest private ­0.01123 0.429311 investor or group Percentage of total shares held by foreign ­0.01252 0.426083 investors Revenues (in 1997 U.S. dollars) in year prior to 1.54E­8 0.063947 7.02E­9 0.072632 privatization (1.25E­10) More than one member of senior management ­1.6232 0.08889 ­1.2541 0.078859 changed within one year of privatization (dummy) (­0.022) Dummy variable for Hungary and Poland ­2.37678 0.028032 ­1.40472 0.06792 (­0.025) Revenue (in 1997 U.S. dollars) per employee 3.50E­6 0.510598 Log likelihood ­20.8 ­31.2 Restricted log likelihood ­42.7 ­48.6 X2 43.9 34.9 Number of observations 116 131 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. that environmental enforcement actions position to enforce environmental laws after occurred. This result was significant at the 2 firms are privatized; in this respect, the percent level in the final model and was occurrence of administrative actions should be robust to a variety of specifications.18 With seen as a positive side effect associated with only 9.3 percent of firms experiencing admin- provision of information to potential investors. istrative actions, provision of environmental Second, regulators may use the data to focus information was associated with a rather especially on those firms for which they have dramatic increase in actions. information. Finally, environmental informa- tion may have been provided for firms with At least three potential explanations can be more significant (but, for our purposes, suggested for this finding. First, governments unobserved) environmental problems. As that invested in information may be in a better already noted, environmental information may Bluffstone and Panayotou 51 Does the Treatment of Environmental Liability during Privatization Really Matter? at least partially serve as a proxy for environ- This section looks at legal, rather than admin- mental problems, and it would not be surpris- istrative, actions brought by the government ing that firms with more problems were cited against privatized firms. Because of conver- more often. gence problems in the Tobit, OLS is used, and the results should be interpreted with caution. In terms of conditioning variables, the prob- Although the OLS and Tobit results typically ability of environmental enforcement actions do not differ dramatically, the dependent appears to exhibit an autonomous downward variable is censured, with a lower bound of trend, but this effect seems to be of a small zero, so we know that the Tobit would be magnitude. Firms with higher percentages of most appropriate. government ownership are estimated to have lower probabilities of facing enforcement No relationship was detected between the actions, but again, the magnitude of the effect provision of environmental information in is small. We might speculate, though, that privatization packets and the number of some of the environmental regulatory distor- environmental legal actions since tions from the era of planned economies that privatization. The same was found for prelimi- caused environmental problems in Central nary and secondary environmental audits. A European countries persist when governments rather strong relationship between the provi- sion of written information on site contamina- hold onto ownership. Perhaps government tion to most or all investors and the number of ownership partially insulates firms from environmental legal actions was estimated to enforcement actions, as it did during the exist. In the final model, firms for which site planned economy period. contamination information was provided are estimated to experience about three times the Higher-revenue firms faced higher probabili- environmental legal actions of the average ties of enforcement action (p = 0.07). Firms firm. This estimate was significant at the 8 that experienced management turnover percent level, suggesting that we can be fairly immediately after privatization were much sure that governments are using site contami- less likely to be the subject of enforcement nation information for legal purposes. actions, with the probability of citation being 2 percent lower (p = 0.08), all else being Other significant, or close to significant, equal. This result is in accord with the appar- variables included the year of privatization. ently important--but still fundamentally An increase of one year in time of mysterious--relationship between manage- privatization is estimated to reduce the ment turnover and better privatization and site number of legal actions by 0.07, approxi- remediation outcomes. Hungarian and Polish mately 50 percent of the mean number of firms were less likely to experience actions actions (p = 0.13). Government ownership than firms in other countries, with a 2.5 share was positively correlated with the percent lower chance of these events occur- number of legal actions (p = 0.13). An ring (p = 0.07). This result could be due either increase in the government share of the firm of to better environmental behavior or to unwill- 10 percent (about one-third compared with ingness of regulators to impose sanctions. the mean) was correlated with an increase in 52 Environment Strategy Papers Econometric Results Table 6.15. OLS regression. Dependent variable: Number of environment-related legal actions brought by the government against firms since privatization Full model Final model Coefficient p-value Coefficient p-value Constant 160.734 0.065021 139.451 0.126861 Whether written information about site contamination was provided to investors (dummy) 0.521366 0.029537 0.432908 0.079874 Year of privatization ­0.08079 0.064695 ­0.07004 0.126726 Chemical industry dummy variable 0.603675 0.303923 0.700813 0.312455 Primary metal manufacturing industry dummy variable 0.617225 0.214175 Mining and oil and gas extraction dummy variable ­0.44008 0.16959 ­0.30961 0.181413 Petroleum and coal products industry dummy variable ­0.29543 0.306207 Percentage of total shares held by government in 1997 0.005493 0.058257 0.004718 0.127533 Percentage of total shares held by largest private investor or group 0.001142 0.648807 Percentage of total shares held by foreign investors 0.010013 0.056792 0.010749 0.054266 Revenues (in 1997 U.S. dollars) in year prior to privatization 3.95E­9 0.296393 More than one member of senior management changed within one year of privatization (dummy) ­0.16169 0.246903 ­0.2069 0.2182 Dummy variable for Hungary and Poland 0.198241 0.158453 0.20752 0.142714 Revenue (in 1997 U.S. dollars) per employee ­1.25E­6 0.186111 R2 0.30 0.20 F statistic 3.77 4.17 Mean of dependent variable 0.16 0.15 Number of observations 129 146 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. the number of actions of 0.04 (about 25 A logit model was also run on a dummy percent of the mean). variable indicating whether any environment- related legal actions had been taken by the Foreign ownership share was positively government against firms. The model could correlated with the number of legal actions. A only be run for the two audit variables be- 1 percent increase in foreign ownership share cause when full-rank matrices were created, (about 11 percent of the mean) was correlated the other environmental information variables with an increase in the number of incidents of explained this dummy variable perfectly. The about 6 percent (p = 0.05). This relationship results with the preliminary audit variable also held when a dummy variable indicating were consistent with the OLS results but for the existence of foreign ownership was used instead of the percentage of foreign owner- preliminary audits were significant only at the ship. Firms in Hungary and Poland experi- 18 percent level. The coefficient estimate for enced more environmental legal actions than secondary audits was not significant at any other countries. Being located in those reasonable level. countries was correlated with an increase in environmental actions of 0.21, approximately Relationship with having a well defined 125 percent of the mean. This estimate was, cleanup plan, and the interaction of environ- however, significant only at the 14 percent mental information and site remediation level. plans Bluffstone and Panayotou 53 Does the Treatment of Environmental Liability during Privatization Really Matter? Our hypothesis in this section is that the measures of pollution the changes in wastewa- existence of a cleanup plan with clearly ter and particulate emissions during the period defined remedial measures would reduce 1994­97 per dollar of firm revenues in 1997. postprivatization conflicts. In none of the This measure adjusts for firm size and is an regressions on any of the four conflict vari- often-used measure of environmental effi- ables did we find a relationship with the ciency. A large positive change indicates existence of a site remediation plan.19 improved environmental performance. Positive environmental measures are the POSTPRIVATIZATION ENVIRONMENTAL establishment of internal environmental PERFORMANCE audits, waste management or pollution prevention programs, and candidacy for ISO Dependent variables: 14000 series certification. ! Change in wastewater emissions An issue with our data is that most firms were (in tons) during the period 1994­ privatized after 1994. Any change in emis- 97 as a proportion of 1997 firm sions or establishment of internal environmen- revenues tal programs may therefore have taken place, ! Change in particulate emissions wholly or partly, before the date of (in tons) during the period 1994­ privatization. The implication of this point is 97 as a proportion of 1997 firm that we cannot really interpret any positive revenues behavior as being the result of steps taken at ! Dummy variable indicating the time of privatization (in the way, perhaps, whether internal environmental that we could in sections 6.2 through 6.6). audit, waste minimization, or pollution prevention programs Romanian firms are omitted from this section had been implemented because of a coding error. To remedy the ! Dummy variable indicating problem, it will be necessary to manually whether an environmental plan examine all questionnaires from the full had been formulated Romanian data set of approximately 600 firms ! Dummy variable indicating and attempt to match them with our privatized whether steps had been taken firms. This section therefore covers only four toward obtaining ISO 14000 countries. series certification Relationship with provision of environmental This section examines the relationship be- information tween two sets of policies--provision of information and site remediation planning-- Ongoing environmental performance and and postprivatization environmental perfor- environmental information are potentially mance. Two types of variables focus on linked because information may help firms environmental performance: reported pollu- inventory and address current environmental tion emissions, and key indicators of positive problems. Government provision of informa- environmental behavior. We utilize as our tion may also signal government interest in the 54 Environment Strategy Papers Econometric Results environmental performance of a firm and for example, to be an autonomous year-on- serve as a starting point for negotiations and year decline in wastewater emissions, with regulatory efforts with regard to environmental firms privatized one year later estimated to performance. have declines in wastewater emissions that are 13 percent larger than the mean decline of The results in table 6.16 suggest that no 0.0036 tons per dollar of revenue (p = 0.04). relationship exists between the provision of environmental information at the time of Larger government participation was associ- privatization and the change in wastewater emissions per dollar of revenue generated in ated with larger declines in wastewater 1997. The same result was found if the emissions. A 1 percent increase in government provision of written site contamination ownership was estimated to be associated information to most or all investors or the with a 4 percent larger drop in emissions per existence of environmental audits was used as dollar of revenue compared with the mean (p a measure of information. = 0.04). This result somewhat contradicts our conclusions with regard to site remediation-- Several conditioning variables appear, how- that higher levels of government ownership ever, to be closely associated. There seems, implied poorer environmental performance. Table 6.16. OLS regression. Dependent variable: Change in wastewater emissions, 1994­97, as a proportion of firm revenues in 1997 Full model Final model Coefficient p-value Coefficient p-value Constant ­0.97546 0.039905 ­0.930035 0.035184 Whether environmental information, including ­0.0011 0.519249 ­0.0009811 0.51019 information on site contamination, was provided in privatization packets (dummy) Year of privatization 0.000491 0.039876 0.0004681 0.035151 Primary metal manufacturing industry dummy 0.001489 0.72309 variable Mining and oil and gas extraction dummy variable ­0.00651 0.0071 ­0.0059156 0.008487 Percentage of total shares held by government in 0.000148 0.081818 0.00015404 0.044879 1997 Percentage of total shares held by foreign 4.38E­5 0.139862 4.04E­5 0.142997 investors Revenues (in 1997 U.S. dollars) in year prior to ­8.93E­12 0.743763 privatization Dummy variable for Hungary and Poland ­0.0035 0.242148 ­0.0036253 0.164265 Revenue (in 1997 U.S. dollars) per employee ­3.28E­9 0.395668 ­4.04E­9 0.011841 Frequency of internal emissions monitoring (6 = 0.000752 0.046458 0.00082272 0.029671 > once a day; 1 = once a year; 0 = never) Frequency of monitoring by regulators (once in __ ­0.00011 0.206262 ­9.27E­5 0.238504 months) Whether inspections are generally announced ­0.00086 0.645809 (dummy) R2 0.35 0.35 F statistic 1.37 2.11 Mean of dependent. variable 0.0037 0.0036 Number of observations 43 46 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 55 Does the Treatment of Environmental Liability during Privatization Really Matter? Foreign ownership was associated with larger Approximately 60 percent of firms reported declines in emissions, but this estimate was that they had introduced internal environmen- only significant at the 14 percent level. tal audit, waste management, or pollution prevention programs. These measures are More frequent internal monitoring was considered important practical steps toward positively correlated with wastewater declines exploiting win-win solutions to environmental (p = 0.03), as was more frequent monitoring problems. In the full model the industry by regulators (p = 0.24), but most other dummies were omitted because they perfectly conditioning variables were associated with explained the dependent variable. increased emissions. For example, firms in the mining sector had declines in emissions that The results in table 6.17 suggest that inclusion were 164 percent lower than for other firms of environmental information in privatization (p ~0). packets was highly positively and significantly correlated (at the 7 percent level) with the Firms located in Hungary and Poland were existence of these programs. The effect of estimated to have declines in wastewater providing environmental information was also emissions per dollar of revenue that were 100 large: the provision of environmental informa- percent lower than for firms in other countries, but this estimate was only significant at the 16 tion that included information on site contami- percent level. Of the several possible explana- nation in privatization packets was correlated tions for this finding, perhaps the most likely is with about a 20 percent increase in the that in our data, firms in these countries had probability that internal environmental lower wastewater emissions in 1994, and all management programs would be put in place else being equal, declines from these lower in firms. Compared with the mean, the effect base emissions would be expected to be is about a 30 percent increase. smaller three years later. This result suggests that providing information Efficiency, measured as revenue per em- at the time of privatization signals a proactive ployee, was correlated with lower wastewater government approach to environmental declines, but we believe that this is an artifact performance in industries that often have of our decision to normalize the change in environmental problems. We have already emissions using 1997 firm revenues. seen that information is correlated with a higher probability of site remediation and the Degrees of freedom were insufficient to run use of environmental regulatory instruments. It models for sulfur dioxide and BOD emissions. seems from this result that bringing in environ- Results for particulate emissions were similar mental issues early in the relationship with to those for wastewater emissions, and we investors also makes owners more likely to therefore conclude that no relationship exists institute programs that may improve their between any of our pollution variables and postprivatization environmental performance. any of our environmental information vari- In addition, instituting such programs helps ables. Results for particulates are presented in insulate firms from criticism for poor environ- appendix D. mental citizenship. 56 Environment Strategy Papers Econometric Results Table 6.17. Binomial logit regression. Dependent variable: Whether firms had introduced internal environmental audits, waste minimization, or pollution prevention programs (dummy) Full model Final model Coefficient p-value Coefficient p-value Constant 27.9826 0.917416 ­0.72351 0.157528 (­0.133) Whether environmental information, including 1.42804 0.101739 1.05904 0.065686 information on site contamination, was provided in (0.195) privatization packets (dummy) Year of privatization ­0.01383 0.918677 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Percentage of total shares held by government in 0.007034 0.707867 1997 Percentage of total shares held by foreign 0.002432 0.85612 investors Revenues (in 1997 U.S. dollars) in year prior to 1.66E­10 0.988672 privatization Dummy variable for Hungary and Poland ­0.51687 0.584491 Revenue (in 1997 U.S. dollars) per employee ­1.63E­6 0.468323 Frequency of internal wastewater emissions 0.210038 0.366934 0.372395 0.022652 monitoring (6 = > once a day; 1 = once a year; (0.0686) 0 = never) Frequency of monitoring by regulators (once in 0.043165 0.491079 0.075401 0.15817 __ months) (0.0139) Whether inspections are generally announced ­0.05952 0.941961 (dummy) Log likelihood ­27.8 ­40.3 Restricted log likelihood ­32.8 ­47.7 X2 8.13 14.7 Number of observations 62 79 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. The other significant variable was frequency of programs of 1.3 percent. The estimate is internal monitoring by firms; increases in significant at the 15 percent level. internal monitoring were correlated with an increased probability of introduction of These results suggest that internal environ- environmental management programs (p = mental management programs are comple- 0.02). By contrast, frequency of monitoring by ments with internal monitoring and substitutes regulators was associated with a decrease in for external monitoring. The first conclusion is the probability that environmental manage- not surprising; better internal monitoring of ment programs would be put in place. A delay emissions is generally an important compo- of one month, for example, was associated nent of internal environmental management with a decrease in the probability of the programs. The second finding supports our presence of environmental management conclusion with regard to the provision of Bluffstone and Panayotou 57 Does the Treatment of Environmental Liability during Privatization Really Matter? environmental information, in that environ- mental performance.21 All environmental mental management systems seem to insulate information variables were tried in the model, firms from regulatory scrutiny. Firms that but the provision to most or all investors of institute management programs are generally written information on site contamination was regarded as better environmental performers, the only one that was significant. It was and it seems that one implication of having associated with an increase of about 19 this label is less frequent wastewater monitor- percent in the probability of firms' having ing by regulators.20 environmental plans after privatization. As only 46 percent of firms had environmental Table 6.18 presents a binomial logit model to plans, this effect is quite large. This result was explain the existence of environmental plans significant at the 10 percent level. in firms. Like environmental management systems, these plans are considered to be Frequency of water pollution monitoring by important measures for improving environ- regulators was negatively and highly signifi- Table 6.18. Binomial logit regression. Dependent variable: Whether firm had an environmen- tal plan in place (dummy) Full model Final model Coefficient p-value Coefficient p-value Constant 125.058 0.74255 ­0.35237 0.416683 (­0.0747) Whether written information on potential site 0.833152 0.261763 0.904115 0.101978 contamination was provided to most or all (0.1919) investors (dummy) Year of privatization ­0.06318 0.74084 Primary metal manufacturing industry dummy ­2.76782 0.11528 ­1.70775 0.23068 variable (­0.362) Mining and oil and gas extraction dummy variable Percentage of total shares held by government ­0.0395 0.057294 ­0.024 0.062696 in 1997 (­0.0051) Percentage of total shares held by foreign ­0.00884 0.465251 investors Revenues (in 1997 U.S. dollars) in year prior to 1.12E­8 0.371969 privatization Dummy variable for Hungary and Poland ­0.17807 0.829021 Revenue (in 1997 U.S. dollars ) per employee 3.49E­6 0.714199 Frequency of internal emissions monitoring (6 0.207028 0.311633 = > once a day; 1 = once a year; 0 = never) Frequency of wastewater monitoring by 0.282658 0.006759 0.224431 0.000606 regulators (once in __ months) (0.0476) Whether inspections are generally announced 1.78951 0.042637 0.627044 0.284311 (dummy) (0.133) Log likelihood ­32.0 ­51.2 Restricted log likelihood ­47.8 ­63.4 X2 31.6 24.6 Number of observations 77 97 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. 