Policy Research Working Paper 9971

State-Owned Enterprises
as Countercyclical Instruments
Experimental Evidence from the Infrastructure Sector

Matias Herrera Dappe
Aldo Musacchio
Carolina Pan
Yadviga Semikolenova
Burak Turkgulu
Jonathan Barboza

Infrastructure Chief Economist Office
March 2022
Policy Research Working Paper 9971

Abstract
This paper examines the effects of a negative macroeconomic estimation—indicate that fully owned SOEs (FSOEs) that
shock on the financial performance of state-owned enter- suffered a negative macroeconomic shock performed worse
prises (SOEs) in infrastructure. It exploits the differential than those that did not. FSOEs that suffered a shock also
effects of a drastic fall in oil prices (in 2014–15) on SOEs received large fiscal transfers from the government to cope
in energy-rich countries relative to SOEs in non-energy-rich with the shock for three years after the shock. Despite the
countries, matching firms based on their fuel expense transfers, they reduced their capital expenditures as a con-
ratio. The results—based on a balanced sample using sequence of the shock.
coarsened exact matching and a differences-in-differences

This paper is a product of the Infrastructure Chief Economist Office. It is part of a larger effort by the World Bank to
provide open access to its research and make a contribution to development policy discussions around the world. Policy
Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The author may be contacted
at mdappe@worldbank.org.

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Produced by the Research Support Team
State-Owned Enterprises as Countercyclical Instruments: Experimental Evidence from the
Infrastructure Sector 1

Matias Herrera Dappe, World Bank
Aldo Musacchio, Brandeis University and NBER
Carolina Pan, Power for All
Yadviga Semikolenova, World Bank
Burak Turkgulu, World Bank
Jonathan Barboza, Independent Consultant

JEL: H32, H4, H54, L9

Keywords: State-owned enterprises, infrastructure, macroeconomic shock, fiscal transfers

1
This paper was peer reviewed by Luis Andres, Lead Economist and Sector Leader, World Bank, and Martin
Melecky, Lead Economist, World Bank. It was prepared as a background paper for the Infrastructure Vice-
Presidency’s flagship Fiscal Costs and Risks from Infrastructure in an Era of High Debt Vulnerabilities and
presented at the corresponding authors’ workshop, held July 13, 2021, benefiting from feedback provided
by the workshop participants. The authors thank in particular Vivien Foster and Teresa Ter-Minassian for
their helpful comments and suggestions.
1 Introduction

Despite privatizations and reform efforts, state-owned enterprises (SOEs) are still key players in
infrastructure around the world. As a result of failed privatizations, re-nationalizations, and failed
reforms to open the sector to private investment, governments still own, control, and influence
the operation of a large portion of the world’s infrastructure firms. For example, Foster et al.
(2017) find that in 2015, of the 88 countries whose energy reforms they tracked, a third had an
SOE monopoly operator and 22 percent had a single-buyer model, which is almost always an SOE.
The remaining half of the countries they study had hybrid models in which SOEs and private
companies interact and compete with one another. This resilience of state ownership in
infrastructure is in part a consequence of the haphazard or incomplete privatization process of
large infrastructure firms, which left governments with full or partial ownership of what were
often the politically sensitive, harder to sell SOEs (Estache, Gomez-Lobo, and Leipziger 2001;
Nellis 2005).

A key policy question is whether these SOEs can play a countercyclical role during
macroeconomic crises. Do negative shocks worsen performance and adversely affect the long-
term health of these companies? The answer may not be straightforward, as SOEs use a variety
of mechanisms to cope with negative shocks, including soft budget constraints (Kornai, Maskin,
and Roland 2003; Musacchio and Lazzarini 2014). SOEs may also have access to soft loans from
other SOEs or the Treasury, which they use to withstand a shock. They can borrow at below-
market rates from the market or from other SOEs or financial intermediaries, and they can
receive operational subsidies, tax breaks, and tax deferrals from the government (Herrera Dappe
et al. 2022). Recent studies that use matching techniques to compare partially privatized SOEs
(PPSOEs) and private firms in a variety of sectors show that during severe recessions, the former
have a slower rate of reduction in labor than similar comparable private firms (Lazzarini and
Musacchio 2018). Still, most of the discussions about the role of SOEs during crises lack
experimental or quasi-experimental studies that shed light on how they behave during
macroeconomic downturns.

This paper focuses on two issues. First, it examines whether SOEs follow policies that are
consistent with a countercyclical role or whether they exacerbate the negative effect of the shock
by generating fiscal risk. Second, it investigates whether these shocks have different effects on
fully owned SOEs (FSOEs) and partially privatized SOEs (PPSOEs). 2

The findings are relevant for the literature that documents how fiscal surprises from SOEs can be
problematic for macroeconomic stability. One strain of the literature examines the fiscal cost of

2
In this study, PPSOEs refer to firms that are majority-owned or controlled but not fully owned by the
state. Full ownership is defined as state control of at least 99.5 percent of the share capital.
2
bailouts in a variety of countries after different types of macroeconomics shocks (Bova et al.
2016, 2019). Another warns about the fiscal risk from contingent liabilities that can arise during
crises (Lewis and Mody 1998; Polackova Brixi 1998; Polackova Brixi and Mody 2002; Polackova
Brixi and Schick 2002). This paper contributes to the literature by conducting a quasi-experiment
that shows the fiscal cost of SOEs (that is, the fiscal risk of SOEs) during a negative macroeconomic
shock using data from multiple countries in a differences-in-differences setting.

We examine the effects of the drastic fall in the price of oil in 2014 as a natural experiment that
treated infrastructure SOEs in energy-rich countries with a negative macroeconomic shock,
especially relative to similar SOEs in non-energy-rich countries. We use data from the World Bank
Infrastructure SOEs Dataset, which tracks the financials of over 100 firms in 19 countries from
2000 to 2018 and includes a more balanced panel for 2009–18. We use data on three sectors—
airports, airlines, and power—to maximize intrasector variation over time and across countries.

Ideally, we would like to isolate the negative macroeconomic effect of a drastic fall in the price
of oil in energy-rich countries from any confounding effect that may come from having firms
enjoy cheaper fuel inputs. Finding a large sample of SOEs for which the latter effect is null or
insignificant is difficult, however. For that reason, we isolate the positive effects stemming from
the fall in the price of oil by comparing firms in treatment countries and control countries
matched according to fuel costs as a percent of expenses. That is, we compare firms with similar
sensitivities to fuel costs within the same industry in treatment (energy-rich) countries and
control (non-energy-rich) countries. Doing so allows us to capture the effect of the negative
macroeconomic shock that is caused by the plummeting in the price of oil, as it negatively affects
the demand of firms in energy-rich countries. Thanks to the detailed nature of our database, we
can compare firms in multiple infrastructure sectors in a variety of treatment and control
countries, controlling for fixed effects for industry and World Bank regions and income levels.

