Pollcy, Planning, and Research WORKING PAPERS' Public Economics Country Economics Department The World Bank April 1989 WPS 190 Macro Performance Under Adjustment Lending Riccardo Faini, Jaime de Melo, Abdel Senhadji-Semiali, and Julie Stanton Bank-Fund adjustment lending apparently improves economic performance mostly by relieving the foreign exchange con- straint and allowing for the purchase of crucial foreign interme- diate goods. The Policy. Plnning, and Research Complex distributes PPR Working Papers to disseminate the ftdi igs of work in progrcss and to encourage the exchange of ideas among Bank staff and all others interested in development issues. These papers carry the names of theauthors. reflect only their views, and should be used and cited acconTengly. lTe findings. interpretations, and conclusions are the authors' own. They should not beaauributed totheWorld Bank, its Board of Directors, its management, orany of its member countries. Polac,Pnning, and Research Public Economics The authors of this paper used simple statistical group. The control group is small, howcvcr, So methods to measure the effect of adjustment the results are not statistically robust. lending (AL) on economic performance. Using eight economic :ndicators, they relied on tradi- The ntensity of Bank-Fund involvement tional "before-after" comparisons of AL recipi- contributes significantly to better performnance ents and a control group of 63 countries. for most indicators - except that intensity of Bank-Fund lending correlates negatively (and How have countries under adjustment significantly) with the share of investment in lending performed? AL countries improved GDP. Gven the pr sitive corrclation of lending their external position, generating enough of a intensity with GDP growth and import growth, it trade balance surplus to service their extemal would appear that Bank-Fund lending improves debt. Fiscal (and inflation) indicators deterio- economic pe.formance mostly by relieving the rated, however, a sign that macroeconomic foreign exchange constraint and allowing for the imbalances remained. Finally, growth rates fell, purchase of crucial foreign intermediate goods. reflecting deteriorating terms of trade and the difficulties of reducing absorption to the re- Adjustment lending was intended to elicit a quired degree. supply response: for a given expenditure- switching policy, the trade balance was expected On nine economic indicators, AL recipients to improve more. Alternatively, AL recipicnts fared better overall than the non-recipients - were expected to achieve a given improvement although improvement varied between 53 and 63 in trade balance (controlled for changes in terms percent, depending on the classification. Some of trade) at less cost than non-recipients in terms improvements were mild, some statistically of forgone growth. Correlations for a group of insignificant. Improvements are stronger for a 30 countries that received their fist Bank adjust- group of 12 AL recipients that received 3 or ment loan by 1984 show that AL recipients more adjustment loans. experienced higher average growth and im- proved their trade balance more than non- Results appear stronger when the analysts recipients. The tradeoff between growth aid control for the potentially negative effects of the improvement in external balance, however, was external environment on performance: AL the same for both groups. recipients appear to do better than the control This paper is a product of the Public Economoics Divsion, Country Economics Department, Copies arc available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Maria Ameal, room N8-073, extension 61466. (Thirty-five pages with tables and charts.) The PPR Working Paper Series disseminates the findings of work under way in the Bank's Policy, Planning, and Research Complex. An o'ojective of the series is to get these findings out quickly, even if presentations are less than fully polished. T'he findings, interpretations, and conclusions in these papers do not necessarily represent official policy of the Bank. Produced at the PPR Dissemination Center Table of Contents 1. Introduction 2. Alternatives to Evaluate Adjustment Lending Effectiveness 2.1 Alternative Approaches 2.2 Implementation 3. The External Environment Under Adjustment Lending 4. Performance Indicators for Bank-Fund Supported Adjustment Programc 5. Performance Indicators for Bank Adjeutment Lending Recipients 6. Growth-External Adjustment Tradeoffs for Recipients of Bank Adjustment Programs 7. Conclusions Footnotes References Appendix 1. Data Sources 2. Country Classifications Paper prepared for the World Bank Symposium, 'Adjustment Lending and Beyond,' April 6-7, 1989. We thank Bela Balassa, Ajay Chibber, Patrick Conway, John Holsen, Dani Rodrik and Vinod Thomas for helpful comments. MACRO PERFORMANCE UNDER ADJUSTMENT LENDING 1. Introduction The objective of adjustment lending (AL) programs supported by international agencies is to promote stabilization with growth. In the recent adjustment programs initiated with the Bank and the IMF, the instruments for achieving internal balance and stabilization included reductions in fiscal deficits through better budgetary management, tighter monetary control, and lower government expenditures. External balance was to be achieved through sustained real exchange rate depreciation which was to help achieve the positive resource balance required to service the foreign debt while positive (but not too high) real interest rates were to mobilize private savings. The microeconomic instruments for achieving and sustaining higher growth included: (i) reforming the system of production incentives; (ii) providing incentives to save and invest; (iii) choosing the appropriate investments, namely net foreign exchange earners (private sector), and appropriate social and physical infrastructure (public sector). This paper assesses macroeconomic performance under Bank-Fund supported adjustment programs. The paper relies on a statistical analysis of performance indicators. Statistical evaluation of adjustment programs is fraught with difficulties. To begin with, an assessment of performance under Bank supported adjustment programs must recognize that most programs were undertaken jointly with the IMF. Any assessment must also recognize that performance will be influenced by the external environment. Countries under adjustment programs which faced a more unfavourable environment would be e7pected to show less improvement in performance. Furthermore, it would - 2 - be desirable to control for what adjustment policies might have been in the absence of Bank-Fund lending. Finally, one should recognize that participation in Bank-Fund supported programs is not independent of a country's own policies. To take into account all these difficulties surrounding any quantitative assessment of adjustment lending is well beyond the scope of the preliminary analysis of adjtstment lending undertaken here. In a companion paper, we attempt to tackle some of these difficulties by postulating a simple, but specific, model of how Bank-Fund lending