58 Environment Strategy Papers Econometric Results cantly correlated (p ~0.0) with the presence ISO 14000 certification. The same result was of environmental plans. A delay of one month found when using any of the other environ- in monitoring was correlated with a 5 percent mental information variables, including audits. increase in the probability of an environmen- tal plan being in place. This result is consistent More frequent internal monitoring by firms with that found for environmental manage- was positively correlated with ISO 14000 ment systems. It seems, therefore, that envi- certification (p = 0.15), as were unannounced ronmental plans also substitute for monitoring visits by environmental inspectors. Indeed, by regulators. companies facing unannounced visits were 20 percent more likely to be seeking ISO 14000 Greater government ownership was associated certification than other firms (p = 0.02). with a decrease in the probability that firms Larger firms were more likely than smaller had environmental plans in place. A 10 firms to seek certification (p ~0). percent increase in government ownership was correlated with a 5 percent decline in the Firms in Hungary and Poland were substan- probability of a firm's having an environmen- tially less likely (13 percent) to seek ISO tal plan (p = 0.06). This result is consistent 14000 certification than were firms in other with our findings related to site remediation countries (p = 0.10). This finding is rather and pollution emissions, which suggest that surprising in view of the important role of more government involvement is correlated foreign direct investment in Hungary and with worse environmental performance. Poland. Perhaps, given these countries' relatively longer histories of economic open- We now turn to analysis of firms seeking ISO ness and their reputations for better environ- 14000 series certification. This, like the Eco- mental performance, their firms had less need Management and Audit Scheme (EMAS) in the to seek certification. European Union framework, is a rigorous and potentially important certification of respon- Also negatively related to ISO 14000 certifica- sible environmental production methods. It tion was government ownership share. In the tells downstream users that a series of comple- final model, a 10 percent increase in govern- mentary environmental protection measures ment ownership was correlated with a 4.4 was used in production processes. Firms percent decline in the probability that a firm seeking ISO 14000 series certification often do was seeking ISO 14000 series certification (p so to attract environmentally conscious = 0.02). This finding supports our other consumers. results showing that more government in- volvement does not seem to be correlated In our analysis we were again unable to with the use of more progressive and innova- include the industry dummies in the regres- tive environmental tools. Indeed, the opposite sion, due to collinearity with the dependent appears to be the case. variable. We see from the logit model results presented in table 6.19 that there does not Relationship with having a well defined appear to be a relationship between environ- cleanup plan, and the interaction of environ- mental information and whether firms sought mental information and site remediation plans Bluffstone and Panayotou 59 Does the Treatment of Environmental Liability during Privatization Really Matter? Table 6.19. Binomial logit regression. Dependent variable: Whether firm management reported seeking iso 14000 certification (dummy) Full model Final model Coefficient p-value Coefficient p-value Constant ­1,077.91 0.126517 ­2.35381 0.024511 (­2.35) Whether environmental information, including ­2.2956 0.132172 0.029233 0.972466 information on site contamination, was provided in (0.0029) privatization packets (dummy) Year of privatization 0.539643 0.127104 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Percentage of total shares held by government in ­0.05622 0.120647 ­2.35381 0.024511 1997 (­0.0043) Percentage of total shares held by foreign 0.00141 0.941968 investors Revenues (in 1997 U.S. dollars) in year prior to 4.02E­8 0.025599 2.18E­8 0.005176 privatization (2.18 E­9) Dummy variable for Hungary and Poland ­2.24561 0.081655 ­1.3251 0.106672 (­0.132) Revenue (in 1997 U.S. dollars) per employee ­5.93E­6 0.555861 Frequency of internal emissions monitoring (6 = 0.458829 0.155264 0.295165 0.15352 > once a day; 1 = once a year; 0 = never) (0.0294) Frequency of monitoring by regulators (once in 0.03351 0.627283 __ months) Whether inspections are generally announced 2.45078 0.034128 2.02523 0.019328 (dummy) (0.202) Log likelihood ­16.6 ­24.1 Restricted log likelihood ­30.5 ­35.9 X2 27.8 23.6 Number of observations 62 74 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Numbers in parentheses are marginal effects computed at the means of independent variables. Our data set does not include a variable that establishment of environmental management explicitly mentions requirements for ongoing programs. environmental performance in privatization contracts. The existence of a site remediation The existence of a site remediation plan with clearly defined measures and plan was not significantly correlated with the responsibilities may be similar, however, in change in any of our pollution variables per that it reflects a potentially high level of care dollar of revenue generated in 1997. It seems, and the inclusion by privatization agencies of therefore, that this type of measure is unre- environmental issues in privatization pro- lated to postprivatization pollution behavior. cesses. Our hypothesis is that the existence of Illustrative results for particulate emissions are such cleanup plans is positively correlated included in appendix D. No relationship was with environmental improvements, such as detected between the existence of a site declines in pollution over time, and with remediation plan and any of the positive positive environmental measures, such as the environmental behaviors--internal environ- 60 Environment Strategy Papers Econometric Results mental management systems, environmental similar to those presented in the previous plans, or ISO 14000 series certification. subsection. Logit model results are therefore Results for the conditioning variable were presented only in appendix D. Bluffstone and Panayotou 61 Chapter 7 Conclusions T his report focuses on firms in industries cases where such information was given, that are likely to have site contamina- environmental information (including informa- tion and other environmental prob- tion on site contamination) was also provided. lems. These firms in "dirty" industries tend to Less frequently, written information explicitly be large, with high employment levels. Most focusing on site contamination problems was investors in these firms purchased majority offered. This does not mean that at the time of shares at the time of privatization. Few of these privatization the information did not exist; investors were foreigners; only 25 percent of indeed, it appears that almost half of investors firms indicated that foreign investors had any themselves financed preliminary audits, and 14 stake at all. In general, governments have percent paid for secondary audits. maintained significant stakes in these firms, with an average of 35 percent government Although indemnification is sometimes advo- ownership after privatization. cated as a particularly useful risk reduction tool, in only a few cases did privatization Privatization of firms in the selected industries agencies do anything but pass 100 percent of generally took less than a year, and it was quite environmental liabilities to new owners (see, unusual for privatization agency officials to for example, Boyd 1996). Using the modeling report that environmental issues had delayed framework of Bluffstone and Panayotou (2000), privatizations. It is also unlikely that it seems that symmetric information between preprivatization management teams erected investors and privatization agencies was roadblocks to privatization, since current presumed by policy. Price discounts were very management rarely believed that the old rarely explicitly given, leaving us to conclude management was hostile to privatization. that not much overt material consideration was Despite this apparent lack of opposition, after given to investors buying into these generally privatization, changes in management were dirty industries. Investors appear to have been quite common.22 on their own in finding out about any site contamination problems and negotiating with General information on firms was typically privatization agencies as best they could. The provided to investors, and in one-third of the likely consequences were probably lower 63 Does the Treatment of Environmental Liability during Privatization Really Matter? prices and less foreign investment, but we privatization prices. Firms in primary metal were unable to test this hypothesis explicitly industries had systematically lower prices per because indemnifications and discounts were dollar of 1997 revenue, and a year-on-year virtually never offered. autonomous decline in prices equal to 70 percent of the mean was also observed. If investors themselves can be believed, it appears that at least some site remediation has The amount of privatization revenues, defined taken place. Approximately 20 percent of as the price per share times the number of firms in the sample reported being engaged in shares sold at the time of privatization, offers a some type of site remediation. Over half of more comprehensive measure of the budget- them said that remediation had been com- ary effects of privatization programs. The pleted, and most of the rest reported that they analysis suggests that by itself, provision of planned to complete it. Although we do not environmental information at the time of know what level of remediation is planned or privatization was correlated with increased implemented, these results to some degree revenues, but this result was significant only at call into question the claim that domestic the 14 percent level. The most robust and investors are unlikely to engage in site interesting results were when both environ- cleanup and other positive environmental mental information and the existence of a site activities (see, for example, Auer, Reuveny, remediation plan were included in the model and Adler 2001). and interacted with each other. In this case privatization revenues were estimated to triple Many firms in the sample engaged in a variety when environmental information, including of positive environmental behaviors. Firms site contamination, was provided and then have internal environmental management followed up with a site remediation plan. Also systems or environmental plans and are of interest is that when the interaction term seeking ISO 14000 series certification. On was included, neither environmental informa- average in the sample, emissions of wastewa- tion alone nor the site remediation plan ter and key pollutants declined significantly variable was significant. This result suggests during the period 1994­97. that a key tool in selling large numbers of shares at higher prices (the two components of The regression results generally validate the privatization revenues) is not just to provide supposed best practices. The literature tells us environmental information but also to truly that making investors aware of potential address site remediation problems through problems should lead to higher prices. Our planning. results suggest that providing investors with environmental information in privatization Metallurgical companies appeared once again packets was associated with increased to be heavily discounted. Firms in Hungary privatization prices, measured as share price and Poland generated four times lower as a proportion of revenues in the year prior to revenues, compared with the mean. privatization. At the mean, the provision of environmental information by governments Foreign investors did not appear to respond to was associated with a 173 percent increase in provision of environmental information by 64 Environment Strategy Papers Conclusions governments; there was no relation between tal information, including information on site the provision of environmental information contamination, was provided sold 10 percent and the percentage of foreign ownership in fewer shares, which is equivalent to about 16 firms. This result may suggest credibility issues percent of the mean block of shares sold. associated with Central European privatization agencies. The same was true for environmen- How is this evidence to be interpreted? It is tal audits. Only the size of total firm revenues possible that once site contamination issues in 1997 (i.e., larger companies) and firms in are raised, due diligence processes take some the chemical industry were positively associ- time to complete. Alternatively, the causality ated with foreign ownership share or the may be reversed if environmental information existence of foreign ownership in a company. tends to be provided for firms with more environmental problems. Finally, it is possible It is sometimes said that privatization agencies that environmental information is a proxy for often do not provide environmental informa- noninsider privatization processes. It certainly tion to investors and do not encourage them seems likely that insider deals would tend to to investigate sites because officials worry that have no environmental information provided privatization processes will be slowed. Yet the and would be quicker than other modes of literature suggests that more information will privatization. facilitate privatizations by reducing risk. The results indicate that provision of environmen- Once again, much more insight is gained tal information by governments to investors is when we include the site remediation plan indeed associated with slower privatization variable and its interactions with environmen- processes, and they seem to be very robust to tal information in one model. Site remediation a variety of measures of privatization effi- plans were associated with shorter ciency. All results were significant at least at privatization times, particularly in the OLS the 5 percent level. model. Fewer auction rounds were also associated with the existence of site For example, when environmental information remediation plans. By far the most powerful was provided, at the mean firms are estimated variable, however, was the interaction term to have taken four months longer to privatize. representing cases in which environmental This difference represents an increase of 50 information was provided and a site percent compared with the mean privatization remediation plan was developed prior to or time of eight months. Providing environmental shortly after privatization. The independent information was also associated with in- effect of environmental information was creased numbers of tender rounds required-- associated with about a 50 percent increase in about 45 percent of the mean value of 1.5 privatization time compared with the mean, rounds--and with an increase in the number but when both instruments were used, of auction rounds that was equivalent to about privatization time was estimated to be re- 70 percent of the mean. When environmental duced by 83 percent of the mean (p = 0.11). information was provided, smaller shares of Although the significance level is only 11 companies seem to have been sold at the time percent, we are reasonably sure that the of privatization. Firms for which environmen- efficiency-enhancing effect of the two policies Bluffstone and Panayotou 65 Does the Treatment of Environmental Liability during Privatization Really Matter? together outweighs the predicted delays associated with site remediation after associated with the provision of environmental privatization. Preliminary and secondary information, since other models (OLS and audits do, however, seem to be important. Tobit adjusted for heteroskedasticity) gave Both types of audit were associated with very similar results. increases in the probability of postprivatization remediation of about 14 When privatization time is measured by percent, about 67 percent of the mean prob- number of tender or auction rounds, the ability of site remediation of 21 percent.23 The results are very similar. There does not appear effect of site remediation planning appears to to be an independent relationship between be even more important. Firms that developed number of tender rounds needed to privatize site remediation plans prior to or shortly after and either environmental information or the privatization had a 25 percent greater prob- development of site remediation plans, but the ability that site remediation occurred, com- use of both policies together seems to be pared with firms with no plans in place. This important. The results indeed suggest that use effect is about 120 percent of the mean. Taken of both instruments together reduced the together, these results suggest that to actually number of tender rounds needed to privatize spur site remediation after privatization, active by about 174 percent of the mean number of and concerted measures are needed. rounds; the estimate is significant at about the p = 0 level of confidence. It therefore seems Two related points are also of note. First, the that combining these policies offers an results suggest that the environmental payoffs opportunity to significantly improve the from audits and remediation planning are efficiency of privatization markets. extremely high. These types of measure appear to dramatically increase the chance Taken in its full context, it appears clear that that remediation will be undertaken after the generally accepted best practice of privatization. Second, most audits were providing environmental information at the financed by investors, and most remediation time of privatization does indeed speed up plans were probably prepared by firms either privatization. For firms with