Infrastructure is a good setting in which to study the effects of large macroeconomic shocks on
SOEs for at least four reasons. First, infrastructure SOEs are asset heavy and need continuous
capital expenses to sustain and renew their existing capacity. Second, infrastructure SOEs are
usually large, if not the largest, employers in the sectors and countries in which they operate. As
such, their workers have significant political power, making it harder for managers and politicians
to make painful adjustments to employment and expenses when they face a negative shock.
Third, infrastructure SOEs are usually strategic suppliers of inputs for companies and households,
making them more prone to having controlled tariffs, which can exacerbate the effects of a
negative shock. Fourth, the infrastructure sector tends to be heterogeneous in terms of the
ownership of large companies. It often includes some dominant firms with full state ownership
as well as large firms with a mix of public and private ownership.

3
One would assume that reforms that facilitate cost recovery in infrastructure firms with more
flexible (and less political intervention in) tariffs should lead to a faster recovery of these firms
during a negative shock. For instance, partial privatization of infrastructure firms is usually
accompanied by a full or partial liberalization of tariffs—by introducing a regulator to set tariffs,
for example—as well as by investment and outreach conditions that improve the quality and
reach of their services (see, for instance, Galiani, Gertler, and Schargrodsky 2005 and Andrés,
Foster, and Guasch 2006). The problem is that during negative shocks, even in these partially
privatized firms, the temptation of the government to intervene to control tariffs or to prevent
these firms from firing redundant labor can be high (Lazzarini and Musacchio 2018).

Therefore, we pay particular attention to the heterogeneity of the effects of a negative shock in
FSOEs and PPSOEs. Our working hypothesis is that FSOEs usually enjoy more guarantees than
similar PPSOEs (Borisova and Megginson 2011; Borisova et al. 2015; Wagner et al. 2018) and
arguably have easier access to government bailouts, providing them with a softer budget
constraint, which allows them to respond differently to negative shocks relative to PPSOEs. In
particular, the paper examines whether FSOEs in treated countries avoided cutting capital
expenditures (CAPEX) or employment when faced with a negative shock, thus behaving like
FSOEs in countries that did not suffer the negative shock.

We find that after a sharp decline in the price of oil, firms in energy-rich countries (the treated
group) behaved differently from SOEs in the control group in at least three ways. First, treated
infrastructure SOEs seem to have experienced worse performance for several years after the
negative shock, particularly in terms of earnings before interest, taxes, depreciation, and
amortization (EBITDA) net of the operations subsidies SOEs receive. Second, we find a decline in
CAPEX in treated SOEs relative to the control group. Third, FSOEs performed much worse in
treated countries than in the control group. We also find some evidence that they received larger
fiscal injections. That is, we find evidence that FSOEs play a budget-contracting, rather than a
budget-expanding, role during crises This finding is consistent with the fact that governments
have to use many tools to help SOEs cope with negative shocks.

The rest of this paper is organized as follows. Section 2 summarizes the literature. Section 3
describes the data and the experimental setting. Section 4 describes the empirical methodology.
Section 5 presents the findings. Section 6 summarizes the paper’s main conclusions.

2 Literature Review

Until now, most of what we know about the weak performance of infrastructure SOEs comes
from quasi-experimental settings that examines performance before and after privatization and
reform. Studies of power utilities show that privatization can lead to improvements in

4
productivity and service quality (Andrés, Foster, and Guasch 2006). 3 The literature also
documents that in developing countries, reforms in the power sector have been slow and
suffered significant reversals, leaving SOEs responsible for the bulk of the generation,
transmission, and distribution of power (Eberhard et al. 2016; Foster et al. 2017). A companion
paper (Herrera Dappe et al. 2022) studies the performance of infrastructure SOEs relative to
comparable private infrastructure firms. It shows that infrastructure SOEs are larger relative to
GDP, have larger liabilities relative to GDP, have higher employment costs as a share of revenues,
are less efficient, and have worse returns on assets than comparable private firms.

This evidence suggests that infrastructure SOEs underperform similar private firms and can
generate significant fiscal risk. But little is known about the countercyclical role they can play
during macroeconomic crises. SOEs can take advantage of their soft budget constraint to
promote employment and investment (Kornai, Maskin, and Roland 2003; Musacchio and
Lazzarini 2014). How much they do so when facing negative shocks is not clear, however.

There is more evidence of the countercyclical role of state-owned banks than on nonfinancial
SOEs. Under certain conditions, for example, state-owned banks lend more than private banks
during a crisis (Coleman and Feler 2015), especially in developing countries. State-owned banks
reduce the procyclicality of credit in developing countries (Panizza 2021), and during severe
recessions, PPSOEs reduce the labor force at a slower rate than similar private firms (Lazzarini
and Musacchio 2018).

This paper provides evidence that sheds light on the fiscal risk of infrastructure SOEs when they
face common macroeconomic risks, such as a large drop in export prices. Until now, knowledge
about the fiscal risks of infrastructure SOEs and public-private partnerships (PPPs) comes from
studies of the aftermath of extremely large macroeconomic crises. For instance, Bova et al. (2019)
identify 21 bailouts of infrastructure SOEs with costs above 1 percent of GDP, most of them
occurring in the aftermath of a major financial crisis. The literature on fiscal risk warns that
government guarantees for PPPs generate significant risk, as contingent liabilities can be sizable
during negative macroeconomic shocks (Polackova Brixi 1998; Polackova Brixi and Mody 2002;
Polackova Brixi and Schick 2002).

This paper contributes to the literature by examining how negative shocks can have
heterogeneous effects in FSOEs and PPSOEs. The working hypotheses are that FSOEs usually
enjoy more guarantees than similar PPSOEs (Borisova and Megginson 2011; Borisova et al. 2015,
Wagner, Jara-Bertin, and Musacchio 2019) and arguably have easier access to government

3They use fixed effects and firm-specific time trends to disentangle improvements in efficiency
that may be caused by ownership changes in the short and long terms.

5
bailouts, which provides them with a softer budget constraint than PPSOEs, allowing them to
respond differently to negative shocks. For instance, FSOEs may have access to government
transfers or credit without having to show good performance for it (Cull and Xu 2003).

The contribution of the paper to the literature is therefore threefold. First, it uses a quasi-
experimental setting to study the behavior of SOEs treated with a negative macroeconomic shock
relative to nontreated SOEs. Second, it studies heterogeneous effects on FSOEs and PPSOEs.
Third, it uses detailed financials of infrastructure SOEs in multiple countries in the same time
period.

