working paper series department of economics alfred … · department of economics alfred lerner...
TRANSCRIPT
Working Paper Series∗
Department of Economics
Alfred Lerner College of Business & EconomicsUniversity of Delaware
Working Paper No. 2003-05
An Assessment of the Effectiveness of
International Financial Intervention†
James L. Butkiewicz and Halit Yanikkaya
March 2003
∗http://www.be.udel.edu/economics/workingpaper.htm† c© 2003 by author(s). All rights reserved.
AN ASSESSMENT OF THE EFFECTIVENESS OF INTERNATIONAL FINANCIAL INTERVENTION
James L. Butkiewicz
Department of Economics University of Delaware
Newark, DE 19716 [email protected]
Halit Yanikkaya
College of Business and Administrative Sciences Department of Economics
Celal Bayar University Manisa, Turkey 45030
JEL Classifications: F34, O40
Key Words: International lending, development aid, empirical growth model
Abstract
The two global international financial institutions, the International Monetary Fund and the World Bank, frequently, and often repeatedly, extend loans to developing nations. Recently, these loans have been blamed for generating adverse economic outcomes. An empirical growth model, which controls for the other determinants of growth, is used to assess the growth impact of Fund and Bank loan programs. The primary focus is on the heavily criticized IMF lending programs. Another unique feature of this study is the use of the value of lending programs rather that the number of programs. The estimates indicate that Bank lending stimulates growth in some cases, primarily by increasing public investment. Fund lending is either neutral or detrimental to growth. The channel for this effect is a negative impact of Fund lending on private investment.
1
I. Introduction
In recent years the lending practices of the International Monetary Fund (IMF or
Fund) and the International Bank for Reconstruction and Development (World Bank or
Bank) have been subjected to ever-increasing public scrutiny, and frequent criticism.
The IMF’s lending practices, in particular, are regularly criticized as “anti-growth” and
“anti-poor”. Indeed, IMF lending, especially in “crisis” situations, typically requires
domestic governments to restrain domestic demand. However, there is frequent
recidivism among the borrowing nations, and the IMF-imposed conditions may be
necessary to address the fundamental problems existent in recidivist nations.
World Bank lending has been less criticized and less studied than IMF lending.
As Kruger (1998) argues, the IMF frequently lends to countries in the midst of a crisis
when the borrowing nations suffer from external and/or internal imbalances. This is not
the case for World Bank lending, which is typically directed toward long-term
development projects. Thus, while the IMF is criticized as too harsh, the World Bank is
often criticized as too soft in its lending practices.
This paper presents the results of an investigation of the growth effects of both
IMF and World Bank lending, within a common empirical framework. A unique feature of
this study is an analysis of the impact of fundamental economic and political conditions
on the efficacy of IMF and World Bank lending programs. The next section discusses
the relevant literature. The model and data are presented in section III. The results are
presented in section IV, followed by a discussion and a concluding section.
II. Literature Review
The IMF was established as a financial institution. Its basic functions, as
specified in the first Article of Agreement, include promoting world trade, international
financial stability, and the macroeconomic stability and growth of member countries. In
2
order to achieve these objectives, the IMF focuses on its core responsibilities: monetary,
fiscal, and exchange rate policies, and their associated institutional and structural
frameworks1. The IMF performs its functions in three ways: surveillance, provision of
technical assistance to members, and lending in three modes (crisis prevention, balance
of payments adjustment, and poverty reduction).
The World Bank has always viewed its mission as developmental. Thus, the
Bank’s core mandate is to help countries reduce poverty, particularly by focusing on the
institutional, structural and social dimensions of development. Birdsall and Londono
(1997) argue that “financial gap” models motivated World Bank lending throughout the
1970s. In the 1980s, the operational work of Bank shifted to adjustment issues. And in
the 1990s, the Bank once again made its primary focus poverty reduction, with a “new”
strategy emphasizing the acceleration of economic growth, the provision of basic social
services to the poor, and the creation of social safety nets.