contamination before or after privatization. Firms themselves problems, this information must be accompa- then tended to follow up with remediation. nied by concrete plans for addressing the problems; otherwise the information may We of course do not know what factors delay privatizations. motivated those decisions, but it seems reasonable to suppose that firms were acting A particularly important reason for providing in their own interests, either by improving the site contamination and other environmental work environment or by protecting themselves information to investors and for developing from potential future liabilities. Firms that felt site remediation plans is to facilitate site they had nothing to gain from site remediation remediation after privatization. We find that were not likely to invest in information and the simple provision of environmental infor- therefore did not remediate. Many of these mation (either general or focused particularly firms may have been left with unremediated on site contamination) is not significantly site contamination that in the future could 66 Environment Strategy Papers Conclusions have negative effects on society.24 With such a to most or all investors, the number of seemingly strong link between environmental postprivatization environment-related legal audits, site remediation planning, and site actions against firms roughly tripled. No other cleanup, it probably would be in the interests environmental information variables were of Central European societies to finance more correlated with postprivatization conflicts, nor environmental audits and the development of was there correlation between conflicts and site remediation plans prior to privatization for the existence of site remediation plans. firms where cleanup seems desirable. Management turnover after privatization was Firms in Hungary and Poland were much associated with a 2 percent lower probability more likely than firms in other countries to that firms had been subjected to administrative engage in cleanup. Location in either of these enforcement actions (about 22 percent of countries was associated with a 16 percent those reporting some type of action). This increase in the probability that some result, in combination with previous findings, remediation was reported, which was 76 again suggests that a privatization environ- percent of the mean probability of 21 percent. ment in which changes can be made by firm Firms with more revenues were more likely to management is important for economic and report remediation, as were firms that reported environmental performance. significant management turnover right after privatization. For these firms, the probability Larger shares of government ownership that remediation was reported was approxi- appear to at least somewhat insulate firms mately doubled. Management turnover from environmental administrative actions therefore appears to signal positive environ- after privatization. By contrast, the share of mental behavior and, judging from the foreign ownership was positively related to analysis of privatization efficiency, positive environmental legal actions. Firms in Hungary economic behavior as well. and Poland were 2.5 percent less likely to face administrative actions than firms in other There is some indication that governments countries--quite a large effect, given that less have utilized the environmental information than 10 percent of firms in the sample faced provided to investors at the time of any sanctions. privatization for environmental enforcement purposes. The provision of environmental Our data on pollution emissions are limited, information in privatization packets was and we hesitate to draw strong conclusions, associated with a 3.4 percent increase in the but there does not appear to be a very strong probability of administrative actions such as relationship between postprivatization envi- notices of violations, fines, and closure. With ronmental behavior and provision of the types 9.3 percent of the sample reporting some type of environmental information we have been of sanction, environmental information was discussing. For example, adjusting for other correlated with a substantial increase in the variables that are likely to be important, our chance that a firm experienced administrative environmental information variables were not actions. Similarly, when written information associated with reductions in wastewater or on potential site contamination was provided particulate emissions. The same was true for the existence of site remediation plans. Bluffstone and Panayotou 67 Does the Treatment of Environmental Liability during Privatization Really Matter? Environmental information was, however, emissions. These outcomes were also corre- positively associated with some environmental lated with higher probabilities of environmen- behaviors that in the future could yield results. tal management systems being in place. These For example, the provision of environmental effects were estimated to be quite large. information in privatization packets was associated with a 20 percent increase in the It seems clear from these results that the probability that firms had environmental audit generally accepted best practices for dealing and other internal environmental management with environmental liabilities at the time of programs in place after privatization (29 privatization indeed have important benefits. percent of the mean of 68 percent of firms Providing environmental information seems to with such programs in place). It was also increase privatization revenues and yield associated with a 19.2 percent increase in the higher prices for firms. Although the provision chance that firms had an environmental plan of environmental information may stimulate in place but was completely unrelated to additional investigations that take time and whether firms sought ISO 14000 series somewhat delay privatizations, for firms with certification. more site contamination problems, the creation of site remediation plans actually Most of the action was with the conditioning appears to be correlated with faster variables. For the two pollutants studied, there privatizations. When environmental informa- appears to be a strong autonomous trend tion and site remediation plans are combined, toward decreased emissions. For example, all privatization is more efficient on a variety of else being equal, a one-year delay in indicators. privatization was associated with 13 percent larger reductions in wastewater emissions. A More extensive investigations, such as audits higher government ownership share implied and site remediation planning, appear to be larger wastewater reductions but lower related to remediation of contaminated sites. reductions in particulates. (A 10 percent The process of generating information also increase in government share was correlated seems to assist governments in their environ- with a decline that was 50 percent smaller.) mental enforcement efforts. Although it may Government share was also correlated with be true that governments focus more attention lower probabilities that environmental plans on firms for which they have information, we would be in place or that firms were seeking must recall that all the firms considered in this ISO 14000 series certification. With the report are in typically highly polluting indus- exception perhaps of wastewater reductions, it tries, and it is likely that information was seems that government retention of shares has provided for firms with more serious prob- not been good for the environment. lems. Greater enforcement attention may therefore not be a problem. The data also suggest that internal monitoring was important in triggering emissions reduc- Despite some caveats, on the whole our tions. All else being equal, more frequent empirical results strongly support the gener- internal monitoring was associated with bigger ally accepted best practices for addressing wastewater declines and lower particulate environmental liabilities at the time of 68 Environment Strategy Papers Conclusions privatization. Governments would do well to to address these problems vigorously. The investigate and bring to light environmental evidence suggests that economic as well as issues in firms that are likely to have or known environmental benefits will result. to have problems and to work with investors Bluffstone and Panayotou 69 Appendix A Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic HUNGARY As is the case in many other Central and Eastern European countries, the civil code of Hungary's privatization program began in 1990, when the State Property Agency was Hungary provides that when property is created to prepare enterprises for privatization transferred, all the rights and duties associ- and complete sales of firms. The State Asset ated with that property change hands as well. Holding Company (AVRt) was established in Early in the Hungarian privatization program, late 1992 to manage enterprises still in state environmental issues were generally ignored hands. With the passage of an amended law unless an investor (usually a foreigner) raised on privatization in 1995, the two institutions the topic (Goldenman 1997; Csanádi and Bell were merged into the Hungarian Privatization 1999). Negotiations with investors on environ- and State Holding Company (APVRt), which mental issues have tended to be ad hoc and took on the functions of its precursor agencies have sometimes generated problems (see box (Goldenman 1996). A.1). BOX A.1 Hungary: A low price for a river-contaminating plant In 1992 the Kordax firm, which was renting the facility of the former Eptek Company, a mineral oil processing and trading company, expressed a desire to purchase the plant. The value of the facility was estimated at US$4 million. During the negotiations, oil was observed floating on the Danube River near the plant. The Hungarian Privatization and State Holding Company (APVRt) appeared ready to take responsibility for the problem by offering Kordax a price discount based on the estimated liability. The APVRt had, however, recently hired an environmental expert, who recommended that an audit be conducted to determine responsibility before a price discount was offered. The US$100,000 study revealed that 96 percent of the responsibility for the contamination lay with Kordax. Kordax ultimately purchased the facility and is believed to have paid substantially less than US$4 million. Part of the reason for the lower price was that Kordax was able to use the existence of the contamina- tion to discourage other investors and create concern within the APVRt that waiting for a better offer was unwise. In 2001 company management was charged with tax evasion, and the firm was closed. 71 Does the Treatment of Environmental Liability during Privatization Really Matter? A well-known instance was the 1993 pur- time that shares were sold internationally chase of the Lehel Refrigerator Factory by the using a leading international stock exchange. Swedish firm Electrolux. In response to To prepare for the sale, the APVRt commis- Electrolux's concerns, but without an envi- sioned a full-scale environmental audit with ronmental audit or clearly defined cleanup the explicit goal of increasing the company's standards, the privatization agency agreed to value. As Tillson (1996) notes, finance all site remediation costs, and 60 percent of the privatization proceeds were MOL Rt used the findings of the placed in escrow for that purpose. Electrolux [environmental] assessment to focus spent all the funds in the escrow account and resources on those parts of the com- requested additional funds to complete the pany with highest environmental risks cleanup. The APVRt responded by conduct- to reduce long term environmental ing a financial audit and found that many of liabilities. . . . Although the total value the expenditures would not be considered of environmental liabilities was large, warranted under Hungarian environmental the assessment was used to demon- law. It refused to pay, and the matter is still strate that liabilities are being well under negotiation (Baka 1995; Csanádi and managed, adequate financial provi- Bell 1999; Bluffstone and Panayotou 2000). sions are being made and long term investments will enhance company Price reductions and earmarked escrow value. accounts have been consistently used tools, but the sophistication of Hungary's policies All of the MOL Rt shares were placed increased throughout the 1990s, as Csanádi successfully and quickly, through the New and Bell (1999) note. For example, in 1994 York Stock Exchange (Goldenman 1997). the APVRt hired a full-time environmental expert to coordinate its environmental audit In an analysis of 2,600 privatization records, program. As part of the change in the Csanádi and Bell (1999) found that in only privatization law in 1995, environmental 147 transactions were environmental aspects provisions were more systematically included included. In 79 cases the APVRt took on in the privatization process. An environmen- cleanup commitments, and in 78 cases tal unit was established to oversee environ- purchasers agreed to environmental measures. mental audits and negotiate environmental For the cases in which escrow accounts were issues with investors. established, only 4.1 percent of the earmarked funds were actually drawn on for cleanup, Perhaps the most telling indication of the probably because of bureaucratic roadblocks government's sophistication after 1995 was erected by the APVRt. Csanádi and Bell (1999) the privatization of the oil and gas firm MOL conclude that the use of escrow accounts Rt in 1995­96. At the time of privatization, probably helped the APVRt close deals, but MOL Rt was the largest firm in Hungary, as the status of contaminated sites now in private measured by revenues, and its sale may be hands is unclear. It also seems that ongoing the largest privatization to date (Tillson pollution issues were not addressed at all 1996). The sale of MOL Rt was also the first within the privatization framework. 72 Environment Strategy Papers Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic LITHUANIA vouchers, and 2,726 (mainly small) firms were sold at auction. The remaining firms Lithuania has a diverse industrial sector that were privatized using auctions and tenders includes the fertilizer, chemical, cement, and (see the Website listed in note 26; see also electronics industries. By 1995, 80 percent of box A.2 for a specific privatization case). the enterprises slated for privatization had been transferred, and 50 percent of the The second round took place under a revised economy was in private hands (Lithuanian privatization law passed in 1995 that focused Government Interministerial Working Group on capital privatization but included no 1995; Brada 1996). By 1997, 70 percent of environmental components (Lithuanian gross domestic product (GDP) was produced Government Interministerial Working Group by the private sector, with most ownership 1995). Under the law, a national property transfers having occurred via mass fund was established to prepare enterprises privatization.25 Certain key facilities were and carry out their privatization. A State privatized under the country's capital Privatization Commission oversees privatization program, but virtually no privatization and is responsible to the transactions included hard currency compo- parliament. During the period 1996­99, nents.26 1,119 additional firms were sold, most of them at auction. As part of these deals, Privatization in Lithuania was carried out in investors agreed to preserve 2,300 jobs for two rounds. The first round began in 1991 three to five years. By the end of 1999, 3,100 under the Law on Initial Privatization of State firms were awaiting privatization, but most Property, passed in 1991, and was carried of the shares available were in enterprises in out under the auspices of the Ministry of which the state did not have a controlling Economy.27 By 1995, 5,714 companies had interest (Lithuanian Free Market Institute been sold; 2,928 were privatized using 2000). BOX A.2 Lithuania: Sale of a fertilizer plant AB Achema is the largest producer of nitrogen fertilizer in the Baltic region; in the late 1980s it pro- duced over 2 million tons a year. The collapse of the Soviet market and steep increases in raw material prices at the beginning of the 1990s caused huge difficulties for the company. The firm was privatized in 1994 using a method that was rather different from the usual Lithuanian mass privatization. For the first time, a public tender to acquire a controlling stake was held, with the tender winner selected on the basis of a business plan that included an environmental component. A consortium led by top management won the tender. Transaction documents included no specific environmental provisions except a requirement of compli- ance with local environmental standards, but the company has made significant progress toward reducing its pollution. In 2001 Achema was awarded ISO 14001 certification for environmental manage- ment. Source: Environmental Policy Center (1998); AB Achema Website, . Bluffstone and Panayotou 73 Does the Treatment of Environmental Liability during Privatization Really Matter? Although the government has looked at 1. The direct privatization component is issues of environmental liability, notably initiated at the regional (voivod) level and through an interministerial working group focuses on the privatization of small and convened in 1995, Lithuania has no specific medium-size enterprises. Transfer occurs legislation dealing with environmental through direct sale, contribution of an damage that occurred during the period of enterprise to (often strategic) potential state ownership. Indeed, it is our understand- owners, and the offer of an enterprise to ing that only one capital privatization employees or others for a period of 10 years contract has explicitly included environmen- in lieu of debts. With regard to the last tal requirements: the 1993 purchase of method, when employees are involved, at Klaipedos Tabakas by Philip Morris. As part least 20 percent of the capital must be of the contract, the Lithuanian government acquired from external investors. This agreed to shut down Kovos Tabakas, which component of the program has basically met was the only other cigarette producer in its goals, with about 92 percent of the more Lithuania, but in return Philip Morris was than 2,000 affected enterprises having been required to cofinance site remediation at sold.29 Kovos Tabakas. 