3 Data and Experimental Setting

3.1 Database
The World Bank Infrastructure SOEs Database was compiled as part of a broader research project
to analyze the fiscal costs and risks of infrastructure. The data come from financial statements
from SOE websites, government websites that include SOE financial statements, annual reports,
and other sources, such as the EMIS Intelligence database and stock exchange websites. The
database covers all SOEs operating infrastructure assets in the power (generation, transmission,
and distribution) and transportation (roads, railroads, and airlines and airports) sectors for 19
countries between 2000 and 2018. The countries were selected based on data availability and to
maximize sectoral coverage. They include the following:

• East Asia and Pacific: Indonesia, Solomon Islands
• Europe and Central Asia: Albania, Bulgaria, Croatia, Georgia, Kosovo, Romania,
Ukraine
• Latin America and Caribbean: Argentina, Brazil, Peru, Uruguay
• South Asia: Bhutan
• Sub-Saharan Africa: Burundi, Ethiopia, Ghana, Kenya, South Africa.

The database classifies an enterprise as an SOE if the state directly or indirectly owns more than
50 percent of its shares or the state is the ultimate controlling entity, through majority ownership
of common stock or any other mechanisms of control. This definition is in line with the European
Union’s definition of public undertakings in Commission Directive 2006/111/EC.

The database provides panel data on the financials of SOEs in the power and transport sectors at
the SOE/year level that are consistent over time and comparable across SOEs regardless of where
they operate. To ensure consistency and reliability, researchers collected the data using a
standardized accounting template that was populated using the information on financial
statements. To ensure that quantities like earnings before interest, taxes, depreciation, and
amortization (EBITDA) or operations subsidies in the database are comparable across SOEs and
6
years, they identified each item as defined by the template using the notes to the financial
statements rather than relying on the way such items are presented in the SOEs’ main financial
tables. Data reliability was further ensured through quality assurance checks by alternate
analysts and accounting experts.

The database provides a standardized representation of the income statement, balance sheet,
and cash flow statement of each SOE. It also contains supplementary items, including currency
risk, debt/loan analysis, maturity profiles of assets and liabilities, and SOE ownership structure.

We restrict the sample to SOEs in the power, airlines, and airports sectors, in order to maximize
comparability within sectors across countries. We focus on SOEs that depend heavily on demand
for their output to produce income, in order to measure the effect of a macroeconomic shock on
demand for their services. We exclude roads because of their dependence on budgetary
transfers, and railways, for which there is no variability across treatment and control groups.

3.2 Variables
Table 1 lists the main variables of interest, which include indicators of size (assets); financial
performance (EBITDA, returns on assets, CAPEX); and fiscal support (operational subsidies and
total fiscal injections). To compare across firms, we normalized the variables by total assets. A
mismatch could arise in the case of financial ratios that compare an income statement/cash flow
amount (a cumulative amount during a period of time) in the numerator to a balance sheet
amount like total assets (reported as the stock at the end of the period) in the denominator. To
avoid such a mismatch, we express total assets as the average of the end of the previous year
and the end of the observed year, in order to make it representative of each observed year. For
example, the return on average assets (ROAA) = net income (t)/average of total assets (t, t –1).
Because firms often receive operational and capital subsidies from the government, their net
income is not an accurate depiction of their performance. For that reason, we create alternative
measures of firm performance net of subsidies. Adjusted EBITDA over average assets uses EBITDA
net of operational subsidies as the numerator and average total assets as the denominator.
Adjusted ROAA uses total comprehensive income net of total subsidies as the numerator and
average total assets as the denominator. To see whether SOEs treated with the negative shock
receive more financial support from the government, we measure operational subsidies over
average assets and create a variable called fiscal injections over average assets to capture
different fiscal support instruments. The numerator of this variable is the sum of operational
subsides, government equity injections, changes in loans from government and SOEs, and
changes in deferred tax liabilities.

7
Table 1 Main variables used in the analysis
Variable Description

Capital expenditure (CAPEX) to average Ratio of CAPEX in period t, measured by the purchase of
assets (%) property, plant, and equipment, to the average of total
assets in t and t – 1

Earnings before interest, taxes, Ratio of EBITDA in period t to the average of total assets in
depreciation, and amortization (EBITDA) t and t – 1
to average assets (%)

Adjusted EBITDA to average assets (%) Ratio of EBITDA minus operational subsidies in period t to
the average of total assets in t and t – 1

Return on average assets (%) Ratio of annual net income in period t to average of total
assets in t and t – 1

Adjusted return on average assets (%) Ratio of annual net income minus total subsidies in period
t to average of total assets in t and t – 1

Operating subsidies to average assets (%) Ratio of operational subsidies in period t to average of
total assets in t and t – 1

Fiscal injections to average assets (%) Ratio of the sum of operational subsidies and government
equity injections in period t plus changes in loans from
government and SOEs and in deferred tax liabilities in
period t to average of total assets in t and t-1

Operating expenditures to average assets Ratio of operating expenditures t to average of total assets
(%) in t and t – 1

Operating income to average assets (%) Ratio of operating income in period t to average of total
assets in t and t – 1

Size (Log) Natural logarithm of total assets

Leverage (%) Ratio of debt to total assets in period t

3.3 Experimental Setting
We exploit the oil and gas price collapse of mid-2014 as an unanticipated negative shock to the
infrastructure SOEs in our database and examine its effects on performance. Brent prices
plummeted from $115 a barrel in June 2014 to $55 by the end of the year; after a short initial
recovery in early 2015, they declined throughout 2015, hitting a low of $27 in early 2016 before
bouncing back to more than $60 a barrel (figure 1, panel a). Gas prices show the same pattern
(figure 1, panel b).

8
Figure 1 Oil and natural gas prices, 2009–19
a. Europe Brent spot price for oil

140

120

100
Dollars per barrel

b. Henry Hub natural gas spot price

6
Dollars per million Btu

Source: Energy Information Administration.
Note: Oil prices are FOB (freight on board).

This oil shock affected countries differently based on their oil- and gas-exporting status and
energy reserves. To estimate the effects of a negative shock on the performance of SOEs, we take
advantage of the difference in energy exports and energy reserves across countries. The premise
is that countries that are energy rich will experience a negative shock to their economies with

9
the price collapse of these commodities vis-à-vis other countries. We categorize countries
represented in the sample as energy-rich and non-energy rich. We set a threshold based on both
exports and total reserves of crude oil and gas. Countries whose oil and gas exports are above
0.5 percent of total exports and have reserves of 1 million or more barrels of oil or 700 billion
cubic meters of gas are categorized as energy rich. Table 2 shows the resulting classification.