and the external environment interact (Faini et al, 1989). Here, we rely on a simpler control group analysis, which is less restrictive in its assumptions, and therefore, less satisfactory in terms of controlling for the methodological problems mentioned above. However, we use several approaches to check for the robustness of our results. In Section 2, we discuss briefly the pitfalls of alternative statistical approaches used in assessing the effectiveness of adjustment programs. In Section 3, we provide summary statistical evidence on the environment under which adjustment lending took place, emphasizing the relative role of commercial and multilateral funds, and the size of external shocks. In Section 4, a first cut statistical evaluation is done by comparing the relative performance of countrits which received a high amount of Bank-Fund lending, with countries which received a low amount of Bank-Fund lendin3. In Section 5, we rely on control group comparisons for 30 countries which received their first World Bank-supported adjustment loan by 1984. Finally, in Section 6, we look at growth/resource balance improvement tradeoffs for the same group of 30 countries. We then compare these tradeoffs with the same tradeoffs for non-adjustment recipient countries. Conclusions follow in Section 7. 2. Alternatives to Evaluate Adjustment Lending Effectiveness This section discusses two issues raised in any assessment of adjustment lendings choice of evaluation method and choice of indicators. A discussion of the choice of evaluation method is necessary because none of the available methods provides a good prediction of the 'counter- factual", i.e., of what policies would have taken place in the absence of adjustment lending and, consequently, of what performance would have been in the absence of adjustment lending. A discussion of the choice of indicators is useful since the intermediate objectives of adjustment lending are many and since there are interactions between indicators. 2.1 Alternative Approaches 1/ Adjustment effectivenebs can be measured in several ways. One way is to define effectiveness of adjustment as the difference ir. the values of a set of indicators (defined below) after adjuetment was initiated with their values before adjustment. Another way is to measure effectiveness in terms of targets. For example, as is often done by the Operations Evaluation Department (OED) of the World Bank in individual country program evaluations, effectiveness is measured in terms of the difference between target values specified in the adjustment programs and realiLed values. The former approach to evaluating effectiveness is known as the before- after approach; the latter approach is known as the actual-versus-target approach. - 4 - Each approach has its advantages and disadvantages. The before- after approach is simple to apply, but is subject to serious biases because it assumes that all exogenous factors such as terms-of-trade variations and movements in interest rates which affect performance do not change between the pre- and post-adjustment periods. In the actual-versus-targets approach, one gets no information about how adjustment lending affects a country's performance. Furthermore, just as the situation prevailing before a country enters into an AL is not likely to be a good predictor of what would have happened in the absence of an AL, loan targets may be achieved because of favorable exogenous developments, irrespective of the policies adopted. Thus, neither of the two approaches give a good handle of the costs and benefits of the adopted policy course. The control-group approach, under its several variants, is designed to overcome the shortcomings described above. The simplest variant of the control. group approach assumes that the AL recipients and the non-adjustment lending (NAL) comparators face the same exogenous factors and the same initial conditions. Then, performance of the NAL group serves as an estimate of what the performance of the AL group would have been in the absence of adjustment programs. Because of its transparency and ease of application, this is the approach used in the statistical results reported in sections 4 and 5. This control group approach still does not provide a good basis for the counterfactual, i.e., of what policies would have been adopted in the absence of adjustment lending. To begin with, it does not take into account the fact that changes in the external environment may not be the same for both country groups. Second, it does not identify what policy choices would have been in the absence of AL. Third, it does not recognize that the amount of lending in addition to the presence or absence of an AL may have an effect on performance. In a companion paper (Faini et. al., 1989), we recognize these complications and evaluate performance using a modified control group approach. In that extended approach, we postulate that policy choice is a linear func.lon of the amount of adjustment lending and of changes in the external environment. Moreover, it may be argued that changes in performance attributed to AL may reflect the country-specific conditions which led countries to participate in AL. 2/ For example, countries with large external disequilibria had to cut public investment expenditures and also had a greater probability of entering AL. This possibility is usually not captured adequately in statistical evaluations of effectiveness. As shown by Golstein and Montiel (1986), if the determinants of macroeconomic performance are positively correlated with the country-specific conditions that led countries to participate in AL, then the control-group estimate of AL effects will overstate the true AL effects. 2.2 Implementation Subject to these caveats, our objective is to assess the effects of AL on macro performance. Of the 48 countries which received World Bank SALs or SECALS, only 30 countries received their first AL before 1985. Hence our evaluation will be for this subgroup of 30 AL recipients for which two to three years of data are generally available after the receipt of the first AL. This group of 30 countries will be compared with a control group of 63 countries. This latter group includes: countries that had not received their first AL by 1984; countries which participated in Ii4F adjustment programs; and countries which did not participate in -6- adjustment programs supported by either the IMF or the World Bank. We also rely on a two-way country group classification (low income: LY; middle income: MY) to control for the potential influence of institutional characteristics on programs' effectiveness. Obviously this definition of the non-adjustment lending (NAL) control group is not very satisfactory. Furthermore, most of the countries which participated iL World Bank-supported adjustment programs also participated in Fund-supported adjustmeiit programs. 