Although the division of labor between the IMF and the World Bank was clear
when these institutions were established, over time, considerable overlap has developed
as the IMF and the Bank have focused on assisting developing countries and transition
economies. After the collapse of the Bretton Woods System, and especially after the
1980s debt crisis, the overlap between roles of these International Financial Institutions
(IFIs) has increased considerably. The IMF began developmental assistance in 1986
with the creation of the Structural Adjustment Facility (SAF), which evolved into the
Enhanced Structural Adjustment Facility (ESAF), and more recently became the Poverty
Reduction and Growth Facility (PRGF). These facilities brought the IMF into what had
traditionally been the purview of the Bank. Recently, in a joint statement the IMF and the
World Bank (IMF and World Bank 2002, 2) announced “the purpose of our institutions is
1 See Krueger (1998) and Bordo and James (2000) for extensive review the of the IMF’s role in the past and present.
3
to help all our member countries develop their human potential and productive
resources, thereby building the foundations for sustainable economic growth.”
Many recent studies (James 1998, Feldstein 1998, Goldstein 2000, Bordo and
James 2000, International Financial Institution Advisory Committee 2000) strongly
criticize the IMF for straying from its core competence of macroeconomic and exchange
rate policies into a host of structural policy areas such as corporate governance, trade
policy, privatization, poverty reduction, and environmental management, areas in which
the Fund does not have necessary expertise and staff resources to make timely and
sound policy recommendations2. For example, Feldstein (1998, 20) argues “(T)he IMF’s
recent emphasis on imposing major structural and institutional reforms as opposed to
focusing on balance-of-payments adjustments will have adverse consequences in the
both the short term and the more distant future.”
As the rise of structural adjustment components of IMF lending encroaches on
the division of labor between the Fund and the Bank, a case has been made for merging
the two institutions. However, the Bretton Woods Commission (1994) rejected this
recommendation, citing their differing objectives. Both the Fund and the Bank recognize
the importance of the potential overlap, and state their intention of jointly increasing
cooperation (Fischer 1997, IMF and World Bank 2002, Mussa and Savastano, 2000).
Fischer (2000a) emphasizes the need for reforming the IMF by sharpening the focus of
its activities to macroeconomic policies and the accompanying structural areas:
monetary, fiscal, and exchange rate policies, and the banking and financial sectors.
However, Easterly (2001) emphasizes that in practice neither institution proceeds with
an adjustment loan unless the other is satisfied with progress in "its" area of
responsibility. For example, Feinberg (1988) reports that only 3 of 17 of the Bank’s
2 Some studies (such as, Goldstein 2000, Stewart and FitzGerald 1996, Thacker 1999) put the blame on the powerful members of IMF for the rise in the scope of IMF programs.
4
sector loans signed during fiscal years 1979-1985 were made to countries not engaged
in Fund programs.
The Fund primarily focuses on macroeconomic policies in developing countries
(including exchange rates, trade regimes, and financial markets), while the Bank takes a
long-term perspective, analyzing real variables and directing its focus toward increasing
supply-side efficiency and domestic investment. However, both Fund and Bank
programs have common characteristics, including currency devaluations and market
determined exchange rate adjustments, the adoption of anti-inflationary and demand
restraining measures, the restoration or construction of market mechanisms including
freeing controlled prices and interest rates, reducing trade barriers, and the privatization
of state enterprises.
Although IMF programs have many different objectives, certain characteristics of
Fund programs are linked to its mandate to confront external payments problems. This
is the so-called “three-pronged approach” (Knight and Santanella 1997, Mussa and
Savastano 2000, Krueger 2000). The first component of this approach involves securing
sustainable external financing. At the onset of a crisis, external financing for borrowing
countries frequently evaporates. The IMF assists countries suffering balance-of-
payments problems with short-term loans through its Stand-by Arrangements (SBA) and
medium-term loans through its Extended Fund Facility (EFF). Another common element
of Fund programs is the adoption of demand restraining measures consistent with
available financing. These first two measures comprise the macroeconomic policies
intended to restore sustainable balance between aggregate expenditure and income in
program country by tightening fiscal and monetary policies. The last component of IMF
programs requires structural reforms intended to promote growth and adjustment in the
medium and long term. These policies aim to reduce government related distortions and
5
structural problems resulting in an inefficient allocation of resources in the economy and
hindering growth3.