Table 2 SOEs in energy-rich and non-energy-rich countries in the sample

Treatment category/ SOE Sector Country Type
Energy-rich
OSHEE Power Albania PPSOE
INFRAERO Airlines and airports Brazil FSOE
CEMIG Power Brazil PPSOE
CELESC Power Brazil PPSOE
ECG Power Ghana FSOE
Angkasa Pura II Airlines and airports Indonesia FSOE
CORPAC Airlines and airports Peru FSOE
EGASA Power Peru FSOE
EGESUR Power Peru FSOE
Electro Puno Power Peru FSOE
ELSE Power Peru FSOE
San Gaban Power Peru FSOE
Bucharest Airport Airlines and airports Romania PPSOE
Transelectrica Power Romania PPSOE
Dnipro CHP Power Ukraine FSOE
Odessa CHP Power Ukraine FSOE
Cntr Energo Power Ukraine PPSOE
Kharkiv OE Power Ukraine PPSOE
Ternopil OE Power Ukraine PPSOE
Mykolaiv OE Power Ukraine PPSOE
Zaporizhia OE Power Ukraine PPSOE

Non-energy-rich
BPC Power Bhutan FSOE
Sofia Airport Airlines and airports Bulgaria FSOE
Plovdiv Airport Airlines and airports Bulgaria FSOE
GOA Airlines and airports Bulgaria FSOE
Maritsa Power Bulgaria FSOE
NEK Power Bulgaria FSOE
NPP Kozloduy Power Bulgaria FSOE
ESO Power Bulgaria FSOE
Zadar Airport Airlines and airports Croatia FSOE
Rijeka Airport Airlines and airports Croatia FSOE
Croatia Airlines Airlines and airports Croatia PPSOE
HE Power Croatia FSOE
KETRACO Power Kenya FSOE
Ken Gen Power Kenya PPSOE
KPLC Power Kenya PPSOE
KEK Power Kosovo FSOE
KOSTT Power Kosovo FSOE
10
Solomon Airlines Airlines and airports Solomon Islands FSOE
SIEA Power Solomon Islands FSOE
UTE Power Uruguay FSOE
Note: FSOE: Fully owned state-owned enterprise. PPSOE: Partially privatized state-owned enterprise.

Our criteria for defining energy-rich countries include both exports and reserves, so that our
sample of treated countries includes gas re-exporters from Eastern Europe (such as Ukraine). As
a drastic fall in the price of oil (and gas) negatively affects these re-exporters, we want to make
sure our definition of energy rich includes them in the treatment group. Leaving them in the non-
energy-rich group would pollute our control group, as they experience a negative
macroeconomic shock when the price of oil drops. 4

The experimental setting relies on the fact that the negative shock to the price of oil in 2014
severely affected the balance of payments of some major oil or gas exporters, creating a
recession—precisely the negative macroeconomic shock we want for the treatment group (table
3). Other countries that rely less heavily on oil or gas experienced milder to no effects. In fact,
there is no significant difference in real GDP growth rates or the current account deficit in the
control group before and after the shock, whereas the treated group had significantly slower GDP
growth after the shock (statistically significant at the 5 percent level). There is also heterogeneity
in the effect on oil importers. The shock benefited emerging economies that heavily subsidize
energy, such as India; the effect on advanced economies was minimal, in part because it occurred
during a period of high savings and low demand.

Table 3 Macroeconomic variables for the treatment and control groups
2010–13 2015–18
T-statistic
Standard Standard of mean
Variable Mean deviation Mean deviation difference
Control group (nine countries)
Current account balance to GDP
–0.28 6.03 0.64 2.24 –0.85
(% annual variation)
Real GDP growth rate (% annual) 4.67 3.58 3.65 3.08 1.28
Treatment group (nine countries)

4
Most of the re-exporting countries in Eastern Europe derive income from the gas pipelines that come
from the Russian Federation, in part because the gas is used domestically and in part because the re-
exported gas generates income for local firms and tax revenue for the government. We examine re-
exporting countries separately, finding that there was a severe shock to the current account and
aggregate demand in 2014–15, justifying the logic that their economies behave like those of oil and gas
exporters. The price of gas coming from Russia, which crosses the pipelines of many of the countries of
our sample in Eastern Europe, is determined using a moving average of the price of oil. See Gazprom’s
pricing policy at https://www.gazprom.com/about/marketing/europe/.
11
Current account balance to GDP
–0.38 2.04 0.25 1.74 –1.40
(% annual variation)
Real GDP growth rate (% annual) 4.67 3.43 2.72 3.36 2.43**
Source: International Monetary Fund, World Economic Outlook (April 2021), and own calculations.
Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

SOEs in the energy-rich countries are expected to be hurt by a fall in the price of oil or gas through
a macroeconomic demand shock. Countries below the threshold of exports and reserves are
categorized as non-energy-rich. SOEs in these countries are used as the control group, as we
would expect them not to be adversely affected by a fall in fuel prices. A concern is that the
declines in fuel prices not only negatively affect net exporters but also positively affect net
importers. If this is the case, the control group may not be a valid counterfactual. Table 3 shows
clear differences in the macroeconomic impact of the shock after the fall in oil prices. Therefore,
the differences in outcomes between the two groups should capture the effects of the negative
macroeconomic shock on the performance of SOEs. All countries in the sample are price-takers
in the international oil and gas markets.

Another concern about the way we create the treatment and control groups is that confounding
effects could arise from changes in the firms’ cost structures. For instance, the decline in the price
of oil could benefit firms that have high fuel costs (like airlines), attenuating the negative effects
that might be observed from reduced sales or fiscal tightening, especially if these types of firms
are more heavily concentrated in energy-rich countries. In order to avoid potential biases and
accurately identify the effects of interest, we initially restricted the sample to infrastructure firms
with low fuel costs as a percent of total expenses (firms below the 20th percentile of the
distribution of fuel costs as a share of total expenses). This approach yielded fewer than 15
observations. To enlarge our sample, we performed a matching exercise, using the coarsened
exact matching (CEM) algorithm (Iacus et al. 2012), to include firms above the 20th percentile of
the share of fuel costs, such that we are comparing firms within bins in which fuel costs to total
expenses are relatively similar.

We rely on the CEM methodology to create bins that are balanced across the treatment and
control groups using this ratio of fuel costs to expenses. We believe this method neutralizes the
possible positive effects stemming from the reductions in fuel costs for infrastructure SOEs. We
also prefer the CEM methodology because it minimizes the imbalances between the treated and
control groups without requiring a prespecified size of the matching solution (as propensity score
matching does). It operates by creating coarsened treated and control groups (as in stratification)
and then implementing exact matching with the coarsened data. The main advantages of the
method are its computational efficiency, especially in large datasets, and the fact that it

12
maximizes the number of matches achieved. We run the matching algorithm on fuel costs and
sectors, as we control for size in the regressions.

3.4 Descriptive Statistics
Both the mean EBITDA over assets and the adjusted EDITDA over assets are positive in most
years. The distributions are skewed to the left, with a few firms experiencing very large losses.
The median firm has positive EBITDA over assets and adjusted EDITDA over assets in all years.
The margins—income and cost—reflect the commercial nature of these SOEs. Operational
subsidies seem to be moderate. About a third of the SOEs in the sample received no subsidies,
and the median subsidy (when considering only firms that received subsidies) was 2.4 percent of
average assets. However, the distribution is skewed rightward, with large subsidies—above 100
percent of assets in some cases—and a mean of 11.5 percent of assets. 5

Figure 2 shows the trends for several indicators of financial performance in energy-rich countries.
Panels a and b show the mean EBITDA and the adjusted EBITDA over average assets. Both began
to increase in 2013, with some volatility, particularly in the case of PPSOEs, which showed weaker
performance than FSOEs. Capital expenditures (CAPEX) decreased sharply during and
immediately after the shock for the mean SOE, with a sharper decrease for FSOEs than for
PPSOEs, leading the former to invest less than PPSOEs as a share of average assets (panel c).
Operating costs as share of average assets of the mean FSOE are at least twice as large as the
mean PPSOEs (panel d).