3/ It is therefore interesting to evaluate both the effectiveness of AL and the effectiveness of Bank-Fund supported adjustment programs. We do both: in section 4 we assess Bank-Fund programs jointly, and in the remaining sections, Bank- supported programs only. It is not easy to decide on a "before and after' cut-off point when Bank-Fund programs are not initiated in the same year. In section 4 we take 1982 as the cut-off point, i.e. 1982 is the first year of assessment of Bank-Fund programs. This choice of cut-off point also corresponds to the year when adjustment lending intensified. In Section 4, we assess effectiveness of countries which received a non-negligible amount of Bank-Fund financial support. In a first round evaluation, we look for robustness by defining the control group as Low Credit (LC) recipient countries. The control group is constructed so as to include one-third of the 93 countries in our sample. Values of performance indicators for LC countries are then compared with the corresponding values for high credit (HC) recipients, also defined so as to include one-third of the countries in the sample. Slicing the sample into terciles and omitting from the comparisons the middle-tier is a crud3 way to add some robustness to the comparisons of the average value of indicators 'before and after" -7- adjustment lending. The first round evaluation relies on non-parametric statistics (e.g. number of countries in the HC group which improved performance vis-&-vis themselves and vis-&-vis the LC control group. The first round evaluation is therefore similar to other previous assessments (e.g. Balassa (1988)). The second round evaluation relies on tests of significance of changes in mean values (correcting for unequal variances and differences in sample size). Two tests are performed. The first test is for change in mean performance among HC recipients after adjustment lending. For indica- tor X (of which there are nine -- see table 3), we test the null hypothesis HC H G HC where subscripts 1 and 2 refer to *before" and "after". The second test is for effectiveness. For indicator x, we now test the null hypothesis #HC _ HC i LC LC ~x,2 - 'x,1 x.2 ,x,l In Section 5 we assess directly effectiveness of World Bank- supported lending for the group of 30 countries defined above. We use the same set of performance indicators. The main difference with the tests in section 4, is that the cut-off point is endogenous in the results reported in section 5. The same combination of non-parametric and parametric statistics is also used in section 5. However, because our main interest is in assessing the effectiveness of Bank-supported AL, we discuss results in greater detail in section 5. We use nine indicators to measure performance in four areas: growth, external balance, internal balance, and external debt. Growth indicators are average real growth rates for GDP, and exports, and changes in the investment/GDP ratio. External balance indicators are average values of the real effective exchange rate index (an increase in the value of the index signifies a real devaluation), and the current account deficit - 8 - as a fraction of GDP. Internal balance indicators are the government budget surplus/deficit as a fract!on of GDP, and inflation. External debt indicators are gross external debt and total debt service payments (interest and principal) both as fractions of exports of goods and nonfactor services. Obviously there are interdependencies among this group of indica- tors (see section 5). To take into account these interdependencies one would have to specify an appropriate structural model as given in Khan and Knight (1981) or Bourguignon, Branson, and de Melo (1989). Sucb approach would have the advantage of lending itself to counterfactual simulation analysis but would be more demanding on the data, especially for a large sample of heterogeneous countries. Furthermore, the nine indica- tors selected here do not assess the full set of targets under adjustment lending (e.g. public sector and financial sector reforms). The choice of indicators was dictated by the desire to have indicator values for as large a group of countries as possible. Because of this focus on breadth rather than depth, the results here are best viewed as complements to the more thorough country studies of effectiveness of adjustment lending. 3. The External Environment Under Adjustment Lending This section asks two questions: (1) was adjustment assistance under IMF and World Bank auspices mostly directed to countries which faced the most unfavourable external environmpnt; (2) was there complementarity or substitutability between commercial bank lending and IMF and World Bank lending. The reason for looking for evidence of the relative importance of private and official funds is that one of the objectives of official lending is that it was supposed to serve as a catalyst for private funding. -9- To measure how significant the deterioration in the external environment was, we quantify the impact of the external disturbances associated with declining terms of trade and rising real interest rates. External disturbances are measured over 1982-6, taking average values over 1976-81 as the base period. 41 The formula to compute the welfare costs of external disturbances is: 5/ (1) &W - iR R) * (D1Y)1 + (iX /PX., - 1) (X-Y)1 w - (PPM /PM1 - 1) (M/Y) = - RIR + TOT where subscripts 2 and 1 refer to 1982-6,and 1976-81 respectively, a bar over a variable means an average value over the relevant period and the variables are: R = average real interest rate (deflator is USGDP deflator) and the nominal interest rate is the weighted interest on concessional and commercial debt. y e real GDP is measured in dollars PX,PM = export and import price indices measured in dollars and deflated by USGDP deflatcr X,M = exports, imports D = gross outstanding debt, net of reserves - 10 - In equation 1, the first term measures the contribution of higher than expected interest payments and the remaining terms measure the welfare effect of changes in the terms of trade. The choice of periods implies that the welfare measure is expressed as a percentage of the average value of GDP during 1976-81. However, because cut-off points for the measurement of external shocks are debatable, we also report in table 1 calculations that use avera_- values over 1978-81 for the base period. 