2. The second privatization method, indirect With no real cases of environmental stipula- privatization, involves the creation of trea- tions in privatization contracts, liabilities are sury corporations. The mechanisms used settled or will be settled using the civil code, include public stock offerings, tenders, which includes provisions for successor invited negotiations with strategic investors, liability. No legal requirements exist for the free distribution of shares to enterprise gathering or provision of environmental employees, voucher privatization, and mass information during privatization, and gener- privatization through the transfer of shares to ally such information is assembled only national investment funds (NIFs). As of 2002, when a foreign investor requires it. Com- 1,521 firms had been reorganized as treasury pared with, for example, Hungary, environ- corporations. Of these, shares had been mental liability policies in Lithuania are distributed for 975 firms; 521 of these rudimentary. distributions took place through NIFs. POLAND National investment funds are a uniquely Polish institution. In 1994, 15 such funds Privatization in Poland is the responsibility of were established to take control of 512 firms. the Ministry of the Treasury. (The State Employees of firms received 15 percent of all Privatization Agency, which had handled stock at no charge.30 By 1996, the 512 firms privatizations, was abolished in March were under the control of the NIFs. During 2002.) Not counting very small enterprises, the second phase of the program (1996­98), during the period August 1990­January 2002, shares in the NIFs were distributed free to the 5,362 state-owned enterprises were under the public, and in 1997 the shares began to be privatization program.28 The program has traded on the Warsaw Stock Exchange. By three main components. the end of 2001, 73 (16 percent) of the 74 Environment Strategy Papers Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic companies given to the NIFs were either disbursed or simply changed owners infor- liquidated or bankrupt. As of June 2000, the mally. When environmental liabilities were NIFs held only 213 of the original 521 firms. incorporated into the privatization process, generally it was on an ad hoc basis (Bell 3. Two types of liquidation make up the final 1993; Thomas 1994; see also box A.3). privatization method. The first involves the bankruptcy of a firm and the sale of its In 1993 an interministerial unit focusing on assets. The second is essentially a manage- environmental liabilities and privatization ment-employee buyout (Bell 1993). Indeed, was established. This unit, which included as Bell (1993) and Frydman and others (1993) representatives from the Privatization Depart- argue, early liquidations were often just a ment and the Ministry of Environmental covert and politically expedient way to Protection, is charged with commissioning transfer ownership directly and quickly to environmental audits and negotiating envi- firm insiders. ronmental issues with investors case by case. Within this process, environmental issues are By January 2002, a total of 1,752 enterprises addressed via price reductions rather than had been targeted for liquidation, and indemnifications. In general, all liabilities disposal of assets had actually been carried pass to the new owners, and the Polish out in 871 firms. Of the 1,752 firms, 656 government does not take responsibility for bankrupt firms remained in state hands, site remediation or other liabilities leaving 225 nonbankrupt firms to be liqui- (Goldenman 1996, 1997). This approach is dated. In 2002, 45 firms are expected to be generally viewed as being sophisticated but liquidated. time consuming. It is often said that most early privatizations In Poland's mass privatizations, environmen- in Poland were conducted using liquidation tal liabilities have probably received rela- (Frydman and others 1993: 168, 192; tively little attention. Exceptions may be the Goldenman 1997). Official data, however, 521 firms distributed to the NIFs. The NIFs suggest that direct privatization was much more important than liquidation, accounting BOX A.3 for at least two-thirds of all privatizations. Poland: Privatization of a paper producer Liquidations (not including bankruptcies) A typical medium-size (250-employee) paper have consistently made up about 20 percent producer faced pollution-related administrative of all privatizations.31 fines prior to privatization. A thorough environ- mental audit was conducted at the expense of Relatively little is known about the treatment prospective investors. of environmental liabilities for the vast Since the completion of privatization in 1996, majority of privatized firms. It is probably one administrative judgment has been brought fair to say that before 1993, little or no against the firm, but the management does not attention was given to environmental issues. view this charge as very serious. Despite past This is likely to have been particularly true concerns, the firm has not yet introduced an environmental management system at the plant. for liquidated firms whose assets were Bluffstone and Panayotou 75 Does the Treatment of Environmental Liability during Privatization Really Matter? generally took out European Bank for Recon- capital of these enterprises was transferred to struction and Development (EBRD) loans to five private ownership funds, which managed help in restructuring companies, and 5 the assets for the owners of roughly 10 percent of those funds was earmarked for million privatization vouchers that were environmental audits. The NIFs may there- distributed free of charge to Romanian fore have considered potential environmental citizens. The remaining 70 percent interest in liabilities when developing their portfolios. the privatization portfolio was transferred to They also had access to EBRD environmental a state ownership fund, which was respon- training (Goldenman 1997). sible for selling the shares to investors. ROMANIA A revised privatization law passed in 1995 organized the five funds as joint stock Privatization in Romania is a fairly new companies. It also expanded the mass phenomenon, having really begun only in privatization program so that citizens who 1993. Of the 3,257 small and medium-size had not previously used their vouchers to enterprises slated for privatization, by 1999 purchase shares in the private ownership approximately 1,189 had had at least some funds could receive shares in 4,000 state- portion of their capital transferred to the owned enterprises included in the new private sector (CESEP 1998, updated using program. A maximum of 60 percent of the MIGA database, 2000).32 The largest compa- authorized capital of these enterprises could nies in Romania have yet to be privatized, be distributed by this method (CESEP 1998). and there has been some dissatisfaction with Enterprises were sold to foreign or domestic the pace and form of privatization investors through a variety of means, includ- (Mosetescu 2001). ing direct negotiation, public share offering, auction, and direct sale. Since 1998, auc- Compared with other countries in the region, tions have been the most common method Romania's private sector is rather small. As used (MIGA database, 2000). of 1997, only 58 percent of gross domestic product (GDP) was generated by the private In 2000 the state ownership fund was restruc- sector, and slightly over half of employment tured into the Authority for Privatization and was private.33 Management of State Ownership (APAPS). Of note is the use of management-employee The privatization method used in Romania buyouts, which were made possible by a has much in common with the methods used government decree in 1994. It is unknown in other Central and Eastern European how many firms have been privatized in this countries but also has specifically Romanian manner, but the number is believed to be at elements. With the passage of the Law on least 2,000 (CESEP 1998). It is also believed Privatization in 1990, a total of 6,300 that environmental issues were generally enterprises was targeted for privatization. ignored in the mass privatization and capital About half of these enterprises were medium- privatization programs (see the case de- size and large companies employing at least scribed in box A.4). A revised law on capital 500 people. Thirty percent of the authorized privatization enacted in 1997 does include 76 Environment Strategy Papers Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic BOX A.4 Romania: Economic and environmental improvements at a chemical plant with foreign investors In 1998 the Turkish firm Transworld Fertilizers Holding agreed to purchase a 51 percent ownership share in Azomures Chemical Plant for US$45 million. Azomures, located near the city of Targu Mures, is the leading producer of fertilizers and photosensitive chemicals in Romania. Transworld initially deposited just over 70 percent of the total amount agreed but failed to pay the balance and was awarded a less than controlling stake in the company. The Authority for Privatization and Management of State Ownership (APAPS) retained an approximately 20 percent share, and the remainder was sold to other foreign investors on the capital market. In 2001, when Azomures increased its capitalization, Transworld raised its share of the company to just over 50 percent. Immediately after privatization, production was halted because of an unfavorable world fertilizer market. The firm was also saddled with significant debts and needed to rehabilitate its facilities. In addition, it had environmental problems, although there is no indication that environmental aspects were part of the privatization agreement. By 2000, the firm was profitable, with export revenues of US$65 million and output virtually at preprivatization levels. In the first quarter of 2001 Azomures earned US$3.9 million on sales of US$42 million, and sales in that quarter were 84 percent higher than in the same quarter of the previous year. During 1997­2000, the firm significantly reduced its pollution levels. Wastewater volume declined by 50 percent, and nitrogen emissions into water fell by over 75 percent. In 2000 the company invested US$1.9 million in environmental protection. Source: CESEP (1998); National Bank of Romania Monthly Bulletin No. 9, September 2001, available at ; Eteba Romania SA Weekly Bulletin, May 5­11, 2001. available at ; Azomures Annual Report, 2000. provisions related to environmental liabili- These assessments were to be performed by ties, and a government decision of 1999 the firms being privatized. The law envi- spelled out steps to be followed in the capital sioned the possibility that more complex privatization program, including a series of environmental audits might be needed in environmental procedures. The government some cases and required that certified decision explicitly included environmental environmental experts be used for those protection as one of the criteria for evaluat- analyses. Based on the environmental ing investor offers. It also made obligatory assessments, firms are required to create the provision of environmental information compliance plans to manage potential to prospective buyers and contained an environmental liabilities and mitigate risks to explicit requirement to discuss potential human health. The law stipulated that liabilities in firm information packets. specific remediation requirements be in- cluded in the purchase and sale agreements The law also required that environmental with investors. Investors are then responsible assessments be initiated during 1999 for all for the expenditures necessary to implement firms in the capital privatization program. the cleanup. Bluffstone and Panayotou 77 Does the Treatment of Environmental Liability during Privatization Really Matter? SLOVAK REPUBLIC Table A.1 Privatization methods used during the second wave of privatization in the Privatization in the Slovak Republic is under Slovak Republic the auspices of the Ministry for Administra- Number of Assets tion and Privatization of National Property, Privatization method companies (U.S. dollars)a which was founded in 1990, prior to the Direct sales 1,105 3,342,557,743 breakup of Czechoslovakia in January 1993. Public tender 230 380,617,971 The National Property Fund was established the following year to prepare enterprises for Auction 10 477,114 privatization, conduct privatizations, and Note: a. At the December 1997 exchange rate of 35 Slovak crowns to 1 U.S. dollar. transfer ownership from itself to private Source: . The Slovak privatization program had both Recently, privatization has accelerated, with mass and capital privatization components.34 2000 being a particularly important year. The The first wave of privatization, in 1991­93 government earned roughly US$600 million (much of it occurring prior to the dissolution from the sale of a 51 percent interest in of Czechoslovakia), focused primarily on Slovak Telecom to Deutsche Telecom AG. voucher privatization; capital privatization Also of significance in 2000 were the sale of played a relatively minor role. Under the metallurgical part of VSZ to the U.S. steel voucher privatization, each citizen had the giant USX for US$495 million, the purchase right to buy a booklet of 1,000 vouchers that of shares in the Slovak oil refinery Slovnaft could be invested in enterprises selected for by Hungary's MOL Rt for US$270 million, that wave of privatization. Approximately 2.6 and the purchase of a 50 percent interest in million citizens (roughly half the adult the Slovak pulp and paper producer SCP population) participated, and 678 companies Ruzomberok by Austria's Neusiedler AG for were privatized in that way.35 US$80 million. As of May 2002, the National Property Fund had ownership shares of 131 Originally, a second round of mass firms in its portfolio. privatization was to have taken place, but because of the dispersed ownership struc- The National Property Fund operates under tures that were resulting, the focus of the its original enabling legislation (typically second wave, beginning in 1993, was shifted called the Large-Scale Privatization Act) and away from mass privatization and toward its several amendments. The act specifies capital privatization methods. Citizens that all assets and liabilities be transferred to registered for the second wave of mass investors at the time of privatization privatization were compensated with bonds (Tillinghast 1993). Part V of the act states with face values of 10,000 Slovak crowns that fund resources can be used "for defrayal (approximately US$333) that matured at the of expenses accrued by the acquirer due to end of 2000.36 Table A.1 shows the break- settlement of environment-related liabilities down of methods used during the second arising prior to the privatization of the wave of privatization. Direct sales were the enterprise, including those that were un- most important vehicle. known to the acquirer before entering into 78 Environment Strategy Papers Privatization and Environmental Liability Policies in Hungary, Lithuania, Poland, Romania, and the Slovak Republic the purchase agreement concerning the address this potential barrier to foreign direct enterprise or a part thereof." It is unclear to investment. what degree this provision has been used. It also seems that no systematic procedure Presently in Slovakia, most of the "major" exists for evaluating potential environmental foreign investors are dealing with the envi- liabilities and assigning responsibility. ronment issue by essentially ignoring it. Foreign investors recognize that there is a A delegation from the Foreign Investment potential risk for future environmental Advisory Service (FIAS) of the International liability, but they simply calculate the risk as Finance Corporation/World Bank assessed "low." . . . In fact, there were several in- the situation in June 1999 during a visit to stances in which the foreign investor was the Slovak Republic: able to negotiate impunity for past environ- mental liability. One investor created a new There remains great uncertainty as to the limited liability company to protect him from potential liability of investors for "previous" environmental liability and then negotiated a environmental damage. This is particularly contract with the government (Ministry of important when a foreign enterprise invests Environment), according to which the in a Slovak company that has already been company could not be held liable for any privatized or will be privatized. The Slovak preexisting environmental damage.37 government needs to establish policies to Bluffstone and Panayotou 79 Appendix B Industries Included in the Sample I ndustrial sectors are listed by North 221 Utilities (in this case, only the 5-digit American Industrial Classification System level is of interest) (NAICS) code. Asterisks indicate indus- tries and their subsectors that are included in *22111 Electric power generation the sample. Boldface type indicates industrial sectors that are particularly likely to exhibit 31­33 Manufacturing site contamination problems. Sectors are broken down into four- and five-digit codes, *311 Food manufacturing for reference, as appropriate. 3111 Animal food manufacturing 3112 Grain and oilseed milling 11 Agriculture, forestry, fishing, and hunting 3114 Fruit and vegetable preserving and specialty food manufacturing *112 Animal raising 3115 Dairy product manufacturing 1121 Cattle ranching and farming 3116 Meat product manufacturing 1122 Hog and pig farming 3117 Seafood product preparation and 1124 Sheep and goat farming packaging 1125 Animal aquaculture 3118 Bakeries and tortilla manufactur- 1129 Other animal production ing 3119 Other food manufacturing 21 Mining *312 Beverage and tobacco product manufac- *211 Oil and gas extraction turing *212 Mining (except oil and gas) 3121 Beverage manufacturing 2121 Coal mining 3122 Tobacco manufacturing 2122 Metal ore mining 2123 Nonmetallic mineral mining and *313 Textile mills quarrying 3131 Fiber, yarn, and thread mills 3132 Fabric mills 22 Utilities 81 Does the Treatment of Environmental Liability during Privatization Really Matter? 3133 Textile and fabric finishing and 3241 Petroleum and coal products fabric coating mills manufacturing 32411 Petroleum refineries *314 Textile product mills 32412 Asphalt paving, roofing, and 3141 Textile furnishings mills saturated materials manufactur- 3149 Other textile product mills ing 32419 Other petroleum and coal *315 Apparel manufacturing products manufacturing 3151 Apparel knitting mills *325 Chemical manufacturing 3152 Cut and sew apparel manufactur- 3251 Basic chemical manufacturing ing 32511 Petrochemical manufacturing 3159 Apparel accessories and other 32512 Industrial gas manufacturing apparel manufacturing 32513 Dye and pigment manufacturing 32518 Other basic inorganic chemical *316 Leather and allied product manufactur- manufacturing ing 32519 Other basic organic chemical 3161 Leather and hide tanning and manufacturing finishing 325191 Gum and wood chemical manu- 3162 Footwear manufacturing facturing 3169 Other leather and allied product 325192 Cyclic crude and intermediate manufacturing manufacturing 325193 Ethyl alcohol manufacturing *321 Wood product manufacturing 325199 All other basic organic chemical 3211 Sawmills and wood preservation manufacturing 3212 Veneer, plywood, and engi- neered wood product manufac- 3252 Resin, synthetic rubber, and turing artificial and synthetic fibers and 3219 Other wood product manufactur- filaments manufacturing ing 32521 Resin and synthetic rubber manufacturing *322 Paper manufacturing 32522 Artificial and synthetic fibers and 3221 Pulp, paper, and paperboard filaments manufacturing mills 3222 Converted paper product manu- 3253 Pesticide, fertilizer, and other facturing agricultural chemical manufac- turing *323 Printing and related support activities 32531 Fertilizer manufacturing 3231 Printing and related support 32532 Pesticide and other agricultural activities chemical manufacturing 324 Petroleum and coal products 3255 Paint, coating, adhesive, and manufacturing sealant manufacturing 82 Environment Strategy Papers Industries Included in the Sample 32551 Paint and coating manufacturing *327 Nonmetallic mineral product manufac- 32552 Adhesive manufacturing turing 3271 Clay product and refractory 3256 Soap, cleaning compound, and manufacturing toilet preparation manufacturing 32711 Pottery, ceramics, and plumbing 32561 Soap and cleaning compound fixture manufacturing manufacturing 32712 Clay building material and 32562 Toilet preparation manufacturing refractories manufacturing 3259 Other chemical product manu- 3272 Glass and glass product manu- facturing facturing 32591 Printing ink manufacturing 32721 Glass and glass product manu- 32592 Explosives manufacturing facturing 32599 All other chemical product manufacturing 3273 Cement and concrete product manufacturing *326 Plastics and rubber products manufac- 32731 Cement manufacturing turing 32732 Ready-mix concrete manufactur- 3261 Plastics product manufacturing ing 32611 Unsupported plastics film, sheet, 32733 Concrete pipe, brick, and block and bag manufacturing manufacturing 32612 Plastics pipe, pipe fitting, and 32739 Other concrete product manufac- unsupported profile shape turing manufacturing 32613 Laminated plastics plate, sheet, 3274 Lime, gypsum, and gypsum and shape manufacturing product manufacturing 32614 Polystyrene foam product 32741 Lime manufacturing manufacturing 32742 Gypsum and gypsum product 32615 Urethane and other foam product manufacturing (except polystyrene) manufactur- ing 3279 Other nonmetallic mineral 32616 Plastics bottle manufacturing product manufacturing 32619 Other plastics product manufac- 32791 Abrasive product manufacturing turing 32799 All other nonmetallic mineral product manufacturing 3262 Rubber product manufacturing 32621 Tire manufacturing *331 Primary metal manufacturing 32622 Rubber and plastics hoses and 3311 Iron and steel mills and belting manufacturing ferroalloy manufacturing 32629 Other rubber product manufac- 33111 Iron and steel mills and turing ferroalloy manufacturing Bluffstone and Panayotou 83 Does the Treatment of Environmental Liability during Privatization Really Matter? 