5 One reason why these firms may perform better, and be less dependent on subsidies, than
the typical SOEs in developing countries is that the sample includes mostly SOEs from middle-
income countries and mostly from countries in which the World Bank has implemented
programs that include reforms to the main infrastructure sectors we study, which may have
improved the efficiency of the national companies.

13
Figure 2 Financial performance of SOEs in energy-rich economies, 2012–18
a. EBITDA b. Adjusted EBITDA
15 15

10 10

Mean as percent of assets
Mean as percent of assets

5 5

0
0

-5
-5

2012 2013 2014 2015 2016 2017 2018
2012 2013 2014 2015 2016 2017 2018

c. Capital expenditures d. Operating expenditures
7
140

6 120
Mean as percent of assets

Mean as percent of assets

100
5

4 60

40
3
2012 2013 2014 2015 2016 2017 2018 2012 2013 2014 2015 2016 2017 2018

All SOEs FSOEs PPSOEs

Source: World Bank Infrastructure SOEs Database and authors’ calculations.
Note: EBITDA: Earnings before interest, taxes, depreciation, and amortization. FSOE: Fully owned state-owned
enterprise. PPSOE: partially privatized state-owned enterprise.

Figure 3 shows the evolution of fiscal support for PPSOEs and FSOEs in energy-rich countries. The
average FSOE has received more operations subsidies as percent of assets than the average
PPSOEs since 2013 (panel a.). Operational subsidies over assets increased sharply in 2012–14 for
all SOEs, reaching almost 90 percent of average assets for the mean SOE, followed by a slight
decrease after 2014. In the case of overall fiscal injections, there is significant volatility in the
period of analysis, particularly for PPSOEs, with the mean fiscal injections to average assets for
all SOEs ranging between 2 and 6 (panel b.).

14
Figure 3 Fiscal transfers to SOEs in energy-rich economies, 2012–18
a. Operational subsidies b. Fiscal injections
10
1.0

8
0.8
Mean as percent of assets

Mean as percent of assets
6
0.6

4
0.4

0.2 2

0.0 0
2012 2013 2014 2015 2016 2017 2018 2012 2013 2014 2015 2016 2017 2018

All SOEs FSOEs PPSOEs

Source: World Bank Infrastructure SOEs Database and authors’ calculations
Note: Fiscal injections are the sum of operational subsides, equity injections, changes in loans from government
and other SOEs, and changes in deferred tax liabilities.

Table 4 presents descriptive statistics of the dependent variables for the treatment and control
groups for the pre-shock period. It shows that before 2013, on average, firms in the treatment
group (energy rich countries) perform similarly in terms of EBITDA, adjusted EBITDA over average
assets, ROAA, and adjusted ROAA as those in the control group. They also have similar CAPEX,
OPEX, operating income over average assets, and sizes (log of assets). SOEs in energy-rich
countries, however, have lower leverage than those in non-energy-rich countries. This difference
is not problematic, because we control for leverage in our regression analysis.

Table 4 Descriptive statistics and mean differences between SOEs in energy-rich and non-
energy rich countries, pre-shock

Non-energy-rich economies Energy-rich economies
T-statistic
Standard Standard of mean
Variable Observations Mean deviation Observations Mean deviation difference
EBITDA (% of
20 4.90 6.62 21 –1.90 30.09 0.99
average assets)
Adjusted EBITDA (%
20 –7.43 43.74 21 –2.54 29.90 –0.42
of average assets)
ROAA (%) 20 0.40 7.11 21 –6.85 29.63 1.07
Adjusted ROAA 15 –0.08 3.90 14 –10.93 35.35 1.18

15
CAPEX (% of average
19 8.50 10.01 21 6.34 5.07 0.87
assets)
Operations
subsidies (% of 20 12.33 42.80 21 0.64 1.34 1.25
average assets)
Fiscal injections (%
16 16.47 47.42 17 6.53 22.71 0.77
of average assets)
OPEX (% of average
20 59.48 73.10 21 91.57 87.10 –1.27
assets)
Operating Income
(% of average 20 52.03 62.12 21 88.72 76.66 –1.68
assets)
Size (log of total
20 13.07 2.55 21 13.58 1.47 –0.79
assets)
Leverage (debt-to-
20 12.31 11.44 21 6.14 10.37 1.81*
assets ratio, %)
Note: Sample includes SOEs in the power, airlines, and airports sectors.
*p < 0.1,**p < 0.05, and ***p < 0.01

Table 4 also reveals that SOEs in energy-rich and non-energy-rich countries have similar ratios of
subsidies to average assets and fiscal injections to average assets. Our setting thus reveals
whether governments react to negative shocks by increasing subsidies and fiscal injections
beyond their traditional levels.

4 Empirical Analysis

To identify the causal effect from the negative macroeconomic shock of the decline in oil and gas
prices in the world markets, we employ a difference-in-differences technique that takes
advantage of the fact that the price shock reduced aggregate demand in energy-rich countries
relative to non-energy-rich countries. We estimate the following equation:

������������������������������������ = ������������ + ������������1 ������������������������������������������������������������ + ������������2 ������������������������������������������������������������������������ ������������������������������������ℎ������������ + ������������3 ������������������������������������������������������������������������ ������������������������������������ℎ ������������ ∗ ������������������������������������������������ ������������ + ������������ ∗ ������������������������(������������������������������������������������������������������������������������������������ ) + ������������ ∗
������������������������������������������������������������������������������������������������������������������������ + ������������������������������������������������������������������������������������������������������������ + ������������������������������������������������������������������������������������������������������������ ∗ ������������������������������������������������������������ + ������������������������������������������������������������������������������������ + ������������������������������������������������oni ∗ Post t +
Income Groupi + Income Groupi ∗ Post t + πi + ϵit ,

where ������������ denotes the SOE, t denotes the year, and ������������������������������������ denote the outcomes of interest.
������������������������������������������������������������ ������������������������������������ℎ is a dummy variable equal to 1 if the firm is in a country classified as a net oil exporter
with high oil reserves before the shock in 2013. ������������������������������������������������������������ is a dummy variable equal to 1 if the
observation is from the post-shock period and 0 if the observation is from the pre-shock period.
The coefficient of interest is β3 , which identifies the average treatment effect on the treated. It
16
captures the differences in outcomes before and after the shock between the treated (T) and
control groups (C):

� 3 = (Δ
β ���� ����
Y)T − (Δ Y)C

We also include industry, region, and income group fixed effects, which we interact with the
������������������������������������t t variable to capture any sector, region, and income group–specific trends that could be
driving changes in the dependent variables. Although differences in firm characteristics between
treated and control groups are expected to be constant over time, we control for firm size using
the log of assets and the leverage ratio. We run this estimation using firm-level fixed effects, ������������������������ .