6/ The estimate of the size of the external shock is shown in table 1. The choice of base period does not affect significantly the estimated impact of higher real interest rates (RIR), but results in a significantly higher estimate for the impact of declining terms-of-trade (TOT) when the base period is 76-81 than when the base period is 78-81. In general, the estimated magnitude of externai shocks is greater for recipients of Bank ALs than for the control group. The same holds when countries are classified by intensity of Bank-Fund lending. The measure of Bank-Fund lending intensity is the average net IMF credit plus Bank SAL + SECAL credit during 1982-6, expressed as a percentage of average 1982-6 GDP. The cutoff points for high credit (HC) and low credit (LC) countries are chosen so that the sample of 93 countries is divided into three groups. The cutoff points are: a gross credit exceeding 0.7 percent of GDP for the HC group and a gross credit below 0.02 percent for the LC group. Middle credit group countries are excluded from the comparisons. The correlation betwen the estimated size of the external shocks (EXTSHCK) and the availability of Bank, Fund and private credit is shown in Table 2. The correlation between the size of the external shock and the intensity of Bank (Fund) credit is in both cases insignificant, suggesting - 11 - Table lt EXTERNAL SHOCKS AS PERCENTAGE OF AVERAGE GDP RIR a/ -TOT bl Group 78-81 76-81 78-81 76-81 All AL (30) c/ .011 (.010) .036 (.052) IAL (12) - .011 (.010) .045 (.065) NAL (63) .006 (.006) .033 (.049) High Credit (31) .012 (.010) .022 (.037) Low Credit (31) .003 (.003) .042 (.063) Low Income AL (12) .015 (.012) .067 (.091) NAL (24) .010 (.008) .016 (.037) Middle Income AL (18) .009 (.009) .017 (.028) NAL (39) .004 (.004) .041 (.054) Notes: A positive value indicates a worsening of the external environment in 1982-6 compared with the base. Number of countries in parentheses. a/ RIR: Real Interest Rate as measured by (R2 - R1) * (DTY)1, the first term of equation 1. b/ TOT: Terms-of-Trade Index as measured by (PM2/PM1 - 1) * (M/Y)1 - (PX2/PX1 - 1)* (X/Y)1 the third and second terms respectively of equation 1 (reversed to show a sign change). Note that the term identified in the title has a negative sign. The numbers are reported as positive ratios to GDP whereas equation 1 defines a deteriorating external environment as a negative number. Both RIR and TOT are measured using 2 different base periods. The 1st column for each is as a percentage of average GDP during 1978-81 and the second column for each is as a percentage of average GDP during 1976-81. c/ Intensive adjustment lending (IAL) countries are countries which received 3 or more ALs. d/ For definitions see World Bank (1989, Chapter 2). See appendix for country classifications. - 12 - Table 2: CORRELATION MATRIX: EXTERNAL SHOCKS AND EXTERNAL FINANCING af Pearson Correlation Coefficients IMF SAL SALIMF NTPRVGDP EXTSHCK -0.06 0.03 -0.01 -0.05 0.56 0.75 0.90 0.62 91 91 91 91 IMF 0.29 0.78 -0.19 0.005 0.0001 0.06 93 93 93 SAL 0.83 -0.23 0.0001 0.03 93 93 SALIMF -0.26 0.01 93 Notes: Results given in order: correlation coefficient, significance level, and number of observations. a/ Definition of variables (mean values in parentheses): All credit flows are cumulated over 1982-6 and are expressed as a ratio of average GDP during 1982-6. NTPRVGDP (-0.002) = Net private credit EXTSHCK (0.05) = S ze of external shock calculated and expressed as in table 1 (EXTSHCK = +RIR - TOT). SAL (0.003) = SAL + SECAL credit IMF (0.003) = Net IMF credit SALIMF (0.006) = SAL + SECAL + IMP - 13 - that Bank-Fund lending was not targeted to countries facing the most unfavorable external environment. However, as expected, there is a strong positive correlation between Bank and Fund credit reflecting the fact that much of Bank adjustment lending took place simultaneously with Fund stabilization programs. The significantly negative correlation between Bank-Fund adjustment lending and net private credit confirms the widely held view that Bank-Fund credit served as a substitute for private credit and suggests that the joint involvement of both institutions was not sufficient to serve as a catalyst towards commercial banks. 4. Performance Indicators for Bank-Fund Supported Adjustment Programs The evaluation of relative performance is summarized in Table 3. The measure of relative performance for each country group (LY, MY) is the change of HC countries relative to that of LC countries during 1982-6 compared with 1978-81. By way of example, the first entry on the top left- hand corner of table 3, 10(+), indicates that 10 out of 16 HC countries in the LY classification improved their GDP growth when compared with the LC control group. The (+) sign indicates that GDP growth was on average higher in the HC group relative to the LC group after 1982-6 than during 1978-81. The non-parametric statistics in Table 3 suggest a mild improvement in relative performance of HC countries vis-&-vis LC countries. If an equal weight is given to each indicator, 63 percent (59 percent) of HC countries in the LY(MY) classification improved vis-&-vis their corresponding countries in the LC group. Interestingly, the pattern is very similar across indicators for both LY and MY classification. - 14 - Table 8: PERFORMANCE INDICATORS FOR BANK-FUND ADJUSTMENT RECIPIENTS pj Low Middle Row Incoeo Own y Relative w/ Income Own k/ Relative S/ Sum (LY) (MY) Column 1 2 a 4 6 6 7 Number high credit (HC) 16 15 81 Number low credit (LC) 10 21 81 GDP growth (GDPK) 10(+) 10(+) e 20 Investment/GDP (GOIGDP) S(-) 2(-) * 7 Export growth (XK) 14(*) 9(o) 23 Real exchango rate (RER) 12(+) 13(+) e 26 Current Acct. Surplus/GDP(CASGDP) 12(+) o 11(+) e 23 Budget surpluu/GDP (BSGODP) 7(-) 8(+) 1S Inflation (INF) 10(-) 10(-) 20 External debt/Exports (DODX) 9(-) 6(-) 15 Debt Service/Exports (DSVX) 11(+) * 10(o) 21 Share showing improvement 0.83 0.69 a/ The numbers In the table show how many HC recipient countries in each classification lmproved each Indicator over LC coeparatore during 1982-4 (compared with 1978-81). The plus and minus signs show the direction of change of the average value of an Indicator in comparison with the average voluo of the sam indicator for the LC comparator (a plus is an improvement). For five indicators -- GDP growh, investment/GDP, export growth, current account surplus/GDP and overall budget surplus/GDP -- a positive difference in an Improvement. The numbers In the table show for each Indicator how many HC reciplint countries In each classification Improved vis-a-vls their LC comparators during 1982-86 (compared with 1978-81). For three Indicators -- inflation, external debt/exports and debt service/exports - a positive difference is a worsening (shown by a minus sign). For the real exchange rate, a greater r cl depreciation betwoen periods than that of the comparators is an Improvement. For example, if the average GDP growth of HC recipients in a subgroup was 0.2 percentage points lose than that of LC countries during 1978-81, and 0.1 percentage points loss during 1982-4 the differonce (0.1) is positive, and the relative performance of HC recipients improved. b/ PI H HC (see text) C ;HC UHC LC LC (e et c/ Ex,2 /s,l t I'x,2 /- x,l (see text) * Asterisks denote 10X (or more) significance level of test of equality of means. A blank denotes statistical Insignificance. - 15 - On the positive side, the patterns suggest relative improvements in: * GDP growth * The current account and export growth via a larger real exchange rate devaluation than in LC countries * Debt service On the negative side, the patterns suggest relative worsening in: * The ratio of real investment to real GDP * The budget deficit over GDP 7/ While the worsening fiscal position is an unambiguous sign of a worsening situation and is against the intents of adjustment programs, it could be argued that the falling investment share in GDP was intended, reflecting a cut in overly ambitious public investment programs. Unfortunately available data do not allow for a breakdown of investment into its public and private components. Even though the value of each performance indicator represents an average over several years, there is much va.