Another goal of both IMF and Bank programs is the mobilization of supporting
private capital flows4. As argued by Rodrik (1995), private capital flows follow
multilateral flows only if the private sector values the information provided by IFIs. Bird
(1996b) and Rowlands (1996) review previous studies analysing the “catalyst” role of
IMF lending and concluded that the catalytic effect of IMF lending is, at best, weak. Bird
(1996b, 489) actually concludes that “the Fund’s seal of approval does not seem to carry
a very high market value.” More importantly, Rodrik (1995) shows that multilateral
lending has actually followed the other sources of capital. While IMF loans to member
countries have followed commercial bank loans, non-IMF multilateral lending has
followed bilateral transfers. Rodrik and Bird eventually claim that to the extent that
multilateral lending follows private flows, a major concern is that the multilateral
institutions end up bailing out private creditors. Although IFIs have claimed the seniority
of their monies, any multilateral lending that helps governments’ service their debt is a
form of subsidy to private capitalists. However, upon further disaggregating IMF lending,
Rowlands (1996) finds that SBA and EFF loans do induce official lending to developing
countries while private lending is neutral. It is possible that SAF/ESAF lending may
actually discourage private lending.
During the 1970s and early 1980s, Fund programs were frequently criticized in
many studies (Bird 1996b, Bordo and James 2000, and Bird 2001) and by developing
countries as being too demand oriented and too short run, not paying enough attention
3 For an extensive overview of IMF structural programs, see Goldstein 2000. He (76) concludes “my reading of the record is that on structural policies the Fund has bitten off more -- in both scope and detail -- than either it or its member countries can chew.” 4 Bird (2001) evaluates counterarguments that make the theoretical basis for catalysis ambiguous. Moreover, Rodrik (1995) argues that the insistence of multilateral lenders that their claims be senior to private claims undercuts the signaling value of their exposure.
6
to economic growth and supply-side reforms. The IMF has responded to these critics by
substantially increasing the number of structural conditions in its programs since the late
1980s and developing long-term packages in which more attention given to growth and
development issues (Knight and Santanella 1997). At the same time, Bordo and James
(2000) argue that the IMF has responded to the fundamental changes in global
environment, including the breakdown of the par value system, the increased mobility of
capital, financial deregulation, and the collapse of Soviet bloc by expanding the scope
and detail of its activities. Goldstein (2000) reports a consensus in the literature
analyzing Fund structural policy conditionality. Over the past fifteen years, conditionality
in IMF programs has increased markedly, at an increasing rate. Furthermore, since the
mid-1980s, IMF as well as Bank programs include as an objective economic growth, and
recently “high-quality growth.” For example, in a number of speeches (1994, 2000)
former IMF Managing Director, Michel Camdessus emphasizes that the IMF’s primary
goal is not only growth but “high quality growth.” Note that even he characterized this
objective as “ambitious” but argues (1994, 2) “it is the only way that the world's economic
and social challenges can be met.”
Recent critics of the Fund argue that the rise in the scope of IMF programs more
likely reduces program effectiveness, further increasing criticism of the Fund. For
example, Mussa and Savastano (2000) claim that the objectives of high output growth
and alleviating poverty are not explicitly among the Fund’s core responsibilities. They
also point out that much of the criticism of the IMF might be due to a disjunction between
its core responsibilities and broader objectives such as a high rate of economic growth,
low inflation, and the alleviation of poverty. These latter goals are medium- to long-term
objectives. Furthermore, Goldstein (2000) claims that the Fund’s charter does not
provide a basis for a broad agenda aimed at high (quality) growth. He (2000, 77) argues
7
that the “IMF mission chiefs have considerable knowledge and experience in
macroeconomic and financial policies but not in structural policy areas beyond this core
competence. Efforts to include in Fund conditionality everything but the kitchen sink
under the a loosely defined agenda of pursuing ‘high quality’ growth have taken the
Fund too far from its comparative advantage and have elicited legitimate charges of
‘mission creep’.” Regarding the argument that the IMF should not be in the poverty
business; poverty is primarily the business of the World Bank and the regional banks,
Fischer (2000b, 4) replies that “the poorest members of the world community belong to
the IMF; they have macroeconomic problems; they have a right like every other member
to access the facilities of the IMF.”