3312 Steel product manufacturing 3324 Boiler, tank, and shipping from purchased steel container manufacturing 33121 Iron and steel pipes and tubes 33241 Power boiler and heat exchanger manufacturing from purchased manufacturing steel 33242 Metal tank (heavy gauge) manu- 33122 Rolling and drawing of pur- facturing chased steel 33243 Metal can, box, and other metal container (light gauge) manufac- 3313 Alumina and aluminum produc- turing tion and processing 33131 Alumina and aluminum produc- 3325 Hardware manufacturing tion and processing 33251 Hardware manufacturing 3314 Nonferrous metal (except alumi- 3326 Spring and wire product manu- num) production and processing facturing 33141 Nonferrous metal (except alumi- 33261 Spring and wire product manu- num) smelting and refining facturing 33142 Copper rolling, drawing, extrud- ing, and alloying 3327 Machine shops, turned product, 33149 Nonferrous metals (except and screw, nut, and bolt manu- copper and aluminum) rolling, facturing drawing, extruding, and alloying 33271 Machine shops 33272 Turned product and screw, nut, 3315 Foundries and bolt manufacturing 33151 Ferrous metal foundries 3328 Coating, engraving, heat treating, 33152 Nonferrous metal foundries and allied activities 33281 Coating, engraving, heat treating, *332 Fabricated metal product manufactur- and allied activities ing 3321 Forging and stamping 3329 Other fabricated metal product 33211 Forging and stamping manufacturing 33291 Metal valve manufacturing 3322 Cutlery and hand tool manufac- 33299 All other fabricated metal turing product manufacturing 33221 Cutlery and hand tool manufac- turing *333 Machinery manufacturing 3331 Agriculture, construction, and 3323 Architectural and structural mining machinery manufacturing metals manufacturing 33311 Agricultural implement manufac- 33231 Plate work and fabricated turing structural product manufacturing 33312 Construction machinery manu- facturing 84 Environment Strategy Papers Industries Included in the Sample 33313 Mining and oil and gas field 33399 All other general purpose ma- machinery manufacturing chinery manufacturing 3332 Industrial machinery manufactur- *334 Computer and electronic product ing manufacturing 33321 Sawmill and woodworking 3341 Computer and peripheral equip- machinery manufacturing ment manufacturing 33322 Rubber and plastics industry machinery manufacturing 3342 Communications equipment 33329 Other industrial machinery manufacturing manufacturing 3343 Audio and video equipment 3333 Commercial and service industry manufacturing machinery manufacturing 33331 Commercial and service industry 3344 Semiconductor and other elec- machinery manufacturing tronic component manufacturing 3334 Ventilation, heating, air-condi- 3345 Navigational, measuring, medi- tioning, and commercial refrig- cal, and control instruments eration equipment manufacturing manufacturing 33341 Ventilation, heating, air-condi- tioning, and commercial refrig- 3346 Manufacturing and reproducing eration equipment manufacturing magnetic and optical media 3335 Metalworking machinery manu- *335 Electrical equipment, appliance, and facturing component manufacturing 33351 Metalworking machinery manu- 3351 Electric lighting equipment facturing manufacturing 33511 Electric lamp bulb and part 3336 Engine, turbine, and power manufacturing transmission equipment manu- 33512 Lighting fixture manufacturing facturing 33361 Engine, turbine, and power 3352 Household appliance manufac- transmission equipment manu- turing facturing 33521 Small electrical appliance manufacturing 3339 Other general purpose machin- 33522 Major appliance manufacturing ery manufacturing 33391 Pump and compressor manufac- 3353 Electrical equipment manufactur- turing ing 33392 Material handling equipment 33531 Electrical equipment manufactur- manufacturing ing Bluffstone and Panayotou 85 Does the Treatment of Environmental Liability during Privatization Really Matter? 3365 Railroad rolling stock manufac- 3359 Other electrical equipment and turing component manufacturing 33591 Battery manufacturing 3366 Ship and boat building 33592 Communication and energy wire and cable manufacturing 3369 Other transportation equipment 33593 Wiring device manufacturing manufacturing 33599 All other electrical equipment and component manufacturing *337 Furniture and related product manufac- turing *336 Transportation equipment manufactur- 3371 Household and institutional ing furniture and kitchen cabinet 3361 Motor vehicle manufacturing manufacturing 33611 Automobile and light duty motor 33711 Wood kitchen cabinet and vehicle manufacturing counter top manufacturing 33612 Heavy duty truck manufacturing 33712 Household and institutional furniture manufacturing 3362 Motor vehicle body and trailer manufacturing 3372 Office furniture (including fixtures) manufacturing 3363 Motor vehicle parts manufactur- ing 3379 Other furniture-related product 33631 Motor vehicle gasoline engine manufacturing and engine parts manufacturing 33791 Mattress manufacturing 33792 Blind and shade manufacturing 33632 Motor vehicle electrical and electronic equipment manufac- *339 Miscellaneous manufacturing turing 3391 Medical equipment and supplies 33633 Motor vehicle steering and manufacturing suspension components (except spring) manufacturing 3399 Other miscellaneous manufactur- 33634 Motor vehicle brake system ing manufacturing 33991 Jewelry and silverware manufac- 33635 Motor vehicle transmission and turing power train parts manufacturing 33992 Sporting and athletic goods 33636 Motor vehicle fabric accessories manufacturing and seat manufacturing 33993 Doll, toy, and game manufactur- 33637 Motor vehicle metal stamping ing 33639 Other motor vehicle parts 33994 Office supplies (except paper) manufacturing manufacturing 33995 Sign manufacturing 3364 Aerospace product and parts 33999 All other miscellaneous manu- manufacturing facturing 86 Environment Strategy Papers Appendix C Survey Instrument DEFINITION 3.1. Four-digit NAICS classifica- tion Privatization is defined as occurring when 4. Year firm was privatized either: 4.1. In that year did a single group or individual purchase at least A. One group, firm or individual purchases at 30 percent of the company outright least a 30 percent share of the company or did a single group or individual directly from the government, or accumulate at least a 51 percent B. One group, firm or individual accumulates stake in the firm? Outright purchase in the secondary market (e.g., through / 51 percent of shares accumulated purchase of vouchers) at least a 51 percent 5. Annual revenues in $US of the firm share of the company. in the year prior to privatization 5.1. Revenues in 1997 in $US The date of privatization is therefore when 6. Total assets of the firm in $US in the one of these criteria is satisfied. year prior to privatization 6.1. Net assets of the firm in $US When converting local currency values to in 1997 $US, please use the 1997 average exchange 7. Number of employees in year prior rate. to privatization 7.1. Number of employees in 1997 SECTION 1: BASIC FIRM INFORMATION 8. Output of the firm's main product in 1. Firm code in data file 1997 as a percentage of output in the 2. Country in which the firm's head year prior to privatization office is located 9. Percentage of firm ownership held by 3. Three-digit industrial (NAICS) foreigners in 1997 classification of firm 10. Percentage of firm ownership held by the government in 1997 87 Does the Treatment of Environmental Liability during Privatization Really Matter? 10.1. Percentage of votes on the 15. Price of one share purchased board of directors in 1997 from the government in $US 11. Percentage of total ownership of 16. Total number of shares pur- the firm held by the largest chased from the government nongovernmental investor 17. If an auction was used, how individual, firm, or group in many auction rounds were 1997 necessary to sell at least 30 11.1. Percentage of votes on the percent of the firm's shares? board of directors in 1997 17.1. How many bidders participated in the round in SECTION 2: INFORMATION which the firm was privatized? ABOUT THE 18. If a tender was used, how many PRIVATIZATION PROCESS tender rounds were necessary to 12. Prior to privatization, was an sell the greater than 30 percent information packet prepared? Yes of firm shares? / No 18.1. How many tenders were 12.1. If yes, who mainly pre- received in the round in which pared that information? Former the firm was privatized? managers / Privatization agency 19. From the time the intended / Consultants / Other privatization by the government 12.2. If yes, was there an was announced, how many environmental section that dealt months elapsed until the signing with site contamination issues? of the purchase and sale agree- Yes / No ment? 20. According to privatization agency officials, did environ- Questions 13­20 deal only with the mental issues delay the case where privatization involved the privatization of the firm? Yes / outright purchase of at least 30 No percent of the company by one 21. Did more than one member of individual or group. If this was not the the senior management team privatization method, skip to question change within one year after 23. privatization occurred? Yes / No 22. According to firm managers in 13. Was the privatization accom- 1997, was the management team plished using a tender or auction at the time of privatization process? Tender / Auction hostile to privatization? Defi- 14. Percentage of all shares pur- nitely yes / Probably yes / chased from the government at Possibly / Probably not / Defi- the time of privatization nitely not 88 Environment Strategy Papers Survey Instrument SECTION 3: HANDLING OF 28.1. If yes, what was the ENVIRONMENTAL LIABILITIES BEFORE estimated total cost of site PRIVATIZATION remediation in $US? 28.2. If yes, what percentage of 23. Was any written information that estimated cost was consid- about possible site contamina- ered the responsibility of the tion provided to most or all investors? investors? Yes / No 28.3. If yes, as of 1997 what 24. Was a preliminary audit (i.e., percentage of the total site one that utilized a questionnaire cleanup envisioned in the plan to learn the history of the was completed? facility) conducted before 29. With or without a site privatization? Yes / No remediation plan, up to what 24.1. If yes, who financed the amount did the government audit? Investors / Privatization exempt investors from cleanup agency / Both investors and costs (in $US)? privatization agency / Other 29.1. If no monetary amount organization was set, up to what percentage of 25. Was a secondary audit (i.e., one the total cleanup costs did the that included sampling and government exempt investors? analysis of soils and other 30. If privatization occurred through materials) conducted before an outright sale of at least 30 privatization? Yes / No percent to a group, firm, or 25.1. If yes, who financed the individual, did the investor(s) audit? Investors / Privatization pay a reduced price to account agency / Both investors and for the existence of site contami- privatization agency / Other nation? Yes / No organization 30.1. If yes, approximately what 26. Estimate of total expenditures on percentage of the purchase price environmental audits prior to did this price reduction repre- privatization (in $US) sent? 27. Estimate of total expenditures on environmental audits in total as of 1997 (in $US) 27.1. Percentage of this total SECTION 4: HANDLING OF cost that was paid by the govern- ENVIRONMENTAL LIABILITIES AFTER ment PRIVATIZATION 28. Prior to or shortly after 31. Was the firm subject to environ- privatization, was a site mental enforcement actions remediation plan developed that (e.g., notification of violations, included clearly defined cleanup measures Yes / No Bluffstone and Panayotou 89 Does the Treatment of Environmental Liability during Privatization Really Matter? fines, partial or full closure, etc.) / Not very serious / No legal after privatization? Yes / No problems existed 32. Has the firm conducted any site 35. Number of environmentally cleanup? Yes / No related legal actions that were 32.1. If yes, is the site cleanup brought by the government completed? Yes / No against the firm since the date of 32.1.1. If no, does the firm privatization intend to continue site 35.1. According to the current remediation? Yes / No firm management, from the 33. Approximate total amount spent perspective of the firm, how as of 1997 on site cleanup ($US) serious were these legal actions? 33.1. If expenditures were Very serious / Somewhat serious greater than zero, approximate / Not very serious / No legal percentage of the total cost of problems existed site cleanup that was paid by 36. Number of environmentally investors related legal actions that were 34. Number of environmentally brought by the firm against the related legal actions that were government since the date of brought by the public against the privatization firm since the date of 36.1. According to the current privatization firm management, from the 34.1. According to the current perspective of the firm, how firm management, from the serious were these legal actions? perspective of the firm, how Very serious / Somewhat serious serious were these legal actions? / Not very serious / No legal Very serious / Somewhat serious problems existed 90 Environment Strategy Papers Appendix D Regression Results Not Included in Chapter 6 OUTCOME 1 -- PRIVATIZATION PRICES Table D.1 Tobit Regression. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 181.196 0.0001107 182.435 7.35E­5 Whether a general information packet was provided ­0.306635 0.0008383 ­0.325861 0.0002362 to potential investors (dummy) Year of privatization ­0.090842 0.0001139 ­0.0914907 7.55E­5 Chemical industry dummy variable ­0.0402406 0.654117 Primary metal manufacturing industry dummy ­0.366165 0.0224492 ­0.361039 0.0260378 variable Mining and oil and gas extraction dummy variable ­0.211079 0.260318 Petroleum and coal products industry dummy ­0.0081832 0.956804 variable Percentage of total shares held by government in 0.005553 0.0029672 0.0062556 0.0004908 1997 Percentage of total shares held by largest private 0.0042818 0.006377 0.0045678 0.0036157 investor or group Percentage of total shares held by foreign 0.0025443 0.107197 0.00269 0.0774072 investors Revenues (in 1997 U.S. dollars) in year prior to 2.07E­10 0.53208 privatization More than one member of senior management 0.139375 0.0119452 0.153803 0.005038 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.0552287 0.473561 Revenue (in 1997 U.S. dollars) per employee 1.20E­7 0.227208 1.59E­7 0.0118516 Log likelihood ­5.62 ­6.86 Mean of dependent variable 0.128 0.136 Number of observations 91 92 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. 91 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.2 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 172.063 0.00236 177.721 0.0001766 Whether a general information packet was provided to ­0.30483 0.000592 ­0.329968 2.02E­5 potential investors (dummy) Year of privatization ­0.08624 0.002404 ­0.0891212 0.0001809 Chemical industry dummy variable ­0.02883 0.898664 Primary metal manufacturing industry dummy ­0.36399 0.653507 ­0.365316 0.510684 variable Mining and oil and gas extraction dummy variable ­0.22816 0.319563 Petroleum and coal products industry dummy ­0.00632 0.991881 variable Percentage of total shares held by government in 0.004933 0.040978 0.0061179 0.0023656 1997 Percentage of total shares held by largest private 0.003964 0.005024 0.0045077 0.0001977 investor or group Percentage of total shares held by foreign investors 0.002005 0.21283 0.0026659 0.0879753 Revenues (in 1997 U.S. dollars) in year prior to 1.87E­10 0.847622 privatization More than one member of senior management 0.127551 0.121686 0.152759 0.0312354 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.04132 0.587025 Revenue (in 1997 U.S. dollars) per employee 1.24E­7 0.844752 1.59E­7 0.72371 Log likelihood ­6.11 ­6.45 Mean of dependent variable 0.128 0.136 Number of observations 91 92 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.3 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 172.063 0.007196 172.397 0.005995 Whether a general information packet was provided to ­0.30483 0.035961 ­0.32124 0.022645 potential investors (dummy) Year of privatization ­0.08624 0.007307 ­0.08643 0.006087 Chemical industry dummy variable ­0.02883 0.428737 Primary metal manufacturing industry dummy variable ­0.36399 0.003726 ­0.35898 0.00427 Mining and oil and gas extraction dummy variable ­0.22816 0.230866 Petroleum and coal products industry dummy variable ­0.00632 0.885582 Percentage of total shares held by government in 1997 0.004933 0.017526 0.0055 0.007852 Percentage of total shares held by largest private investor 0.003964 0.08851 0.00422 0.068563 or group Percentage of total shares held by foreign investors 0.002005 0.293771 0.002199 0.201724 Revenues (in 1997 U.S. dollars) in year prior to 1.87E­10 0.334948 privatization More than one member of senior management changed 0.127551 0.012742 0.143034 0.005985 within one year of privatization (dummy) (continued) 92 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.3 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization (continued) Full model Final model y p ( y) Coefficient p-value Coefficient p-value Dummy variable for Hungary and Poland ­0.04132 0.668947 Revenue (in 1997 U.S. dollars) per employee 1.24E­7 0.015897 1.60E­7 0.003391 R2 0.46 0.47 F statistic 5.11 9.29 Mean of dependent variable 0.128 0.136 Number of observations 91 92 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.4 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 222.837 3.47E­5 219.238 0.0006057 Whether environmental information is included in 0.246912 0.0056173 0.275427 4.97E­5 information packet (dummy) Year of privatization ­0.111817 3.50E­5 ­0.110088 0.0005995 Chemical industry dummy variable ­0.0359486 0.786596 ­0.117704 0.504207 Primary metal manufacturing industry dummy ­0.36408 0.675545 ­0.456275 0.338257 variable Mining and oil and gas extraction dummy variable 0.0760527 1 Petroleum and coal products industry dummy ­0.141993 0.688231 ­0.202535 0.654558 variable Percentage of total shares held by government in 0.0052661 0.013361 0.0071631 0.0039236 1997 Percentage of total shares held by largest private 0.0022074 0.123901 0.0050194 0.0121063 investor or group Percentage of total shares held by foreign ­0.0014441 0.716783 investors Revenues (in 1997 U.S. dollars) in year prior to ­1.05E­10 0.87612 privatization More than one member of senior management 0.105467 0.234062 0.0604721 0.545252 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.0558328 0.515981 Revenue (in 1997 U.S. dollars) per employee ­1.80E­7 0.694502 Log likelihood ­15.59 ­12.92 Mean of dependent variable 0.054 0.106 Number of observations 77 87 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 93 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.5 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 222.837 0.00483 219.238 0.006234 Whether environmental information is included in 0.246912 0.017364 0.275427 0.010452 information packet (dummy) Year of privatization ­0.11182 0.004825 ­0.11009 0.00622 Chemical industry dummy variable ­0.03595 0.375433 ­0.1177 0.049927 Primary metal manufacturing industry dummy ­0.36408 0.004199 ­0.45628 0.001003 variable Mining and oil and gas extraction dummy variable 0.076053 0.163528 Petroleum and coal products industry dummy ­0.14199 0.05751 ­0.20254 0.038809 variable Percentage of total shares held by government in 0.005266 0.022067 0.007163 0.006112 1997 Percentage of total shares held by largest private 0.002207 0.296153 0.005019 0.031388 investor or group Percentage of total shares held by foreign investors ­0.00144 0.166956 Revenues (in 1997 U.S. dollars) in year prior to ­1.05E­10 0.64571 privatization More than one member of senior management 0.105467 0.016451 0.060472 0.237747 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.05583 0.535118 Revenue (in 1997 U.S. dollars) per employee ­1.80E­7 0.44354 R2 0.38 0.30 F statistic 2.94 4.08 Mean of dependent variable 0.054 0.105 Number of observations 77 87 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.6 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 248.222 0.0006751 185.75 0.0029301 Whether a site remediation plan was in place ­0.0743818 0.825557 ­0.0679833 0.842934 (dummy) Year of privatization ­0.124552 0.0006803 ­0.093216 0.0029437 Chemical industry dummy variable ­0.0226691 0.942187 Primary metal manufacturing industry ­0.54129 0.32007 ­0.454371 0.387391 dummy variable Mining and oil and gas extraction dummy 0.0920916 1 variable Petroleum and coal products industry 0.0213045 0.985117 dummy variable Percentage of total shares held by 0.0057182 0.0340979 0.0045551 0.111195 government in 1997 Percentage of total shares held by largest 0.0045955 0.0013095 0.0048169 0.0005279 private investor or group (continued) 94 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.6 Tobit Regression with Correction for Heteroskedasticity in Year of Privatization, 1997 Revenues, or Both. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization (continued) Full model Final model Coefficient p-value Coefficient p-value Revenues (in 1997 U.S. dollars) in year prior 1.66E­10 0.867853 2.67E­10 0.697477 to privatization More than one member of senior 0.142948 0.0577181 8.47E­2 0.303375 management changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.0319937 0.70294 Revenue (in 1997 U.S. dollars) per employee 2.19E­7 0.583518 1.75E­7 0.675668 Log likelihood ­6.27 ­14.66 Mean of dependent variable 0.13 0.13 Number of observations 83 90 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.7 Tobit Regression. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 250.087 4.80E­7 189.353 9.34E­5 Whether a site remediation plan was in place ­0.0738852 0.445096 ­0.0656806 0.505882 (dummy) Year of privatization ­0.125503 4.86E­7 ­0.0950378 9.38E­5 Chemical industry dummy variable ­0.0178198 0.85419 Primary metal manufacturing industry dummy ­0.536892 0.0008413 ­0.453545 0.0091204 variable Mining and oil and gas extraction dummy 0.0932171 0.738261 variable Petroleum and coal products industry dummy 0.0215444 0.89005 variable Percentage of total shares held by government 0.006042 0.002495 0.0049164 0.0072318 in 1997 Percentage of total shares held by largest 0.0049092 0.0023798 0.0050999 0.0017016 private investor or group Percentage of total shares held by foreign 0.0016289 0.295702 investors Revenues (in 1997 U.S. dollars) in year prior to 1.65E­10 0.627728 2.82E­10 0.414184 privatization More than one member of senior management 0.146268 0.016146 0.0871447 0.167326 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.0447439 0.603571 Revenue (in 1997 U.S. dollars) per employee 2.19E­7 0.0259054 1.72E­7 0.0984479 Log likelihood ­5.76 ­14.5 Mean of dependent variable 0.13 0.13 Number of observations 83 90 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 95 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.8 Ordinary Least Squares. Dependent Variable: Share Price as a Proportion of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 248.222 0.000124 185.75 0.003212 Whether a site remediation plan was in place ­0.07438 0.082345 ­0.06798 0.077687 (dummy) Year of privatization ­0.12455 0.000124 ­0.09322 0.003211 Chemical industry dummy variable ­0.02267 0.667968 Primary metal manufacturing industry dummy ­0.54129 1.41E­5 ­0.45437 0.000213 variable Mining and oil and gas extraction dummy variable 0.092092 0.158072 Petroleum and coal products industry dummy 0.021305 0.497557 variable Percentage of total shares held by government in 0.005718 0.011021 0.004555 0.030002 1997 Percentage of total shares held by largest private 0.004596 0.049191 0.004817 0.044455 investor or group Percentage of total shares held by foreign 0.00136 0.528195 investors Revenues (in 1997 U.S. dollars) in year prior to 1.66E­10 0.402807 2.67E­10 0.227203 privatization More than one member of senior management 0.142948 0.02937 0.084658 0.172033 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.03199 0.797803 Revenue (in 1997 U.S. dollars) per employee 2.19E­7 1.30E­5 1.75E­7 0.002676 R2 0.46 0.34 F statistic 4.55 5.16 Mean of dependent variable 0.132 0.132 Number of observations 83 90 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. OUTCOME 2 -- PRIVATIZATION REVENUES Table D.9 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 3,825.83 0.002102 3,812.2 9.88E­6 Whether an information packet was provided to ­5.95881 0.09665 ­5.93256 0.09677 potential investors (dummy) Year of privatization ­1.91148 0.002229 ­1.90461 1.07E­5 Chemical industry dummy variable 2.33834 0.225244 2.32376 0.218803 Primary metal manufacturing industry dummy ­5.5673 0.341817 ­5.48604 0.319933 variable Mining and oil and gas extraction dummy variable ­1.86641 0.731619 ­1.92628 0.722141 Petroleum and coal products industry dummy ­0.91734 0.771338 ­0.92709 0.768867 variable Percentage of total shares held by government in 0.000559 0.990817 1997 Percentage of total shares held by largest private 0.053563 0.206767 0.053264 0.058276 investor or group (continued) 96 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.9 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization Full model Final model g p Coefficient p-value Coefficient(continued p-value ) Percentage of total shares held by foreign investors ­0.05434 0.221219 ­0.05477 0.200255 Revenues (in 1997 U.S. dollars) in year prior to 4.66E­9 0.516565 4.62E­9 0.519877 privatization More than one member of senior management 0.152312 0.907941 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­10.9403 5.97E­8 ­10.9164 4.90E­9 Revenue (in 1997 U.S. dollars) per employee 4.10E­6 0.60332 4.22E­6 0.587538 Log likelihood ­217.07 ­217.08 Mean of dependent variable 2.85 2.85 Number of observations 83 73 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.10 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 3,867.48 0.039833 4,008.22 0.004301 Whether an information packet was provided to potential ­5.11873 0.320691 ­5.07167 0.322089 investors (dummy) Year of privatization ­1.93231 0.040846 ­2.00315 0.004421 Chemical industry dummy variable 2.5428 0.160988 2.59532 0.155233 Primary metal manufacturing industry dummy variable ­6.14885 0.114089 ­6.43748 0.015431 Mining and oil and gas extraction dummy variable ­1.79824 0.04984 ­1.80431 0.024762 Petroleum and coal products industry dummy variable ­0.87003 0.178666 ­0.87798 0.173136 Percentage of total shares held by government in 1997 ­0.00743 0.894638 Percentage of total shares held by largest private 0.045572 0.426671 0.050257 0.194156 investor or group Percentage of total shares held by foreign investors ­0.06762 0.098934 ­0.06591 0.127725 Revenues (in 1997 U.S. dollars) in year prior to 4.34E­9 0.265922 4.33E­9 0.25939 privatization More than one member of senior management changed ­0.01333 0.991827 within one year of privatization (dummy) Dummy variable for Hungary and Poland ­10.4213 0.001045 ­10.3012 0.000835 Revenue (in 1997 U.S. dollars) per employee 2.98E­6 0.207368 3.09E­6 0.157751 R2 0.46 0.46 F statistic 3.94 4.81 Mean of dependent variable 2.85 2.85 Number of observations 73 73 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.11 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization Full model Final model Coefficient p-value Coefficient p-value Constant 4,056.71 0.001598 4,762.89 2.76E­10 Whether environmental information is included in 1.56663 0.482914 2.58878 0.14784 information packet (dummy) Year of privatization ­2.02972 0.00165 ­2.3833 2.84E­10 (continued) Bluffstone and Panayotou 97 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.11 Tobit Regression. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of Firm Revenues in the Year prior to Privatization (continued) Full model Final model p Coefficient p-value Coefficient p-value Chemical industry dummy variable 1.87737 0.316083 1.69737 0.36676 Primary metal manufacturing industry dummy ­5.32131 0.339228 ­4.93839 0.356282 variable Mining and oil and gas extraction dummy variable ­1.40504 0.789957 ­0.99903 0.845077 Petroleum and coal products industry dummy ­1.71115 0.598188 ­2.39717 0.454178 variable Percentage of total shares held by government in ­0.00305 0.944845 1997 Percentage of total shares held by largest private 0.038776 0.295521 investor or group Percentage of total shares held by foreign investors ­0.06248 0.135265 ­0.06107 0.132484 Revenues (in 1997 U.S. dollars) in year prior to 1.89E­9 0.802431 privatization More than one member of senior management ­0.43245 0.733483 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­10.6456 8.73E­8 ­10.3751 2.43E­8 Revenue (in 1997 U.S. dollars) per employee 1.49E­6 0.85073 Log likelihood ­208.29 ­209.25 Mean of dependent variable 2.36 2.36 Number of observations 72 72 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.12 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of 1990 Firm Revenues Full model Final model Coefficient p-value Coefficient p-value Constant 4,392.21 0.043125 4,573.8 0.006001 Whether environmental information was provided in 1.77932 0.358734 1.8827 0.181078 privatization packets (dummy) Year of privatization ­2.19818 0.04361 ­2.28858 0.006016 Chemical industry dummy variable 2.17797 0.213276 1.93306 0.249264 Primary metal manufacturing industry dummy ­6.34609 0.110198 ­6.1115 0.0221 variable Mining and oil and gas extraction dummy variable ­1.18214 0.318953 ­1.15738 0.059597 Petroleum and coal products industry dummy ­1.74052 0.170218 ­2.0848 0.057761 variable Percentage of total shares held by government in 0.006859 0.911851 1997 Percentage of total shares held by largest private 0.045537 0.470015 investor or group Percentage of total shares held by foreign investors ­0.06853 0.078537 ­0.06973 0.082528 Revenues (in 1997 U.S. dollars) in year prior to 1.44E­9 0.727435 privatization More than one member of senior management ­0.34388 0.797301 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­10.2858 0.001883 ­8.792 0.002299 Revenue (in 1997 U.S. dollars) per employee 6.40E­7 0.820714 R2 0.44 0.40 (continued) 98 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.12 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of 1990 Firm Revenues (continued) Full model Final model Coefficient p-value Coefficient p-value F statistic 3.34 5.19 Mean of dependent variable 2.43 2.36 Number of observations 70 72 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.13 Ordinary Least Squares. Dependent Variable: Privatization Revenues in 1997 U.S. Dollars as a Percentage of 1990 Firm Revenues Full model Final model Coefficient p-value Coefficient p-value Constant 5,717.48 0.032788 4,725.62 0.014919 Whether environmental information, 1.97226 0.315339 0.50915 0.66456 including information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was in ­0.58365 0.523126 ­1.24651 0.107427 place (dummy) Interaction term between the existence 5.52287 0.086424 7.27055 0.004218 of a site remediation plan and the provision of environmental information at the time of privatization Year of privatization ­2.86406 0.033075 ­2.36453 0.014952 Chemical industry dummy variable 2.79181 0.123805 2.37839 0.150301 Primary metal manufacturing industry ­9.65447 0.104533 ­7.42439 0.042028 dummy variable Mining and oil and gas extraction ­0.10506 0.93051 dummy variable Petroleum and coal products industry ­1.54166 0.249288 dummy variable Percentage of total shares held by 0.058542 0.465714 government in 1997 Percentage of total shares held by 0.070245 0.367934 largest private investor or group Percentage of total shares held by ­0.09033 0.022577 ­0.10088 0.0304 foreign investors Revenues (in 1997 U.S. dollars) in ­7.23E­10 0.807576 year prior to privatization More than one member of senior ­0.12806 0.928881 management changed within one year of privatization (dummy) Dummy variable for Hungary and ­8.63816 0.004204 ­8.00002 0.003632 Poland Revenue (in 1997 U.S. dollars) per ­6.01E­7 0.852166 employee R2 0.45 0.41 F statistic 2.78 5.23 Mean of dependent variable 2.33 2.27 Number of observations 66 68 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 99 Does the Treatment of Environmental Liability during Privatization Really Matter? OUTCOME 4 -- EFFICIENCY OF PRIVATIZATION PROCESS Table D.14 Ordinary Least Squares. Dependent Variable: Time Required to Privatize, in Months Full model Final model Coefficient p-value Coefficient p-value Constant 1,602.56 0.012756 1,469.71 0.003303 Environmental information, including site contamination 4.64923 0.014259 4.31308 0.009969 information, provided to investors in privatization packets (dummy) Year of privatization ­0.79926 0.013372 ­0.7321 0.00343 Chemical industry dummy variable 0.022569 0.990142 Primary metal manufacturing industry dummy variable ­3.62828 0.176963 ­3.58616 0.190845 Mining and oil and gas extraction dummy variable 6.65161 0.031858 6.78665 0.030552 Petroleum and coal products industry dummy variable ­1.17734 0.551184 Percentage of total shares held by government in 1997 0.014732 0.698987 Percentage of total shares held by largest private ­0.05772 0.056621 ­0.06601 0.003633 investor or group Percentage of total shares held by foreign investors 0.042155 0.184721 0.040774 0.212096 Revenues (in 1997 U.S. dollars) in year prior to ­1.18E­8 0.030943 ­1.15E­8 0.038107 privatization More than one member of senior management ­1.66077 0.207844 ­1.62832 0.197933 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 4.33992 0.00683 4.13522 0.005381 Revenue (in 1997 U.S. dollars) per employee ­1.03E­5 0.007177 ­1.03E­5 0.003627 R2 0.38 0.37 F statistic 3.24 4.34 Mean of dependent variable 8.38 8.38 Number of observations 85 85 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.15 Tobit Regression with Heteroskedasticity (as a Function of Firm Revenues in 1997) Correction. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm Full model Final model Coefficient p-value Coefficient p-value Constant 558.423 0.756736 0.777099 0.493529 Whether environmental information, 1.50107 0.470721 1.19908 0.141773 including information on site remediation, was provided in privatization packets (dummy) Year of privatization ­0.28048 0.756688 Chemical industry dummy variable 0.877066 0.000326 Primary metal manufacturing ­1.0136 0.932278 industry dummy variable Mining and oil and gas extraction ­0.07031 0.993649 dummy variable Petroleum and coal products ­0.57066 0.771702 industry dummy variable Percentage of total shares held by 0.03072 0.641388 0.003974 0.726202 government in 1997 Percentage of total shares held by 0.016165 0.629484 0.012898 0.377453 largest private investor or group (continued) 100 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.15 Tobit Regression with Heteroskedasticity (as a Function of Firm Revenues in 1997) Correction. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm (continued) Full model Final model g p g p Coefficient p-value Coefficient p-value Percentage of total shares held by ­0.04798 0.027421 ­0.03811 0.194845 foreign investors Revenues (in 1997 U.S. dollars) in 9.13E­10 0.808031 year prior to privatization More than one member of senior 0.406496 1.04E­1 0.151763 0.436869 management changed within one year of privatization (dummy) Dummy variable for Hungary and 1.21E+00 0.518518 Poland Revenue (in 1997 U.S. dollars) per ­3.35E­6 0.792708 ­2.45E­6 0.866411 employee Log likelihood ­45.8 ­44.3 Mean of dependent variable 1.45 1.45 Number of observations 49 49 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.16 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm Full model Final model Coefficient p-value Coefficient p-value Constant 36.241 0.557767 0.22135 0.548159 Environmental information, including site 0.530334 0.090867 0.702345 0.013076 contamination information, provided to investors in privatization packets (dummy) Year of privatization ­0.01823 0.559292 Chemical industry dummy variable 0.418717 0.404176 Primary metal manufacturing industry dummy ­0.01264 0.957947 variable Mining and oil and gas extraction dummy variable 0.232265 0.642939 Petroleum and coal products industry dummy ­0.07341 0.889694 variable Percentage of total shares held by government in 0.014259 0.039428 0.011129 0.003455 1997 Percentage of total shares held by largest private 0.013263 0.011601 0.013611 0.009119 investor or group Percentage of total shares held by foreign investors ­0.00923 0.101604 ­0.00719 0.082537 Revenues (in 1997 U.S. dollars) in year prior to 9.34E­10 0.339829 privatization More than one member of senior management 0.263988 0.200657 0.245536 0.215538 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 0.397957 0.3504 Revenue (in 1997 U.S. dollars) per employee ­1.29E­6 0.009262 ­1.11E­6 0.036959 R2 0.22 0.19 (continued) Bluffstone and Panayotou 101 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.16 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize the Firm (continued) Full model Final model ( ) p p y Coefficient p-value Coefficient p-value R2 0.22 0.19 F statistic 1.16 2.47 Mean of dependent variable 1.47 1.49 Number of observations 68 69 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.17 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Round in Which Firms Were Privatized Full model Final model Coefficient p-value Coefficient p-value Constant ­81.6817 0.511219 ­0.04688 0.952181 Environmental information, including site 0.18763 0.672949 0.110353 0.78438 contamination information, provided to investors in privatization packets (dummy) Year of privatization 0.040733 0.515883 Chemical industry dummy variable ­0.69438 0.177249 ­0.59007 0.218346 Primary metal manufacturing industry dummy ­1.01552 0.032992 ­1.04092 0.00848 variable Mining and oil and gas extraction dummy variable ­1.2009 0.001386 ­1.2572 0.000326 Petroleum and coal products industry dummy ­0.54442 0.421581 variable Percentage of total shares held by government in 0.042993 0.005035 0.040983 4.39E­6 1997 Percentage of total shares held by largest private 0.019983 0.091623 0.020241 0.103479 investor or group Whether firm had any foreign investment (dummy) 1.28348 0.023364 1.13806 0.031322 Revenues (in 1997 U.S. dollars) in year prior to ­3.95E­9 0.001568 ­4.03E­9 0.001504 privatization More than one member of senior management 0.265442 0.478643 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 0.239809 0.783637 Revenue (in 1997 U.S. dollars) per employee ­4.57E­7 0.604904 R2 0.40 0.39 F statistic 2.72 4.63 Mean of dependent variable 2.88 2.88 Number of observations 68 68 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. 