-iance in the calculated means, especially among LY countries. As indicated in column 2, none of the changes in means are significant for the HC recipients in the LY classification. Since most of the countries in this classification are Sub-Saharan African countries where performance varied a lot and data is quite unreliable, this should come as no surprise. On the other hand, the changes in performance are statistically significant for approximately half - 16 - of the indicators in the MY category. Note the significant improvement in GDP growth during 82-86 (compared with 78-81) and the worsening investment performance. In relative terms, HC recipients among LY countries significantly raised their relative performance in terms of export growth, debt service and the current account. However, this relative improvement may not be a sign of sustainable recovery since investment dropped. HC recipients among MY countries improved upon their own performance for half of the indicators. However, in relative terms, their performance showed no significant improvement (except for the current account) and a significant r,lative worsening for investment. Here, too, in spite of a mild relative improvement, there are no signs of a sustainable recovery. As mentioned in section 2, this simple control group approach suffers from several shortcomings. In our companion paper (Faini et al. (1989), we specified a simple statistical model for each of the nine indicators examined here, in which we control for changes in the external environment by using the external shock indicator presented in section 3 and in which we model autonomous policy changes as a function of external shocks. For each indicator, we also controlled for Bank-Fund participation and for the intensity of adjustment (measured by the amount of Bank-Fund lending defined in section 3 and in the appendix). The main results of controlling for these effects can be summarized as follows: * Countries that participated in Bank-Fund programs had less growth loss. * The growth performance of countries that participated in Bank-Fund programs was positively correlated with the amount of Bank-Fund lending. - 17 - * PFrticipant countries fared worse on investment than the control group and the fall in investment was positively correlated with the amount of Bank-Fund lending. The positive effect of intensity of Bank-Fund lending on growth may come as a surprise since the intensity of adjustment lending was also negatively related to the investment to GDP ratio. As mentioned above, it could be argued that the intent of adjustment lending was to cut down on (largely) inefficient investment, and hence that the marginal efficiency of aggregate investment increased greatly in AL recipient countries. However, further investigation of the channels through which adjustment lending affected performance, revealed that intensity of adjustment lending significantly affected import growth. Given the severity of the foreign exchange constraint, we concluded that Bank-Fund lending contributed to growth mostly by allowing for the purchase of spare parts and other needed foreign-made inputs. To summarize, against the background of overall worsening indica- tors for developing countries as a whole, Bank-Fund supported adjustment programs appear to have resulted in a relative improvement in growth but at the expense of lower investment rates and a slightly worsening fiscal situation. It would also appear that the positive correlation between intensity of adjustment lending and improvement in growth can be attributed to a relaxation of the foreign exchange constraint. 5. Performance for Bank Adjustment Lending Recipients The remainder of the paper abstracts from interactions between Bank and Fund adjustment programs. The analysis therefore deals only with - 18 - the group of 30 recipients of Bank SALs and SECALs identified earlier. For this group of 30 countries, enough time has elapsed for a preliminary 'before and after" analysis. The focus here is on how the 30 AL countries (which had loans before 1985) compare with non-adjustment lending (NAL) countries. In this exercise, NAL countries are now defined as countries which had not received an adjustment loan from the Bank until 1985. We use the same LY and MY classification as before but do not control for Fund participation or for lending :ntensity. We use the same set of indicators as before, and the same averaging method, namely each performance indicator is an average value, now taken over the three years preceding receipt of the first Bank AL. As in Balassa (1988), the year of receipt of the first Bank AL is excluded from the comparisons and the average value for each indicator during the loan period is taken over three years following receipt of the first AL (excluding the year of the loan). Even though we are not controlling for IMF participation, the 30 AL countries which received adjustment lending before 1985 account for about 84 percent of the commitments of adjustment lending and 59 percent of net IMF credit disbursements through 1986. These comparisons should therefore be viewed as complementary with, and a check for, the comparisons reported in section 4 above. Table 4 gives the average value of the performance indicators for the group of 30 AL countries, and the comparator group of 63 NAL countries. Both groups are further broken into the LY and MY classifications used earlier. Comparing the average values for the indicators for the AL and NAL groups enerally reveals poorer performance in the years preceding adjustment lending. With the exception of exports, on average, the growth indicators show no improvement. As before, there is a marked decline in Table 4: AVERAGE VALUES OF PERFORMANCE INDICATORS FOR AL COUNTRIES Low Income (LY) a/ Middle Income (MY) b/ 3 years 3 years 3 years 3 years before after before after AL NAL AL NAL AL NAL AL NAL 1 GDP growth 0.030 0.032 0.023 0.029 0.023 0.046 0.022 0.024 2 Invest/GDP 0.19 0.21 0.18 0.19 0.25 0.28 0.19 0.25 * 3 Export growth -0.02 0.06 -0.00 0.01 0.03 0.05 0.06 0.03 Oh 4 Real exchange rate 1.06 1.00 1.94 1.04 1.02 1.00 1.09 1.01 5 Current account -0.09 -0.08 -0.07 -0.07 -0.07 -0.05 -0.05 -0.05 surplus/GDP 6 Budget surplus -0.08 -0.04 -0.08 -0.04 -0.06 -0.06 -0.06 -0.06 7 Inflation 0.26 0.14 0.33 0.13 0.26 0.20 0.38 0.32 8 External debt/exports 3.96 2.03 4.45 3.03 1.35 1.06 2.10 1.68 9 Debt services/exports 0.21 0.10 0.23 0.16 0.28 0.17 2.31 0.22 a/ Number of AL recipients: 12. Number of NAL recipients: 24. b/ Number of AL recipients: 18. Number of NAL recipients: 39. - 20 - the share of investment in GDP. On the other hand the external balance indicators show an improvement, a reflection of the fact that AL countries generated the necessary non-interest current account surplus to service the larger interest payments on their external debt. As has been analyzed elsewhere (e.g. Cohen (1988), Rodrik (1988)), the burden of external debt servicing has been accompanied by a deteriora- tion (inflation) or by a stagnation (budget deficit) in the indicators of internal balance. Finally, for both LY and MY recipients, external debt indicators deteriorated. The combination of stagnating and/or deteriora- ting indicators for internal balance and external debt raise the issue of the long-run sustainability of the adjustment program achieved to date. Table 5 reports values of adjustment indicators for the AL and NAL groups using the same format as in Table 3. Comparing the results in both tables indicates similar results for MY countries. However, for LY countries, average value comparisons for 4 out of the 9 indicators change sign. As in Table 3, tests of change in own performance show insignificance for LY recipients. Finally, when the same comparisons are made for countries that received 3 or more ALs, the share of indicators showing improvement rises to 62 percent for the LY group and to 67 percent for the MY group. It was mentioned earlier that there are interdependencies among performance indicators. Table 6 shows the correlations for the nine indicators of table 4. (All Pearson correlations and significance levels appear for correlations at the 5Z significance level or better). Somewhat surprisingly, there is no significant correlation between real exchange rate depreciation and export growth. This could be explained by the fact that partial correlations omit other relevant variations like an activity - 21 - Table 5: PERFORMANCE INDICATORS FOR BANK ADJUSTMENT RECIPIENTS Low Middle Incom Incom Row (LY) Own b/ IAL c/ (MY) Own b/ IAL SI Sum IAL Number of AL recipients 12 6 18 7 30 12 Number of NAL countries 24 24 39 89 6s 63 1 GDP growth 6(-) a 11 (+) 6 16 8 2 Invest/GDP 6 (.) 2 6 (-) e 2 11 4 3 Export grosth 8 C.) 4 9 (W) 5 17 9 4 Real exchange ratt 9 C.) 5 11 C.) 6 20 11 S Current sect. surplus/GDP 6 C.) 8 16 Ci) 6 21 9 6 Budget surplus/GDP a (.) 8 6 (W) 4 12 7 7 Inflation 8 (-) 1 18 C-) B 16 7 8 Ext. debt/exports 7 (+) 4 9 (-) a 3 16 7 9 Debt service/exports 8 (W) 8 9 (.) S 17 8 Share showing improveme nt 0.58 0.62 0.66 0.67 0.64 0.65 ,/ The numbers in the table show for each indicator how many AL countries in each classification improved vis-a-vis their NAL comparators after AL (compared with the pre-loan period). In this table, the *before' and after' perlod is determined *ndogenously with the cut-off point determined by the year of receipt of the first SAL or SECAL. (The year of receipt of the first loan Is excluded from the calculations.) The plus and minus signs show the direction of change of the average value of an Indicator in comparison with the average value of the same indicator for the NAL comparator (a plus Is an lmprov-ent). Indicators and interprotation of results are the same as In Table 8. b/ As In Table 8, asterisks denote lOX (or more) significance le-l of test of *quality of means for the AL and NAL results, and a blank denotes statistical Insignificance. Only the test AL, # L2 is reported (in the column labeled "own%), since tlo test for effectiveness cannot be performed because the averaging over the control group (NAL countries) Is done repeatedly. c/ IAL refers to 12 Intensive Adjustmont Lending recipients (8 or more ALS). - 22 - variable in an equation tkhat would relate export growth to the real exchange rate. Keeping these omissions in mind (which apply to all correlations in table 6), growth is positively correlated with the share of investment in GDP and negatively with inflation (which is itself negatively correlated with the budget surplus). 8/ Though partial, these correlations give some support to the widely shared view that stabilization helps growth. The real exchange rate is negatively correlated with both external debt indicators suggesting that countries which had the largest real exchange rate depreciation had the greatest improvement in their external debt indicators (ratios expressed in constant dollars). Finally, the partial correlations suggest that countries which improved the most their current account did so by reducing the investment component of absorption. 6. Growth-External Adjustment Tradeoffs for Recipients of Bank Adjustment Programs For most countries, foreign borrowing was foreclosed after 1981, which implied that the absorption-income gap had to be reduced. The resulting reduction in absorption is referred to by Corden (1989) as the primary (or inevitable) cost of reducing a current account deficit. The goal of adjustment lending in the medium term was to raise capacity utilization and to shift resources towards tradables beyond levels attainable in the absence of adjustment lending. This goal was to be achieved by, among others, incentive reform and institutional reforms, both of which would reduce the extent of relative price rigidities. Since relative price rigidities cause a secondary cost of adjustment, one can say that adjustment lending was intended to reduce the secondary cost component of adjustment by increasing supply responsiveness. 9/ - 23 - Table 6: CORRELATIONS BETWEEN ECONOMIC PERFORMANCE INDICATORS: (SIGN AND SIGNIFICANCE LEVEL IN PARENTHESES) Low Income Middle Income 30 "ALI Countries 30 "ALI Countries 30 "ALI Countries GDP GDIGDP (+, .02) INF (-, .006) GDIGDP (+, .03) GDIGDP CASGDP (-, .03) CASGDP (-, .005) DODX (-, .01) DSVX (-, .03) RER DODX (-, .0001) DODX (-, .001) DODX C- .03) DSVX (-, .001) DSVX (-, .0004) BSGDP (+ .02) BSGDP INF (-, .003) INF (-, .01) DODX (-, .01) INF DODX (+, .002) DODX DSVX (+, .0002) DSVX (+, .0003) Definition of Variables: GDP - GDP Growth GDIGDP - Investment/GDP RER - Real Exchange Rate Index BSGDP - Budget Surplus/GDP INF = Inflation DODX - External Debt/Exports DSVX = Debt Service/Exports CASGDP - Current Account Surplus/GDP - 24 - In this section, we look for evidence of whether the group of adjustment lending recipients were able to achieve external adjustment at less cost in terms of foregone growth. To test whether adjustment lending achieved its objectives, we investigate whether the change in growth and change in current account is statistically different for AL recipient countries. To test this hypothesis we estimated the following reduced form equations (2) A GDPKi - aO + al D + a2 A (RESBAL/GDPK)i + a3 D A (RESBAJIGDPK)i + Ei where i = 1, ... 