We focus on the following financial performance outcomes: return on assets, return on assets
net of subsidies (adjusted ROAA), EBITDA to average assets, EBITDA net of operational subsidies
(adjusted EBITDA) to average assets, and investment (CAPEX over average assets). We also look
at government support through (a) operational subsidies over average assets and (b) total fiscal
injections (the sum of operational subsidies, government equity injections, changes in loans from
the government and other SOEs, and changes in deferred tax liabilities) over average assets.

We expect that infrastructure SOEs would give up performance and CAPEX during a negative
shock while preserving employment, with a consequent drop in productivity. But it could also be
the case that these firms show more resilience, by taking advantage of their soft budget
constraint (by, for example, accessing more subsidies, deferring payment of taxes, or exercising
their privileged access to capital to continue paying CAPEX and employment costs). In this case,
we would expect to find that treated firms would have an increase in fiscal injections.

Initially, we look at the immediate reaction post shock, by comparing outcomes in 2013 and 2014.
However, as SOEs often hide or delay losses, we also examine whether the changes in SOE
policies happened in the years right after the shock. For that purpose, we run regressions in which
we assume the post-shock period extends to 2015 or 2016.

Doing so is important because SOEs have the capacity to borrow from other SOEs and from state-
owned banks and can ask governments to defer tax liabilities as a way to withstand a negative
shock or hide losses. As these tactics are often just ways of delaying the accounting loss, we check
to see whether the coefficient for the post-shock years is significant for treated SOEs. The
delaying or coping tactics can end up manifesting themselves in the accounting in other ways,
such as losses down the road, smaller investments, and greater government support.

Agency theories of SOE inefficiency predict that negative shocks should hurt financial
performance and productivity (Megginson and Neter 2001). But the soft budget constraints of
SOEs allow them to buffer some of the shock and continue to pay for government priorities such
as sustained CAPEX and employment costs (Kornai 2003). Given that the direction of causality
17
between a negative shock and our dependent variables is not clearly determined, we prefer to
let the data speak, to find the actual behavior of the infrastructure SOEs in our treated group
relative to the infrastructure SOEs in the control group.

We explore a few mechanisms through which the negative shock might affect revenues and
operating expenses. In treated countries, the negative macroeconomic shock could reduce
revenues for infrastructure SOEs through lower demand, with untreated firms expected to
remain unaffected. Regarding operating expenses, we do not have a clear prediction, because
we are matching firms by industry and fuel cost in the pre-shock period. There could, however,
be different reactions in terms because the shock depends on the pass-through of oil prices.

In a separate regression, we restrict our sample to include only FSOEs. We expect these firms to
have additional means of coping with the shock, in particular using their access to the
government and other SOEs as lenders of last resort, especially when governments ask SOEs to
undertake quasi-fiscal operations, as it is usually understood that in exchange the government
will be there to help when needed (Musacchio and Pineda Ayerbe 2019; Ter-Minassian 2019).

5 Findings

Table 5 displays the effects of the negative shock on all our outcomes of interest for the fixed-
effects estimation of our difference-in-differences model using the CEM sample. It does not show
evidence of a decline in financial performance in the aftermath of the shock for treated SOEs. It
does show an increase in subsidies over average assets in the first year after the shock and a
substantial increase in the overall measure of fiscal injections (i.e., through a range of
instruments beyond subsidies) two and three years after the shock.

18
Table 5 Baseline estimations of differences-in-differences with matching: Effects for treated SOEs one, two, and three years after
the oil shock of 2014

(1) (2) (3) (4) (5) (6) (7) (8) (9)
Item Adjusted Adjusted Operational Fiscal Operating
EBITDA/AA EBITDA/AA ROAA ROAA CAPEX/AA subsidies/AA injections/AA OPEX/AA Income/AA

2013–14 8.634 7.403 7.330 10.59 –1.047 1.231* 2.956 –4.447 2.951

Standard error (8.012) (8.139) (7.302) (13.20) (1.400) (0.619) (4.147) (4.656) (5.806)

Observations 82 82 82 58 80 82 66 82 82

Mean of dependent variable 3.362 –2.566 –1.951 –2.828 7.545 5.928 9.359 70.53 67.93

2013–15 17.38 13.33 17.29 24.98 –1.122 4.042 15.07*** –11.96 0.510

Standard error (18.91) (19.57) (18.21) (25.78) (1.530) (2.339) (3.482) (11.02) (13.35)

Observations 80 80 80 58 78 80 62 80 80

Mean of dependent variable 3.835 –2.036 –1.068 –2.129 7.423 5.872 9.072 72.65 70.77

2013–16 11.16 –2.333 12.75 14.62 –2.572 13.50 15.89** 9.424 6.775

Standard error (13.72) (16.16) (13.78) (20.24) (3.411) (8.972) (6.464) (10.98) (8.791)

Observations 80 80 80 58 78 80 60 80 80

Mean of dependent variable 4.513 –0.442 –0.514 –1.173 7.722 4.955 10.03 71.57 71.24

Note: EBITDA: Earnings before interest, taxes, depreciation, and amortization. AA: Average assets. ROAA: Return on average assets. CAPEX: Capital
expenditures. OPEX: Operating expenses. Robust standard errors are in parentheses. Regressions include a constant, a control for ln(assets), leverage, and
fixed effects for industry and World Bank region and income groups. Matching is done using industry and fuel cost expense over total expenses.
* p < 0.1, ** p < 0.05, *** p < 0.01.

19
A possible problem of mixing PPSOEs and FSOEs in the same sample is that the PPSOEs may
behave more like private firms than FSOEs and could be receiving less support from the
government to cope with the shock. PPSOEs may be able to adjust more rapidly than FSOEs to a
shock, perhaps reducing the labor force at a faster pace and thus not showing a decline in return
on assets. They may also have less access to subsidies or cheap financing. Finally, one would
expect that both types of firms would reduce CAPEX, causing us to pick up a negative and
significant coefficient in some of the specifications.

To avoid some of these confounding effects, we exclude PPSOEs from the sample and focus
exclusively on FSOEs (table 6). The results seem to be more robust statistically and point to the
following effects. First, there is a clear decline in profitability. Column 1 shows no change on
EBITDA over assets after the oil shock. However, the result can be masked by operational
subsidies from the government. Column 2 shows a significant decrease in adjusted EBITDA over
average assets one and three years after the shock. In the first year, adjusted EBITDA over
average assets decreased by 3.9 percentage points, with a mean EBITDA over average assets of
–4.9 percent. Three years after the shock the impact is even stronger, with a decrease of 15.7
percentage points in EBITDA over average assets and a mean of –2.2 percent for the sample.
Adjusted EBITDA over average assets is negative but insignificant two years after the shock.
Columns 4 and 5 show that adjusted ROAA decreased two years after the shock.