102 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.18 Ordinary Least Squares. Dependent Variable: Percentage of All Shares Purchased from the Government at the Time of Privatization Full model Final model Coefficient p-value Coefficient p-value Constant ­3,257.56 0.399607 ­3,509.81 0.30482 Whether environmental information, including ­12.096 0.058475 ­10.1546 0.04917 information on site contamination, was provided in privatization packets to potential investors (dummy) Year of privatization 1.67656 0.387835 1.80299 0.293659 Chemical industry dummy variable ­3.51619 0.535845 Primary metal manufacturing industry dummy ­21.2484 5.72E­5 ­22.4396 2.30E­6 variable Mining and oil and gas extraction dummy variable ­2.30552 0.767237 Petroleum and coal products industry dummy 12.7186 0.426989 variable Percentage of total shares held by government in ­0.43203 0.002165 ­0.43327 0.000663 1997 Percentage of total shares held by largest private ­0.02991 0.76083 investor or group Percentage of total shares held by foreign investors ­0.02991 0.807569 Revenues (in 1997 U.S. dollars) in year prior to ­7.42E­9 0.765305 privatization More than one member of senior management ­0.51371 0.905961 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 8.61411 0.294022 6.51077 0.338035 Revenue (in 1997 U.S. dollars) per employee 1.38E­5 0.19201 ­8.17E­6 0.047498 R2 0.41 0.41 F statistic 3.91 13.23 Mean of dependent variable 63.1 65 Number of observations 90 103 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.19 Tobit Regression. Dependent Variable: Time Required to Privatize, in Months Full model Final model Coefficient p-value Coefficient p-value Constant 1,108.37 0.258223 7.08448 9.99E­15 Whether a site remediation plan was ­2.25708 0.287659 ­2.10492 0.304946 in place prior to or shortly after privatization (dummy) Year of privatization ­0.55278 0.261131 Chemical industry dummy variable 1.65413 0.495184 Primary metal manufacturing ­5.1431 0.188003 industry dummy variable Mining and oil and gas extraction 2.63457 6.91E­1 1.89992 0.76728 dummy variable Petroleum and coal products 1.552 0.676621 1.48444 0.680873 industry dummy variable Percentage of total shares held by 0.036225 0.367322 government in 1997 Bluffstone and Panayotou 103 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.19 Tobit Regression. Dependent Variable: Time Required to Privatize, in Months Full model Final model g Coefficient p-value Coefficient p-value Percentage of total shares held by 0.003526 0.913978 largest private investor or group Percentage of total shares held by ­0.00577 8.56E­1 foreign investors Revenues (in 1997 U.S. dollars) in ­2.81E­9 0.732441 year prior to privatization More than one member of senior ­1.96566 0.16507 ­1.72E+00 0.167727 management changed within one year of privatization (dummy) Dummy variable for Hungary and 7.49476 0.000528 6.15752 4.41E­6 Poland Revenue (in 1997 U.S. dollars) per ­7.05E­6 0.439958 2.35E­6 0.487423 employee Log likelihood ­272.11 ­326.49 Mean of dependent variable 8.23 8.23 Number of observations 85 102 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.20 Ordinary Least Squares. Dependent Variable: Time Required to Privatize, in Months Full model Final model Coefficient p-value Coefficient p-value Constant 1,130.17 0.489844 7.06451 0 Whether a site remediation plan was in place ­2.27742 0.11494 ­2.14276 0.144413 (dummy) Year of privatization ­0.5638 0.492421 Chemical industry dummy variable 1.70236 0.407395 Primary metal manufacturing industry dummy ­4.25618 0.370292 variable Mining and oil and gas extraction dummy variable 2.71418 0.095836 1.9625 0.223721 Petroleum and coal products industry dummy 1.57859 0.073138 1.49915 0.049988 variable Percentage of total shares held by government in 0.038028 0.457781 1997 Percentage of total shares held by largest private 0.004961 0.89703 investor or group Percentage of total shares held by foreign investors ­0.00731 0.84627 Revenues (in 1997 U.S. dollars) in year prior to ­2.94E­9 0.472206 privatization More than one member of senior management ­1.90744 0.214519 ­1.65813 0.198253 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 7.6115 0.003062 6.23619 0.000661 Revenue (in 1997 U.S. dollars) per employee ­7.22E­6 0.099227 2.25E­6 0.321065 R2 0.23 0.22 F statistic 1.63 4.34 Mean of dependent variable 8.23 8.23 Number of observations 85 102 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. 104 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.21 Ordinary Least Squares. Dependent Variable: Time Required to Privatize, in Months Full model Final model Coefficient p-value Coefficient p-value Constant 1172.03 0.528201 7.07846 0 Whether environmental information, including information 3.57151 0.07983 3.19405 0.115823 on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was in place (dummy) ­1.64651 0.194052 ­1.82431 0.070687 Interaction term between the existence of a site ­6.35758 0.083372 ­7.01674 0.010659 remediation plan and the provision of environmental information at the time of privatization Year of privatization ­0.58338 0.53151 Chemical industry dummy variable 0.157903 0.934982 Primary metal manufacturing industry dummy variable ­4.46939 0.217813 Mining and oil and gas extraction dummy variable 3.75178 0.017885 3.81434 0.000528 Petroleum and coal products industry dummy variable ­0.49394 0.79147 Percentage of total shares held by government in 1997 0.005785 0.925548 Percentage of total shares held by largest private ­0.06482 0.114646 ­0.05414 0.003892 investor or group Percentage of total shares held by foreign investors 0.035568 0.278463 0.043581 0.228651 Revenues (in 1997 U.S. dollars) in year prior to ­7.40E­9 0.103305 ­7.20E­9 0.089758 privatization More than one member of senior management changed ­0.59338 0.682068 within one year of privatization (dummy) Dummy variable for Hungary and Poland 6.54526 0.005381 6.73521 0.003699 Revenue (in 1997 U.S. dollars) per employee ­1.31E­5 0.009601 ­1.27E­5 0.000594 R2 0.35 0.32 F statistic 2.15 3.47 Mean of dependent variable 7.8 7.8 Number of observations 75 75 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.22 Tobit Regression with Heteroskedasticity Correction. Dependent Variable: Number of Tender Rounds Required to Privatize Full model Final model Coefficient p-value Coefficient p-value Constant 1,142.3 0.647905 1.13485 1.73E­8 Whether environmental information, 2.00946 0.552243 0.278173 0.395492 including information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was 0.276845 0.134132 0.245251 0.482985 in place prior to or shortly after privatization (dummy) (continued) Bluffstone and Panayotou 105 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.22 Tobit Regression with Heteroskedasticity Correction. Dependent Variable: Number of Tender Rounds Required to Privatize Full model Final model p ( y) Coefficient p-value Coefficient p-value Interaction term between the ­1.16381 0.795891 ­2.87251 0.0674462 existence of a site remediation plan and the provision of environmental information in privatization packets Year of privatization ­0.573196 0.648331 Chemical industry dummy variable 0.969068 0.0060463 0.551117 0.0510751 Primary metal manufacturing ­1.51E+00 1 industry dummy variable Mining and oil and gas extraction ­0.344399 1 ­0.840683 1 dummy variable Petroleum and coal products ­0.653605 0.838603 industry dummy variable Percentage of total shares held by 0.0400492 0.690214 government in 1997 Percentage of total shares held by 0.0183265 0.507662 largest private investor or group Percentage of total shares held by ­ 0.473253 foreign investors 0.0471078 Revenues (in 1997 U.S. dollars) in 9.66E­10 0.887207 2.34E­9 0.660247 year prior to privatization More than one member of senior 0.226754 0.185686 0.326828 0.0905554 management changed within one year of privatization (dummy) Dummy variable for Hungary and 0.836314 0.831857 Poland Revenue (in 1997 U.S. dollars) per ­3.60E­6 0.828609 employee Log 1997 revenues 0.255708 0.0010139 0.132776 0.108218 (heteroskedasticity term) Log likelihood ­37.99 ­48.56 Mean of dependent variable 1.46 1.57 Number of observations 49 68 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.23 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize Full model Final model Coefficient p-value Coefficient p-value Constant 110.219 0.73704 1.14377 8.95E­14 Whether environmental information, 0.634871 0.107572 0.212872 0.371448 including information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was 0.385442 0.416817 0.177226 0.71289 in place (dummy) (continued) 106 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.23 Ordinary Least Squares. Dependent Variable: Number of Tender Rounds Required to Privatize (continued) Full model Final model p ( y) Coefficient p-value Coefficient p-value Interaction term between the ­2.1477 0.010391 ­2.38477 0.001716 existence of a site remediation plan and the provision of environmental information at the time of privatization Year of privatization ­0.05488 0.739311 Chemical industry dummy variable 0.67475 0.180856 0.625214 0.108348 Primary metal manufacturing ­0.04442 0.937203 industry dummy variable Mining and oil and gas extraction ­0.76485 0.079588 ­0.80207 0.073658 dummy variable Petroleum and coal products ­0.23131 0.713807 industry dummy variable Percentage of total shares held by 0.005319 0.688044 government in 1997 Percentage of total shares held by 0.004667 0.435769 largest private investor or group Percentage of total shares held by ­0.0075 0.231168 foreign investors Revenues (in 1997 U.S. dollars) in 1.78E­9 0.007372 2.44E­9 4.51E­5 year prior to privatization More than one member of senior 0.360311 0.060794 0.495293 0.015284 management changed within one year of privatization (dummy) Dummy variable for Hungary and ­0.07225 0.885961 Poland Revenue (in 1997 U.S. dollars) per ­1.39E­6 0.016301 employee R2 0.32 0.26 F statistic 1.35 2.96 Mean of dependent variable 1.46 1.57 Number of observations 59 68 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.24 Tobit Regression. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized Full model Final model Coefficient p-value Coefficient p-value Constant ­505.63 0.253014 ­480.207 0.173344 Whether environmental information, 0.70067 0.273476 0.468171 0.371859 including information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was ­0.432307 0.486248 ­0.980193 0.0749391 in place prior to or shortly after privatization (dummy) (continued) Bluffstone and Panayotou 107 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.24 Tobit Regression. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized (continued) Full model Final model p ( y) Coefficient p-value Coefficient p-value Interaction term between the ­10.2259 0.972587 existence of a site remediation plan and the provision of environmental information in privatization packets Year of privatization 0.254597 0.251097 0.241586 0.171976 Chemical industry dummy variable ­0.794142 0.16854 ­0.943235 0.0911024 Primary metal manufacturing ­7.23E­3 0.994978 industry dummy variable Mining and oil and gas extraction ­0.866379 0.56694 ­0.411218 0.776992 dummy variable Petroleum and coal products ­0.786559 0.383443 ­0.55695 0.52373 industry dummy variable Percentage of total shares held by 0.0109858 0.49444 0.018023 0.122705 government in 1997 Percentage of total shares held by 0.0031519 0.791143 largest private investor or group Percentage of total shares held by 0.0240352 0.0186242 0.024003 0.0108678 foreign investors Revenues (in 1997 U.S. dollars) in ­4.36E­9 0.0473951 ­4.53E­9 0.0178045 year prior to privatization More than one member of senior 0.0199094 0.958006 management changed within one year of privatization (dummy) Dummy variable for Hungary and ­1.59658 0.128952 ­0.818289 0.341109 Poland Revenue (in 1997 U.S. dollars) per 7.66E­8 0.971725 employee Log likelihood ­100.57 ­108.56 Mean of dependent variable 2.91 2.82 Number of observations 61 65 Note: Firms included in this sample are those for which at least 30 percent stakes were sold outright and firms were privatized by tender. Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.25 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized Full model Final model Coefficient p-value Coefficient p-value Constant ­431.77 0.086761 ­421.336 0.040085 Whether a site remediation plan was in place (dummy) ­0.75024 0.179346 ­0.74774 0.156878 Year of privatization 0.216731 0.086164 0.211792 0.039634 Chemical industry dummy variable ­0.63409 0.204206 ­0.53791 0.256365 Primary metal manufacturing industry dummy ­0.30807 0.448846 variable Mining and oil and gas extraction dummy variable ­0.63607 0.302982 ­0.58222 0.271585 Petroleum and coal products industry dummy ­0.25634 0.572154 variable (continued) 108 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.25 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Rounds in Which Firms Were Privatized (continued) Full model Final model Coefficient p-value Coefficient p-value Percentage of total shares held by government in 0.028466 0.00764 0.023052 1.66E­5 1997 Percentage of total shares held by largest private 0.009375 0.32421 investor or group Whether firm had any foreign investment (dummy) 1.46352 0.008006 1.34673 0.003117 Revenues (in 1997 U.S. dollars) in year prior to ­3.67E­9 4.94E­5 ­4.06E­9 0.000336 privatization More than one member of senior management 0.120838 0.735438 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­0.06858 0.919828 Revenue (in 1997 U.S. dollars) per employee 7.22E­8 0.928437 R2 0.48 0.49 F statistic 3.59 8.31 Mean of dependent variable 2.81 2.73 Number of observations 64 68 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.26 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Round in Which Firms Were Privatized Full model Final model Coefficient p-value Coefficient p-value Constant ­551.293 0.066238 ­508.171 0.033815 Whether environmental information, 0.650512 0.032158 0.586609 0.019069 including information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was in ­0.45431 0.467381 place (dummy) Interaction term between the existence ­1.48612 0.114469 ­2.03598 0.00697 of a site remediation plan and the provision of environmental information at the time of privatization Year of privatization 0.277348 0.065391 0.255806 0.033143 Chemical industry dummy variable ­0.78211 0.082709 ­0.79167 0.087815 Primary metal manufacturing industry 0.002841 0.996437 dummy variable Mining and oil and gas extraction ­0.87357 0.164045 ­1.3556 1.12E­6 dummy variable Petroleum and coal products industry ­0.76089 0.083579 ­0.83751 0.091439 dummy variable Percentage of total shares held by 0.012842 0.130658 0.011639 0.045403 government in 1997 Percentage of total shares held by 0.003378 0.627716 largest private investor or group (continued) Bluffstone and Panayotou 109 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.26 Ordinary Least Squares. Dependent Variable: Number of Tenders Received in the Round in Which Firms Were Privatized (continued) Full model Final model g p g p Coefficient p-value Coefficient p-value Percentage of total shares held by 0.024593 0.024486 0.026523 0.020032 foreign investors Revenues (in 1997 U.S. dollars) in ­4.09E­9 6.37E­7 ­3.74E­9 8.25E­6 year prior to privatization More than one member of senior 0.063345 0.860526 management changed within one year of privatization (dummy) Dummy variable for Hungary and ­1.32362 0.000532 ­1.24528 0.000773 Poland Revenue (in 1997 U.S. dollars) per 2.57E­8 0.969047 employee R2 0.51 0.53 F statistic 3.13 5.98 Mean of dependent variable 2.92 2.82 Number of observations 61 65 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.27 Ordinary Least Squares. Dependent Variable: Number of Auction Rounds Required to Sell at Least 30 Percent of the Firm's Shares Full model Final model Coefficient p-value Coefficient p-value Constant ­149.767 0.644351 2.12072 5.71E­9 Whether a site remediation plan was in place ­1.95461 0.015912 ­0.9934 0.000124 (dummy) Year of privatization 0.075934 0.641821 Chemical industry dummy variable 0.796127 0.90288 Primary metal manufacturing industry dummy ­1.08724 0.070123 ­1.02902 0.000137 variable Mining and oil and gas extraction dummy variable 0.001017 0.941814 Petroleum and coal products industry dummy 0.008514 0.356717 variable Percentage of total shares held by government in ­0.61936 0.538481 1997 Percentage of total shares held by largest private ­2.20E­9 0.93837 investor or group Whether firm had any foreign investment (dummy) 0.371446 0.510354 Revenues (in 1997 U.S. dollars) in year prior to 0.079009 0.931002 privatization More than one member of senior management ­8.84E­6 0.211868 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­149.767 0.644351 Revenue (in 1997 U.S. dollars) per employee ­1.95461 0.015912 ­1.18E­5 0.025466 R2 0.22 0.10 F statistic 0.26 1.09 Mean of dependent variable 1.73 1.82 Number of observations 22 33 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. 110 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.28 Ordinary Least Squares. Dependent Variable: Number of Auction Rounds Required to Sell at Least 30 Percent of the Firm's Shares Full model Final model Coefficient p-value Coefficient p-value Constant 420.888 0.21563 468.402 0.18059 Whether environmental information, including 2.87713 0.085855 2.86375 0.068882 information on site remediation, was provided in privatization packets (dummy) Whether a site remediation plan was in place (dummy) Interaction term between the existence of a site ­1.17078 0.124954 ­1.35853 0.020557 remediation plan and the provision of environmental information at the time of privatization Year of privatization ­0.20943 0.218582 ­0.23352 0.182742 Chemical industry dummy variable ­25.2371 0.056163 ­24.0015 0.055626 Primary metal manufacturing industry ­2.6842 0.014999 ­2.56882 0.01035 dummy variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by ­0.02677 0.111802 ­0.02104 0.075122 government in 1997 Percentage of total shares held by largest ­0.0055 0.595933 private investor or group Percentage of total shares held by foreign ­3.73405 0.036022 ­3.57544 0.027757 investors Revenues (in 1997 U.S. dollars) in year prior 1.04E­7 0.056064 9.86E­8 0.055027 to privatization More than one member of senior 0.763175 0.218289 0.785357 0.148246 management changed within one year of privatization (dummy) Dummy variable for Hungary and Poland ­4.19767 0.055514 ­3.9987 0.038898 Revenue (in 1997 U.S. dollars) per employee ­1.81E­6 0.826207 ­3.57544 0.027757 R2 0.59 0.58 F statistic 0.51 1.11 Mean of dependent variable 1.52 1.52 Number of observations 19 19 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 111 Does the Treatment of Environmental Liability during Privatization Really Matter? OUTCOME 5 -- SITE REMEDIATION Table D.29 Binomial Logit. Dependent Variable: Whether Any Site Remediation Has Been Conducted since Privatization (Dummy Variable) Full model Final model Coefficient p-value Coefficient p-value Constant 73.8025 0.831968 ­2.65724 2.79E­8 Whether a secondary audit was conducted (dummy) 1.41124 0.069201 0.972487 0.141185 Year of privatization ­0.03787 0.828294 Chemical industry dummy variable ­0.75866 0.462566 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Petroleum and coal products industry dummy variable Percentage of total shares held by government in ­0.01176 0.416999 1997 Percentage of total shares held by largest private ­0.0175 0.132528 investor or group Percentage of total shares held by foreign investors 0.006412 0.637833 Revenues (in 1997 U.S. dollars) in year prior to 8.32E­9 0.048373 8.02E­9 0.032237 privatization More than one member of senior management 1.00731 0.087305 1.36E+00 0.005375 changed within one year of privatization (dummy) Dummy variable for Hungary and Poland 1.10793 0.140211 0.817577 0.076556 Revenue (in 1997 U.S. dollars) per employee ­4.87E­6 0.27569 ­4.79E­6 0.19699 Log likelihood ­44.9 ­63.5 Restricted log likelihood ­56.4 ­73.5 X2 23.1 19.9 Number of observations 110 143 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. OUTCOME 7 -- POSTPRIVATIZATION ENVIRONMENTAL PERFORMANCE Table D.30 Ordinary Least Squares. Dependent Variable: Change in Particulate Emissions, 1994­97, as a Proportion of Firm Revenues in 1997 Full model Final model Coefficient p-value Coefficient p-value Constant ­0.00285 0.095518 3.27E­6 0.073245 Whether written information on site contamination 1.59E­6 0.134289 ­5.65E­7 0.355816 was provided to most or all investors (dummy) Year of privatization 1.43E­6 0.095317 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy ­2.61E­6 0.221708 variable Percentage of total shares held by government 1.13E­7 0.23048 in 1997 (continued) 112 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.30 Ordinary Least Squares. Dependent Variable: Change in Particulate Emissions, 1994­97, as a Proportion of Firm Revenues in 1997 (continued) Full model Final model Coefficient p-value Coefficient p-value Percentage of total shares held by foreign 2.88E­8 0.147074 investors Revenues (in 1997 U.S. dollars) in year prior to ­1.32E­13 0.139642 privatization Dummy variable for Hungary and Poland ­3.00E­6 0.092282 ­1.71E­6 0.07432 Revenue (in 1997 U.S. dollars) per employee ­8.53E­11 0.103251 ­5.92E­11 0.09447 Frequency of internal emissions monitoring (6 = 3.43E­7 0.115496 3.04E­7 0.199478 > once a day; 1 = once a year; 0 = never) Frequency of Monitoring by Regulators (once in ­3.89E­7 0.070228 ­2.82E­7 0.126172 __ months) Whether inspections are generally announced 3.80E­6 0.065388 1.78E­6 0.188372 (dummy) R2 0.60 