93 countries D - 1 for 30 AL countries 0 for 63 NAL countries a is the percent change between the three-year average after receipt of first AL and the three-year average before receipt of the first AL. and A (RESBAL/GDP) is the terms of trade adjusted change in the resource balance over GDP; AGDPK is the change in real GDP. The estimation is for the data appearing in Figure 1. For countries that did not receive a Bank adjustment loan, 1982 is the year chosen to measure the change in GDP. After correcting for heteroskedasticity, the results are (t-values in parentheses): Figure 1: GDP/RESOURCE BALANCE TRADE-OFF AGDPK 0.14 + 0.12 + 0.10 1 0~~~~ | o 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 0.06 + 0 0 1 0 0 0 0 0.024+ ° 1 0 00.02 I. 0~~~~~~~~~~~~~~~~~~~~~ -0.100 1 -0.12 + -0.14 + I 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -0.16 + I 4 -0.16 -0.11 -0.06 -0.01 0.04 0.09 0.14 0.19 0.24 - 26 - (3) A GDPKi - -0.01 + 0.02 D (-2.69) (1.94) -0.19 A RESBALi - 0.71 D . A RESBALi (-3.03) (-0.31) F - 10.8 The intercept dummy is of the expected sign "upward shift for AL). However, the slope dummy is of the wrong sign (i.e. steeper resource balance to GDP tradeoff for AL countries) but insignificant. Because of suspected multicollinearity between the slope and intercept dummies, we exclude the slope dummy, resulting in a further improvement (4) A GDPKj - -0.01 + 0.02 D - 0.20 A RESBALj (-2.71) (2.13) (-3.50) F - 14.5 The GDP/Resource balance trade off is reported in Figure 1. The regression results do not give direct support to the hypothesis that adjustment lending helped reduce the secondary burdens of adjustment since the trade-off between growth and external balance improvement is the same for both groups of countries. However, the greater trade balance improvement among Bank AL recipients is consistent with the earlier results in Table 4 which show a higher initial debt burden and consequently greater required current account improvement among Bank AL recipients. And the higher growth among Bank AL recipients is consistent with foreign exchange availability, reducing the tightness of the foreign exchange constraint. - 27 - 7. Conclusions This paper has used a simple control group approach to assess quantitatively the economic performance of countries that received adjust- ment lending (AL) from the Bank and the Fund. The statistical evaluation was designed to address the following: (1) was there complementarity between Bank-Fund lending and private lending? (2) Did AL recipients improve their performance vis-a-vis their own past performance? (3) Did AL recipients improve their performance vis-a-vis suitably chosen comparators? With respect to the complementarity between Bank-Fund lending and commercial lending, there was negative correlation between the amount of Bank-Fund lending and private lending during 1982-6. It would thus appear that Bank-Fund lending came as a substitute rather than as a complement to private sector lending. With respect to their own past performance, adjustment lending countries improved their external position, a reflection of the fact that they generated a sufficient trade balance surplus to service their external debt. However, fiscal (and inflation) indicators deteriorated, a sign of remaining macroeconomic imbalances, thus casting some doubts cn the sustainability of adjustment. Finally, growth rates fell, a reflection of deteriorating terms-of-trade and of the difficulties in achieving the required reduction in absorption. With respect to performance vis-a-vis comparators, non-parametric statistics for the selected group of nine economic indicators show that, overall, AL recipients improved relative to comparators. However, the share of indicators showing an improvement (the average of the number of countries -- over all indicators -- showing an improvement) varied between 53 percent and 63 percent depending on classifications. For the group of - 28 - 12 countries which received 3 or more ALs, the shar of indicators showing an improvement varies between 62 percent and 67 percent. Statistical tests of difference of means for individual indicators suggest that these improvements were often statistically insignificant. The evidence is that some mild relative improvement occurred. These results are confirmed and strengthened in related work where an attempt is made to control for the potentially negative effects of the external environment on performance. Controlling for this effect, AL recipients appear to do better than their comparators now defined as countries which received neither Bank nor Fund adjustment lending. - 29 - Footnotes 1/ The following discussion draws on Faini, et. al. (1989). 2/ This is known as sample-selectivity bias. For a further discussion, see Coldstein and Montiel (1986), Khan (1988), Faini, et. al. (1989). 3/ The group of 30 AL recipients account for 84 percent of the commitments of adjustment lending, and for 59 percent of net IMF credit disbursements through 1986. 4/ 1982 is chosen as breakpoint to correspoond with the breakpoint selected for measuring performance. 5/ The formula is derived from a two-period maximization by firms and households under assumptions of perfect competition and wage-price flexibility. See Dornbusch (1986, pp. 354-6). 6/ Note that the definitions of both RIR and TOT are dependent on the signs used. Specifically, RIR is interpreted as a worsening if it is positive (higher interest rates) and an improvement if negative. Conversely, TOT is defined as X-M, therefore showing improvement by a positive number and worsening by a negative. Thus, in equation 1, improvement in the external environment is a positive number and a worsening is a negative number, whereas for convenience, the opposite sign convention is used in table 1. 7/ It should be noted, however, that the fiscal data is very weak and sparse. 8/ See Kormendi and Meguire (1985) for a more complete cross-section model of the determinants of growth. 9/ Ort}.odox stabilization programs usually concentrated on expenditure redt.cing policies to reduce external imbalances. By contrast, the Bank-Fund supported programs initiated in the early eighties have placed greater emphasis on expenditure switching and particularly on output augmenting policies, as a component of adjustment. Hence the characterization of these programs as 'adjustment with growth." Thus Bank-Fund supported adjustment programs were supposed to increase demnud and supply responsiveness to relative price shifts so as to increase the effectiveness of expenditure switching policies. This objective was to be achieved by greater reliance on the price mechanism and by a reduction of distortions. - 30 - References Balassa, B. 1988. "Quantitative Appraisal of Adjustment Lending," PPR Working Paper Series No. 79, World Bank, Washington, D.C. Bourguignon, F., W. Branson, and J. de Melo. 1989. "Macroeconomic Adjustment and Income Distribution: A Macro-Micro Framework," OECD, forthcoming. Calvo, G. 1986. "Incredible Reforms," mimeo, University of Pennsylvania. Cline, W.R. 1985. "International Debt: From Crisis to Recovery," American Economic Review, pp. 185-90. Cohen, D. 1988. 'The Management of the Developing Countries' Debt Guidelines and Applications to Brazil," World Bank Economic Review, pp. 77-104 Corden, M. 1989a. "Macroeconomic Adjustment in Developing Countries," World Bank Research Observer, pp. 51-64. Dell, S. 1983. "Stabilization: The Political Economy of Overkill," in J. Williamson, ed., IMF Conditionality, pp. 7-46. Institute for International Economics, Washington, D.C. Diaz-Alejandro, C. 1979. "Southern Cone Stabilization Plans," in W. Cline and S. Weintraub, eds., Stabilization Policies in Developing Countries, Brookings Institute, Washington, D.C. Dornbusch, R. 1985. "Policy and Performance Links Between LDC Debtors and Industrial Nations," Brookings Papers on Economic Activity, No. 2, pp. 303-68. Dornbusch, R. 1986. "The Effects of OECD Macroeconomic Policies on Non- Oil Developing Countries," World Bank Staff Working Papers, No. 793, World Bank. Faini, R., J. de Melo, A. Senhadji-Semlali, and J. Stanton 1989. "A Statistical Analysis of Performance Under Bank-Fund Adjustment Lending" (mimeo), World Bank. Genberg, A. and A.K. Swoboda. 