Second, we would expect to find a reduction in capex in both PPSOEs and FSOEs. Yet none of the
coefficients in column 5 in Table 5 is significant. Looking exclusively at the negative shock for
FSOEs in column 5 of Table 6, however, reveals a significant difference in the CAPEX of the treated
and control firms two and three years after the shock. CAPEX as a percent of average assets
decreased by 3.2 percentage points two years after the shock, which is close to half the mean of
the sample. The decrease is even larger three years after the shock, with a coefficient of –7.2 and
a mean of 8.1. These findings indicate that the negative shock led to a progressive reduction in
CAPEX of FSOEs in treated countries, which likely led to a decrease in productivity and operational
performance over several years after the shock.

20
Table 6 Matching estimates of the difference-in-differences for the fully owned SOE sample

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Adjusted Adjusted Operational Fiscal Operating
Item EBITDA/AA EBITDA/AA ROAA ROAA CAPEX/AA subsidies/AA injections/AA OPEX/AA Income/AA

2013–14 –2.351 –3.897** –2.718 –3.873 –0.807 1.546* 7.962** 0.528 –3.680

Standard error (1.381) (1.487) (1.811) (2.728) (2.344) (0.763) (2.624) (7.646) (7.388)

Number of observations 56 56 56 42 54 56 46 56 56

Mean of dependent variable 2.829 –4.874 –1.420 –2.200 7.787 7.704 11.34 52.10 47.29

2013–15 0.995 –3.240 0.280 –4.585* –3.231** 4.235 15.27*** –11.80 –16.02

Standard error (4.245) (5.054) (2.910) (2.283) (1.236) (2.400) (3.253) (12.19) (9.490)

Number of observations 56 56 56 42 54 56 46 56 56

Mean of dependent variable 2.687 –4.707 –1.425 –2.436 7.857 7.395 10.77 51.74 47.38

2013–16 –2.913 –15.68* –2.595 –7.706 –7.158** 12.77 17.44* 8.923 –7.288

Standard error (4.144) (8.547) (3.950) (5.836) (2.496) (8.551) (8.601) (17.22) (13.17)

Number of observations 56 56 56 42 54 56 44 56 56

Mean of dependent variable 3.981 –2.179 –0.373 –0.746 8.079 6.160 12.37 49.06 47.09

Note: Note: EBITDA: Earnings before interest, taxes, depreciation, and amortization. AA: Average assets. ROAA: Return on average assets. CAPEX: Capital expenditures. OPEX:
Operating expenses. Robust standard errors are in parentheses. Regressions include a constant, a control for ln(assets), leverage, fixed effects for industry, and World Bank
region and income groups. Matching is done using industry and fuel cost expense over total expenses.
* p < 0.1, ** p < 0.05, *** p < 0.01

21
We explore the effect of the oil price shock on operational subsidies as a percentage of average
assets and fiscal injections as a percentage of average assets. Operational subsidies as a
percentage of average assets increase by 1.55 percentage points immediately after the shock,
with a mean of 7.7 (column 6). Governments used different instruments to support SOEs after
the shock, with support increasing over time (column 7). Fiscal injections as a percentage of
average assets increased by 8.0–17.4 percentage points one, two, and three years after the
shock, with the increases in the last two years 41 percent above the respective sample means for
each of the regressions. The fiscal injections needed by the treated firms were thus large, even
three years after the shock.

We believe that the main mechanism making treated firms underperform control firms is the
decrease in demand in treated countries as a result of the macroeconomic shock. But the
confounding effect of a decrease in the cost of fuel inputs could be reducing operating expenses
in both treatment and control firms. One possible outcome is that operating expenses may fall
more in control firms and do not fall significantly in energy exporters, because a national oil or
gas company is key for fiscal purposes and SOEs that buy energy inputs do not get the
passthrough effect of the fall in energy prices. We checked to see whether part of our findings
are explained by the differentials in the fall of operating expenses to average assets in column 8
of Table 6. As expected, the coefficient is not significant, signaling that there are no differences
in the passthrough of oil prices between the treated and control groups.

6 Conclusions

Hit with negative shocks, infrastructure SOEs incur losses and require government support, in the
form of fiscal transfers. In times of crisis, when governments want to use countercyclical fiscal
policy to stimulate the economy, infrastructure SOEs do not help the government by using their
buffer reserves to withstand the shock. On the contrary, they seem to operate with little financial
slack, such that they need significant fiscal resources to survive following a large macroeconomic
shock. Better-performing SOEs would help cushion negative shocks by not acting as a fiscal drag.

Recent discussion in the infrastructure literature suggests subsidizing energy as an efficient
countercyclical measure during crises. Our results suggest that SOEs would not be able to help
buffer a crisis and that aid would end up being funded by the government. In our findings, the
bill comes back to the government in year 1 of the shock.

The negative effects of the macroeconomic shock we observe are stronger for FSOEs, which
perform worse than comparable PPSOEs. We find evidence of the soft budget constraint for both
FSOEs and PPSOEs, as we document significant increases in subsidies and other fiscal injections
to both types of SOEs in the immediate aftermath of the shock.

22
Our evidence suggests that SOEs can amplify negative macroeconomic shocks and increase fiscal
risk. This finding is particularly important because the effect happens precisely when
governments are under pressure from the fall in total tax revenues. The losses of FSOEs after the
negative shock are significant. Moreover, these firms require economically significant levels of
support to continue financing operations at the very time that belt tightening is important. FSOEs
reduce CAPEX significantly, even after receiving substantial fiscal support, which can lead to
lasting effects of the negative macroeconomic shock.

Our findings provide more evidence for the debate over the advantages and disadvantages of
government ownership of infrastructure. Much of the literature on fiscal risks and negative
shocks has focused on the incompleteness of contracts in PPPs. The Asian financial crisis showed
that negative shocks usually led to bailouts, extensions of contracts, and debt guarantees for
PPPs. 6 This discussion rarely extended to the fiscal costs of having to come to the aid of
infrastructure SOEs during a crisis, at least not with systematic empirical evidence of the fiscal
costs of SOEs and usually not using experimental techniques. This paper shows that SOEs also
suffer from negative shocks. Their effects shocks are persistent, because of the difficulties SOEs
have cutting expenses. We also document how a negative shock increases fiscal risk, by
increasing contingent liabilities for the government, which has to send subsidies immediately, in
amounts that are large in economic terms.

The debate on the fiscal costs of negative shocks in infrastructure needs to be expanded to
include the costs that SOEs impose on governments versus the costs that PPP guarantees impose
on governments during negative shocks. 7 We hope that our paper represents a step in that
direction.