0.37 F statistic 0.95 1.89 Mean of dependent variable 0.88E­6 0.85E­6 Number of observations 19 26 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.31 Ordinary Least Squares. Dependent Variable: Change in Particulate Emissions, 1994­97, as a Proportion of Firm Revenues in 1997 Full model Final model Coefficient p-value Coefficient p-value Constant ­7.55E­4 0.074853 ­0.0005 0.116379 Whether a preliminary audit was conducted (dummy) ­2.02E­6 0.242331 ­8.41E­7 0.275074 Year of privatization 3.82E­7 0.073639 2.55E­7 0.113606 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable ­1.31E­6 0.16895 ­8.09E­7 0.062111 Percentage of total shares held by government in 1997 ­5.46E­8 0.314149 ­5.26E­8 0.017967 Percentage of total shares held by foreign investors 1.98E­8 0.210946 Revenues (in 1997 U.S. dollars) in year prior to privatization 5.82E­15 0.754952 Dummy variable for Hungary and Poland ­4.83E­6 0.073628 ­3.46E­6 0.048872 Revenue (in 1997 U.S. dollars) per employee ­1.73E­11 0.637966 Frequency of internal emissions monitoring (6 = > once a day; 1 = once a year; 0 = never) 3.51E­7 0.124523 3.75E­7 0.110626 Frequency of monitoring by regulators (once in __ months) ­5.07E­7 0.090547 ­3.73E­7 0.10953 Whether inspections are generally announced (dummy) ­1.42E­6 0.39389 R2 0.50 0.40 F statistic 0.81 1.82 Mean of dependent variable 0.11E­5 0.95E­6 Number of observations 21 27 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 113 Does the Treatment of Environmental Liability during Privatization Really Matter? Table D.32 Ordinary Least Squares. Dependent Variable: Change in Tons of Particulate Emissions, 1994­97, per Dollar of Revenue Generated by Firms in 1997 Full model Final model Coefficient p-value Coefficient p-value Constant ­0.0043 0.004845 ­0.00395 0.000928 Whether a site contamination plan was in place 6.16E­7 0.351393 2.07E­7 0.480943 (dummy) Year of privatization 2.16E­6 0.004832 1.98E­6 0.000926 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Percentage of total shares held by government 3.53E­8 0.382955 in 1997 Percentage of total shares held by foreign 3.18E­8 0.019367 2.63E­8 0.001084 investors Revenues (in 1997 U.S. dollars) in year prior to ­4.87E­14 0.030294 ­3.35E­14 0.031177 privatization Dummy variable for Hungary and Poland 1.60E­6 0.123162 1.40E­6 0.057198 Revenue (in 1997 U.S. dollars) per employee ­2.16E­10 0.004121 ­2.07E­10 0.000755 Frequency of internal emissions monitoring (6 1.61E­7 0.435979 = > once a day; 1 = once a year; 0 = never) Frequency of monitoring by regulators (once 6.03E­8 0.377248 8.14E­8 0.15419 in __ months) Whether inspections are generally announced 8.48E­6 0.002864 8.18E­6 0.001613 (dummy) R2 0.88 0.87 F statistic 3.52 5.61 Mean of dependent variable 0.12E­5 0.12E­5 Number of observations 16 16 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.33 Binomial Logit. Dependent Variable: Whether Firm Had an Environmental Plan in Place (Dummy) Full model Final model Coefficient p-value Coefficient p-value Constant 595.788 0.200364 896.139 0.013043 Whether a site remediation plan was in place 0.42165 0.749914 0.617616 0.597567 (dummy) Year of privatization ­0.29894 0.200272 ­0.44944 0.013069 Primary metal manufacturing industry dummy ­3.10006 0.084764 ­3.33502 0.032362 variable Mining and oil and gas extraction dummy variable Percentage of total shares held by government ­0.02361 0.159276 ­0.03068 0.015314 in 1997 Percentage of total shares held by foreign ­0.0174 0.169714 ­0.01386 0.210009 investors Revenues (in 1997 U.S. dollars) in year prior to 1.11E­8 0.422167 6.00E­9 0.037575 privatization (continued) 114 Environment Strategy Papers Regression Results Not Included in Chapter 6 Table D.33 Binomial Logit. Dependent Variable: Whether Firm Had an Environmental Plan in Place (Dummy) (continued) Full model Final model Coefficient p-value Coefficient p-value Dummy variable for Hungary and Poland 1.92404 0.039617 2.72226 3.05E­5 Revenue (in 1997 U.S. dollars) per employee 9.63E­6 0.615382 Frequency of internal emissions monitoring (6 0.056251 0.780569 = > once a day; 1 = once a year; 0 = never) Frequency of monitoring by regulators (once in 0.020991 0.773426 __ months) Whether inspections are generally announced 0.564505 0.523861 (dummy) Log likelihood ­31.8 ­42.4 Restricted log likelihood ­42.9 ­98.1 X2 22.1 111 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Table D.34 Binomial Logit. Dependent Variable: Whether Firms Were Seeking ISO 14000 Series Certification (Dummy) Full model Final model Coefficient p-value Coefficient p-value Constant ­826.005 0.124689 ­506.623 0.233475 Whether a site contamination plan was in place ­2.44561 0.672502 ­0.21272 0.85608 (dummy) Year of privatization 0.412697 0.125969 0.252385 0.236524 Primary metal manufacturing industry dummy variable Mining and oil and gas extraction dummy variable Percentage of total shares held by government in ­0.02918 0.256294 1997 Percentage of total shares held by foreign 0.017569 0.23038 0.027478 0.009598 investors Revenues (in 1997 U.S. dollars) in year prior to 3.36E­8 0.038892 privatization Dummy variable for Hungary and Poland ­2.42187 0.047069 ­1.16337 0.124881 Revenue (in 1997 U.S. dollars) per employee ­1.33E­5 0.515184 Frequency of internal emissions monitoring (6 = 0.587633 0.03468 0.579756 0.002394 once a day; 1 = once a year; 0 = never) Frequency of monitoring by regulators (once in __ 0.049909 0.446678 months) Whether inspections are generally announced 2.68338 0.007015 2.20155 0.003901 (dummy) Log likelihood ­22.5 ­32.6 Restricted log likelihood ­39.1 ­46.6 X2 33.1 28.0 Number of observations 71 86 Note: Boldface type indicates that the variable is significant at least at the 10 percent level. Bluffstone and Panayotou 115 Notes 1. The Treuhandanstalt has been criticized, simply the estimated coefficients, as however, for focusing too much on costs would be the case for OLS. The so-called and for too often choosing containment marginal effects depend on the values of rather than site remediation measures all independent variables at which effects (Goldenman 1995). are estimated, and as is standard, mar- 2. Boyd (1996) argues, however, that ginal effects are computed at the means exclusive reliance on price discounts is of all independent variables (Kennedy undesirable because such methods do 2001: 238). These are quoted in the text, not necessarily specify the level of and all marginal effects are included in purchaser liability and so invite adverse appendix D. All OLS regressions were selection. corrected for heteroskedasticity using the 3. On the financing side, environmental method of White (1980). Tobit regres- escrow accounts have been extensively sions were also corrected for used in the region. National liability heteroskedasticity, with pools are more efficient mechanisms heteroskedasticity modeled as a function because funds are less narrowly ear- of log firm revenues in 1997 (see Greene marked (Boyd 1996). 1998: 674). We experienced model 4. The data were often difficult for in- convergence problems in several cases country researchers to obtain. This was and therefore were often not able to not unexpected. Some survey report results. The heteroskedasticity- implementers were more successful than corrected Tobit models also often re- others in collecting large, robust samples. sulted in estimates of coefficients for 5. This stratification follows the various industry dummies that were totally "screening" methods proposed in the insignificant even when in uncorrected literature. See, for example, Rötschke models coefficients were highly signifi- (1996); Chvojka and Tylova (1996). cant. For this reason, even in cases where 6. Because Tobit regressions are nonlinear convergence was not a problem, estimation techniques, the marginal heteroskedasticity-corrected results are effects of independent variables are not presented only in appendix D. Results for 117 Does the Treatment of Environmental Liability during Privatization Really Matter? the Tobit models can be compared with that preparation of plans was probably the OLS estimates, which were cor- often initiated by environment ministries rected for heteroskedasticity. It was also or investors. considered possible that three vari- 12. Logit and probit models were also run ables--foreign ownership, government, that used all conditioning variables to try and largest investor ownership shares-- to predict the probability that a firm were determined simultaneously with conducted either primary or secondary outcomes 1­4. We therefore tested for audits. Foreign ownership was not this possibility in Tobit models using the associated with the probability that an method of Greene (1998: 685), but in no audit was conducted. The only variable case was simultaneity even remotely that was significant in these models was suggested. the year of privatization, with the 7. OLS regressions were run first, followed regressions suggesting that over time the by the Tobit, Tobit corrected for probability that a firm would conduct heteroskedasticity, and Tobit with tests either type of audit increased. for endogeneity. To allow full compara- 13. The finding on management change bility of results, full and final Tobit and suggests that this variable is a partial OLS models include exactly the same proxy for firm quality, with higher- variables. In some cases this implies that quality firms more likely to experience insignificant variables were retained in management change. Tobit regressions. 14. In the OLS model the metal and mining 8. The analysis was also done with share industry dummies were negatively and prices deflated by 1997 annual rev- significantly correlated with number of enues, but the results were very similar. tenders received, suggesting that these 9. It was not possible to include general industries may be less attractive. and environmental information variables 15. In logit and probit models, no relation- simultaneously because environmental ship between our information variables information is a subset of general and whether privatization officials information. reported that environmental issues 10. Similar results are presented in Tobit and created delays was observed. OLS models corrected for 16. No relationship between the existence heteroskedasticity, although the esti- of a site remediation plan alone and the mated coefficient magnitudes are much number of tender rounds required to larger. privatize firms was found. 11. We emphasize that we know which 17. As was noted earlier, in logit and Tobit firms in the data set developed models coefficients do not show the remediation plans before or shortly after magnitude of effects. In Tobit models privatization, but we do not know marginal effects are typically similar to whether governments or investors estimated coefficients, but in logit prepared the plans. We would expect models the marginal effects often differ that investors and privatization agencies dramatically from the coefficient esti- were, in general, jointly responsible and mates. 118 Environment Strategy Papers Notes 18. No relationship was detected between . information about site contamination to 26. See the country data on the ECSPF most or all investors or whether prelimi- Website, at . conducted. No relationship was detected 27. The Ministry of Economy was abolished between any of the environmental in 1998. information variables and legal actions 28. Firms were brought into the program brought by the public against firms. over time. For example, by the end of 19. We also included environmental infor- 1995, only 3,515 firms had been mation and the existence of a site included in the program. remediation plan, along with an interac- 29. Unless otherwise noted, the source for tion term, in a logit model explaining information in this section is the the existence of administrative actions. Privatization Website of the Ministry of Neither the site remediation plan the Treasury, . significant. We then dropped the site 30. As of March 2002, excluding the sugar contamination plan variable, but the industry, almost 443 million shares in results did not change. The same was 1,035 firms had been or were slated to true when we included only environ- be distributed. mental information and the interaction 31. It may be true that liquidations were term in the model. more important by value, since direct 20. No relationship was found with fre- privatization was generally used for quency of air pollution monitoring. small and medium-size enterprises. 21. The mining sector dummy variable is 32. MIGA, Multilateral Investment Guaran- omitted from the full model because tee Agency. The database is available at when included, it was perfectly corre- . lated with the dependent variable. 33. See the country information on the 22. It could, of course, be the case that ECSPF Website, at preprivatization and postprivatization . viduals. 34. There also existed a small-scale program 23. Approximately 21 percent of firms in the to privatize retail stores, service enter- sample had completed or were engaged prises, and the like. in site remediation. 35. See . site remediation plans, we know that 36. See the Website of the National Property firms were responsible for all cleanup. Fund of the Slovak Republic, . Europe and Central Asia Region Private 37. Source: . Bluffstone and Panayotou 119 Bibliography The word processed describes informally Privatization in Central and Eastern Europe: reproduced works that may not be com- Toward an Optimal Policy." Environmental monly available through libraries. and Resource Economics 17: 335­52. Auer, Matthew, Rafael Reuveny, and Lisa Boyd, James. 1996. "Environmental Liability Adler. 2001. "Environmental Liability and Reform and Privatization in Central and Foreign Direct Investment in Central and Eastern Europe." European Journal of Law Eastern Europe." Journal of Environment and Economics. 3 (1): 39­60. and Development 10 (1): 5­34. Brada, Joseph. 1996. "Privatization Is Transi- Baka, Eva. 1995. "Implementation of Environ- tion--or Is It?" Journal of Economic Per- mental Assessment and Performance Audits spectives 10 (2): 67­86. at the Hungarian State Property Agency: CESEP (Center for Environmentally Sustainable Final Report to the Harvard Institute for Economics Policy). 1998. "Final Report on International Development." Processed. Privatization Survey." Bucharest. Pro- Balaban, Vladislav, Randall A. Bluffstone, and Theodore Panayotou. 1994. "Lemons and cessed. Liabilities: Privatization, Foreign Invest- Chvojka, Richard, and Eva Tylova. 1996. "The ment and Environmental Liability in Activities of the Ministry of Environment of Central and Eastern Europe." Environmental the Czech Republic in Addressing Past Impact Assessment Review 14 (May): 157­ Environmental Damage." Prepared for the 68. Organisation for Economic Co-operation Bell, Ruth. 1993. "Capital Privatization and and Development (OECD)/PHARE Work- the Management of Environmental Liability shop on Policies to Address Environmental Issues in Poland." In Gretta Goldenman, Liability in Privatization, May 9­10, ed., Environmental Liability and Prague. Privatization in Central and Eastern Europe. Csanádi, Maria, and Ruth Greenspan Bell. International Environmental Law and Policy 1999. "Environmental Liability in Transi- Series. London: Graham and Trotman. tion: A Look at the Record in Hungary." Bluffstone, Randall, and Theodore Panayotou. Resources 134 (winter): 10­13. Washing- 2000. "Environmental Liability and ton, D.C.: Resources for the Future. 121 Does the Treatment of Environmental Liability during Privatization Really Matter? Dodds, Paul, and Gerd Wächter. 1993. Conference on Environmental Aspects of "Privatization Contracts with the German Privatization in the Period of Economic Treuhandanstalt: An Insiders' Guide." Transition, March 13­15, Stare Jablonki, International Lawyer 27 (1, spring): 65­90. Poland. EAP Task Force. 1998. "Environment in the ------. 1997. "Privatization and Environmen- Transition to a Market Economy: Address- tal Liability in CEECs and the NIS: Review- ing Environmental Challenges in Central ing Progress since the 1992 Warsaw and Eastern Europe and the NIS." Discus- Conference." Prepared for Organisation for sion document for Organisation for Economic Co-operation and Development Economic Co-operation and Development (OECD) Expert Workshop on Privatisation (OECD) Expert Workshop, December. and Environment in Central and Eastern Environmental Policy Center. 1998. "Final Europe, Paris, September 29­30. Report on Privatization Survey." Vilnius. Gray, Cheryl. 1996. "In Search of Owners: Processed. Privatization and Corporate Governance in Frydman, Roman, and Andrzej Rapaczynski. Transition Economies." World Bank 1994. Privatization in Eastern Europe: Is Research Observer 11 (2): 179­97. the State Withering Away? London: Greene, William H. 1998. LIMDEP Version Central European University Press. 7.0 User's Manual. Plainview, N.Y.: Frydman, Roman, Andrzej Rapaczynski, John Econometric Software, Inc. S. Earle, and others. 1993. The Kennedy, Peter. 2001. A Guide to Economet- Privatization Process in Russia, Ukraine, rics. 4th ed. Cambridge, Mass.: MIT Press. and the Baltic States: Economic Environ- Klavens, Jonathan, and Anthony Zamparutti. ment, Legal and Ownership Structure, 1995. Foreign Direct Investment and Institutions for State Regulation, Overview Environment in Central and Eastern of Privatization Programs, Initial Transfor- Europe: A Survey. A report for the Environ- mation of Enterprises. Budapest: Central mental Action Programme for Central and European University Press. Eastern Europe. Washington, D.C.: World Goldenman, Gretta. 1993. "Environmental Bank. Liability in Central and Eastern Europe: Lithuanian Free Market Institute. 2000. The Legal Issues." In Gretta Goldenman, "Privatization in Lithuania: Analysis March ed., Environmental Liability and 2000." Vilnius. Privatization in Central and Eastern Lithuanian Government Interministerial Europe. International Environmental Law Working Group. 1995. "Recommendations and Policy Series. London: Graham and to the Ministry of Environmental Protec- Trotman. tion." Vilnius. ------. 1995. "Lithuania's Current Environ- Lovei, Magda, and Bradford Gentry. 2002. mental Liability Regime." Vilnius. The Environmental Implications of ------. 1996. "Legal, Economic and Environ- Privatization: Lessons for Developing mental Dimensions of Privatization of Countries. World Bank Discussion Paper Industrial Enterprises in Central and 426. Washington, D.C. Eastern Europe." Prepared for the Harvard Mosetescu, Ovidiu. 2001. "Large Businesses Institute for International Development Begin with Small Things: Introductory 122 Environment Strategy Papers Bibliography Note." APAPS Newsletter (January). Republic." In Gretta Goldenman, ed., Bucharest: Authority for Privatization and Environmental Liability and Privatization Management of State Ownership. in Central and Eastern Europe. Interna- Rötschke, Harald. 1996. "Treuhandanstalt's tional Environmental Law and Policy Quick Assessment Method for Estimating Series. London: Graham and Trotman. Cleanup Costs for Past Environmental Tillson, Tim. 1996. "Environmental Issues in Damage in the New Lander." Prepared for the Privatization of MOL Rt." Presentation the Third Organisation for Economic Co- to the Third OECD/PHARE Workshop to operation and Development (OECD)/ Address Environmental Issues in PHARE Workshop to Address Environmen- Privatization and Economic Reconstruc- tal Issues in Privatization and Economic tion, November 7­8, Budapest. Reconstruction, November 7­8, Budapest. White, H. 1980. "A Heteroskedasticity- Thomas, Randall. 1994. "The Impact of Consistent Covariance Matrix Estimator Environmental Liabilities on Privatization and a Direct Test for Heteroskedasticity." in Central and Eastern Europe: A Case Econometrica 48: 817­38. Study of Poland." University of California World Bank. 2002. Transition--The First Ten Davis Law Review 28 (1): 165­217. Years: Analysis and Lessons for Eastern Tillinghast, Pamela. 1993. "Privatization and Europe and the Former Soviet Union. Environmental Liability: The Laws and Washington, D.C. Policies of the Czech and Slovak Federal Bluffstone and Panayotou 123