1987. "The Medium-Term Relationship Between Performance Indicators and Policy: A Cross Section Approach," EPD Discussion Paper No. EPD-01, The World Bank. Goldstein, M. and P. Montiel. 1986. "Evaluating Fund Stabilization Programs with Multicountry Data: Some Methodological Pitfalls," IMF Staff Papers, Vol. 33, No. 2, pr. 314-44. Khan, H. 1988. "The Macroeconomic Effects of Fund-Supported Adjustment Programs: An Empirical Assessment," (mimeo). - 31 - Khan, M. and M. Knight. 1981. 'Stabilization Programs in Developing Countries: A Formal Framework," Staff Papers, Vol. 28, No. 1, pp. 1-53. Kormendi, R. and P. Meguire. 1985. "Macroeconomic Determinants of Growth: Cross-Country Evidence,' Journal of Monetary Economics. Mitra, P.K. 1984. 'Adjustment to External Shocks in Selected Semi- Industzial Countries, 1974-1981," mimeo, Public Economics Division, The World Bank. Rodrik, D. 1988. 'The Welfare Economics of Debt Service," NBER Discussion Paper No. 2655. Sachs, J. and H. Huizinga. 1987. "U.S. Commercial Banks and the Developing Country Debt Crisis," Brookings Papers on Economic Activity. World Bank. 1989. AdJustment Lending: An Evaluation of Ten Years of Experience, Policy and Research Series No. 1, Country Economics Department, The World Bank, Washington, DC. - 32 - Appendix Al. Data Sources All data were extracted from the World Bank's BESD and ANDREX data bases. Data in constant dollars were obtained by using the World Bank's atlas exchange rate converstion factor. In the calculation of external shocks (equation 1), terms of trade indices were obtained by dividing current exports and imports (expressed in dollars) by the constant values. Similar results were obtained when the terms-of-trade indices were calculated from current and constant local currency values from National Accounts data. To calculate the effective interest rate on external debt, we applied LIBOR + 1 to the share of total non-concessional debt and the implicit interest rate from interest payments on concessional debt. For Bank-Fund lending, we constructed two variables: one based on gross IMF credit (results in the text); another on net IMF credit, where credit was calculated as IMF purchases less IMF repurchases. In both cases, Bank SAL credits are the sum of SAL and SECAL commitments. Results based on gross IMF credit are not reported because they are extremely close to those obtained with the net IMF credit definition. A2. Country Classifications Table Al describes the sample of 93 countries and the classifi- cations used in the text. For the group of 30 Bank AL recipients, the year of receipt of the first loan is indicated in parentheses. The subset of intensive adjustment lending (IAL) recipients is defined as those countries which received three or more ALs. The IAL group includes 12 countries. - 33 - Table A.l: SAMPLE CLASSIFICATIONS (93 COUNTRIES) Nation Nation H Argentina H Morocco (1984, 6) Burundi H Madagascar L Benin Mexico (1983, 3) Bangladesh H Mali Bolivia (1980, 3) L Malta Brazil (1983, 3) H Mauritania Barbados H Mauritius (1981, 3) L Burma H Malawi (1981, 4? L Botswana L Malaysia H Central African Republic H Niger H Chile Nigeria (1983, 2) L China L Nicaragua H COte d'Ivoire (1981, 3) L Nepal L Cameroon Pakistan (1980, 4) Congo H Panama (1983, 2) Colombia Peru H Costa Rica (1983, 2) Philippines (1980, 4) L Cyprus L Papua New Guinea H Dominican Republic Portugal L Algeria L Paraguay H Ecuador L Rwanda L Egypt Sudan (1980, 2) L Ethiopia H Senegal (1980, 3) Fiji L Singapore Gabon Sierra Leone (1984, 1) H Ghana (1983, 6) L El Salvador H Guinea H Somalia Gambia L Seychelles H Guinea-Bissau (1984, 2) ' Syria L Greece L Chad L Guatemala H Togo (1983, 2) L Guyana (1981, 1) Thailand (1982, 2) L Hong Kong L Trinidad and Tobago Honduras H Tunisia Haiti Turkey (1980, 9) H Hungary Tanzan4a (1981, 2) Burkina Faso H Uganda (1983, 1) Indonesia H Uruguay (1984, 2) India L Venezuela L Israel Yemen H Jamaica (1979, 8) Democratic Yemen Jordan Yugoslavia (1983, 2) H Kenya (1980, 3) L South Africa Korea (1981, 3) H Zaire H Liberia H Zambia (1984, 4) L Sri Lanka H Zimbabwe (1983, 1) L Lesotho Notes: H = high credit; L = low credit (classification used in Table 3). PPR Working Paper Series Title Author Date Contact WPS166 Achieving and Sustaining Universal Primary Education: International Experience Relevant to India Nat J. Colletta March 1989 M. Philliph Margaret Sutton 75366 WPS167 Wage Determination In Rural Bangladesh: The Welfare Implications Martin Ravallion WPS168 Technological Change from Inside: A Review of Breakthroughsl Ashoka Mody March 1989 W. Young 33618 WPS169 Financial Sector Reforms In Adjustment Programs Alan Gelb Patrick Honohan WPS170 General Training Under Asymmetric Information Eliakim Katz Adrian Ziderman WPS171 Cost-Effectiveness of National Training Systems In Developing Countries Christopher Dougherty March 1989 C. Cristobal 33640 WPS172 The Effects of Peru's Push to Improve Education Elizabeth M. King March 1989 C. Cristobal Rosemary T. Bellew 33640 WPS173 Staffing and Training Aspects of Hospital Management: Some Issues for Research Julio Frenk Enrique Ruelas WPS174 Trade Restrictions with Imported Intermediate Inputs: When Does the Trade Balance Improve? Ramon E. Lopez Dani Rodrik WPS175 An Integrated Model of Perennial and Annual Crop Production for Sub-Saharan Countries Robert D. Weaver WPS176 Credit Rationing, Tenancy, Productivity, and the Dynamics of Inequality Avishay Braverman Joseph E. Stigiltz PPR Working Paper Series Title Author Date Contact WPS177 Cash-Flow or Income? The Choice of Base for Company Taxation Jack M. Mintz Jesus Seade WPS178 Tax Holidays and Investment Jack M. Mintz WPS179 Public Sector Pricing In a Fiscal Context Christopher Heady WPSI8O Structural Changes in Metals Consumption: Evidence from U.S. Data Boum-Jong Choe WPS1S1 Public Finance, Trade and Development: What Have We Learned? Johannes F. Linn Deborah L. Wetzel WPS182 The Experience of Latin America With Export Subsidies Jullo Nogues WPS183 Private Investment. Macromanagement and Trade Liberal;zatlon: The Case of Mexico Alberto R. Musalem WPS184 Women and Forestry: Operational Issues Augusta Molnar WPS185 Uniform Trade Taxes, Devaluation, and the Real Exchange Rate: A Theoretical Analysis Stephen A. O'Connell WPS186 The Uruguay Negotiations on Subsidies and Countervailing Measures: Past and Future Constraints Patrick A. Messerlin WPS187 The Flexibility of Wages When the Output Price Changes: An Empirical Study of 13 Industrial Countries Menahem Prywes WPS188 International Comparisons of Wage and Non-Wage Costs of Labor Luis A. Riveros WPS189 The Treatment of Companies under Cash Flow Taxes: Some Administrative, Transitional, and Internationa, Issues Emil M. Sunley WPS190 Macro Performance Under Adjustment Lending Riccardo Faini April 1989 M. Ameal Jaime de Melo 61466 Abdel Senhadji-Semiali Julie Stanton