References

Andrés, L. A., V. Foster, and J. L. Guasch. 2006. “The Impact of Privatization on the Performance
of the Infrastructure Sector: the Case of Electricity Distribution in Latin American
Countries.” Policy Research Working Paper 3936, World Bank, Washington, DC.
Borisova, G., V. Fotak, K. Holland, and W. L. Megginson. 2015. “Government Ownership and the
Cost of Debt: Evidence from Government Investments in Publicly Traded Firms.” Journal of
Financial Economics 118(1): 168–191.
Borisova, G., and W. L. Megginson. 2011. “Does Government Ownership Affect the Cost of
Debt? Evidence from Privatization.” Review of Financial Studies 24(8): 2693–2737.
Bova, M. E., M. Ruiz-Arranz, M. F. G. Toscani, and H. E. Ture. 2016. The Fiscal Costs of
Contingent Liabilities: A New Dataset. International Monetary Fund, Washington, DC.

6
See, for instance, Hemming and staff (2006); Lewis and Mody (1998); and Mody (2000).
7
For some initial discussions of the topic, see Engel, Fischer, and Galetovic (2010, 2019).
23
——. 2019. “The Impact of Contingent Liability Realizations on Public Finances.” International
Tax and Public Finance 26(2): 381–417.
Coleman, N., and L. Feler. 2015. “Bank Ownership, Lending, and Local Economic Performance
During the 2008–2009 Financial Crisis.” Journal of Monetary Economics 71: 50–66.
Cull, R., and L. C. Xu. 2003. “Who Gets Credit? The Behavior of Bureaucrats and State Banks in
Allocating Credit to Chinese State-Owned Enterprises.” Journal of Development Economics
71(2): 533–559.
Eberhard, A., K. Gratwick, E. Morella, and P. Antmann. 2016. Independent Power Projects in
Sub-Saharan Africa: Lessons from Five Key Countries. Washington, DC: World Bank.
Engel, Eduardo, Ronald D. Fischer, and Alexander Galetovic. 2010. “The Economics of
Infrastructure Finance: Public-Private Partnerships versus Public Provision.” EIB Papers
15(1): 40–69.
———. 2019. “Soft Budgets and Endogenous Renegotiations in Transport PPPs: An Equilibrium
Analysis.” Economics of Transportation 17(C): 40–50.
Estache, A., A. Gomez-Lobo, and D. Leipziger. 2001. “Utilities Privatization and the Poor:
Lessons and Evidence from Latin America.” World Development 29(7): 1179–1198.
Foster, V., S. Witte, S. G. Banerjee and A. Moreno. 2017. Charting the Diffusion of Power Sector
Reforms across the Developing World. Washington, DC: World Bank.
Galiani, S., P. Gertler, and E. Schargrodsky. 2005. “Water for Life: the Impact of the Privatization
of Water Services on Child Mortality.” Journal of Political Economy 113(1): 83 and 120.
Hemming, Richard, and staff 2006. Public-Private Partnerships, Government Guarantees, and
Fiscal Risk. International Monetary Fund, Washington, DC.
Herrera Dappe, M., A. Musacchio, C. Pan, Y. V. Semikolenova, B. Turkgulu and J. Barboza. 2022.
“Smoke and Mirrors: Infrastructure SOEs and Fiscal Risks.” Policy Research Working Paper
9970, World Bank, Washington, DC.
Iacus, S. M., G. King and G. Porro. 2012. “Causal Inference without Balance Checking: Coarsened
Exact Matching.” Political Analysis 20(1): 1 and 24.
Kornai Kornai, Janos. "Resource-constrained versus demand-constrained
systems." Econometrica: Journal of the Econometric Society (1979): 801-819.
Kornai, J., E. Maskin, and G. Roland. 2003. “Understanding the Soft Budget Constraint.” Journal
of Economic Literature 41(4): 1095–1136.
Lazzarini, S. G., and A. Musacchio. 2018. “State Ownership Reinvented? Explaining Performance
Differences between State‐Owned and Private Firms.” Corporate Governance: An
International Review.
Lewis, Christopher M., and Ashoka Mody. 1998. "The Management of Contingent Liabilities: A
Risk Management Framework for National Governments." In Dealing with Public Risk In
Private Infrastructure, edited by T. Irwin, M. Klen, G. E. Perry, and M. Thobani, 131–53.
———. 1998. “Contingent Liabilities for Infrastructure Projects: Implementing a Risk
Management Framework for Governments.” Viewpoint.
24
Megginson, William L., and Jeffry M. Netter. "From State to Market: A Survey of Empirical
Studies on Privatization." Journal of Economic Literature 39(2): 321–389.
Mody, Ashoka. 2000. Contingent Liabilities in infrastructure: Lessons of the East Asian Crisis.
World Bank, Washington, DC.
Musacchio, A., and S. G. Lazzarini. 2014. Reinventing State Capitalism: Leviathan in Business,
Brazil and Beyond. Cambridge, MA: Harvard University Press.
Musacchio, A., and E. Pineda Ayerbe, eds. 2019. Fixing State-Owned Enterprises: New Policy
Solutions to Old Problems. Washington, DC: Inter-American Development Bank.
Nellis, J. 2005. The Evolution of Enterprise Reform in Africa: From State-Owned Enterprises to
Private Participation in Infrastructure—and Back? Energy Sector Management Assistance
Programme (ESMAP) Technical Paper 84, World Bank, Washington, DC.
Panizza, U. 2021. “State-Owned Commercial Banks.” Working Paper 09-2021, Graduate
Institute of International and Development Studies (IHEID), Geneva. .
Polackova Brixi, H. 1998. “Contingent Liabilities: A Threat to Fiscal Stability.” PREM Note 9,
World Bank, Washington, DC.
Polackova Brixi, H., and A. Mody. 2002. “Dealing with Government Fiscal Risk: An Overview.” In
Government at Risk: Contingent Liabilities and Fiscal Risk, ed. H. Polackova Brixi and A.
Schick, 21–58. Washington, DC: World Bank.
Polackova Brixi, H., and A. Schick, eds. 2002. Government at Risk: Contingent Liabilities and
Fiscal Risk. Washington, DC: World Bank.
Ter-Minassian, T. 2019. “Identifying and Mitigating Fiscal Risks from State-Owned Enterprises.
SOEs.” In Fixing State-Owned Enterprises: New Policy Solutions to Old Problems, ed. A.
Musacchio and E. Pineda Ayerbe, 49–72. Washington, DC: Inter-American Development
Bank.
Wagner, R., M. Jara-Bertin, and A. Musacchio. 2019. “Do Bond Markets Discipline SOEs?” In
Fixing State-Owned Enterprises: New Policy Solutions to Old Problems, ed. John Huber,
Mauricio Jara, Hyungon Kim, Teresa Ter-Minassian, and Rodrigo Wagner, 73–89,
Washington, DC: Inter-American Development Bank.