The literature on IMF programs can be divided into two parts. One strand studies
the determinants of IMF lending5; while the other strand examines the effects of IMF
lending on borrowing countries. Since IMF programs involve a number of
macroeconomic objectives including lowering inflation; restoring balance of payments
equilibrium; attaining a certain level of international reserves; resumption or expansion of
private capital inflows, exports, and investment; and a sustained high rate of growth,
5 On the determinants of IMF lending, see Bird 1996a, Knight and Santanella 1997, Thacker 1999, Przeworski and Vreeland 2000, Barro and Lee, 2002. For example, Knight and Santanella (1997) find that countries with higher external debt, depreciated currencies, low rates of investment, lower levels of GDP growth and per capita GDP are more likely to seek the Fund’s support. At the same time, they find that policy measures to improve fiscal revenues, to reduce government expenditures, to tighten domestic credits, and to adjust the exchange rate are significant factors that increase the probability of Fund approval. Thacker (1999) and Barro and Lee (2002) focus on the political determinants of IMF lending and find that political variables are important determinants of IMF lending. At the same time, Przeworski and Vreeland (2000) conclude that none of the economic variables matter for the government’s decision to remain under agreement with the IMF. The only variables that have significant effects on the government’s decision to continue participation in an IMF agreement are the sum of past years under agreements for a country and number of other countries currently under IMF agreements. Regarding the decision of the Fund to retain countries under its programs, only the balance of payments has a significant effect. Note that a number of studies (Boone 1996, Burnside and Dollar 2000, and Alesina and Dollar 2000) investigating the determinants of official aid find that aid flows are, to a large extent, explained by political factors such as colonial links, alliances, strategic interests, etc., but the recipient’s economic policies and political institutions are less important.
8
there are a number of different criteria by which IMF programs can be judged. Mussa
and Savastano (2000) argue that since encouraging growth and reducing inflation are
not among the Fund’s core objectives, it might be unfair to evaluate these programs by
looking at their effects on growth and inflation. Moreover, as Fischer (1997)
emphasizes, when evaluating the effects of IMF programs, it is important to distinguish
two issues. The first is whether IMF programs are appropriate prescriptions to
encourage growth if implemented. The second question is whether the agreed-upon
program is implemented as designed.
As reviewed in Haque and Khan (1998), Bordo and Schwartz (2000), and Bird
(2001), most studies find that IMF lending improves the balance of payments and current
account. This finding is not surprising given the fact that the primary focus of the IMF is
to correct maladjustments in its member countries’ balance of payments without
resorting to measures destructive of national or international prosperity. However, the
reported effect of IMF lending on inflation is negative but weak. With regard to the
growth effects of these programs, the literature is far from consensus. Haque and Khan
(1998) find negative effects on growth over the short-run, probably due to the demand
restraining nature of IMF conditions, while over the long-run growth increases.
Alternatively, Bird (2001,1861) concludes that “Fund programs seem to have a negative
effect on investment and possibly economic growth, often do not enable countries to
graduate from a reliance on IMF resources, more often than not remain uncompleted,
and do not catalyze external finance from other sources.”
Empirical studies analyzing the effects of IMF programs have employed a variety
of methodologies. As discussed in many studies (Haque and Khan 1998, Goldstein
2000, Bird 2001, Barro and Lee 2002), some of these methodologies have important
shortcomings. “Before-after” comparisons are not reliable because they implicitly
9
assume no change occurs except for an IMF program between the two periods being
compared. Comparison of program outcomes to targets may not be useful because
there are pronounced problems with the implementation of IMF programs and program
targets may be set too ambitiously or not ambitiously enough. Simulations using
economic models can tell us something about the effect of Fund-type policies but not
about the effects of actual Fund programs. Comparisons of outcomes for program and
nonprogram countries suffer the limitation that the two groups of countries can differ
systematically in ways that matter for economic performance. Haque and Khan (1998)
conclude that studies using before-after and with-without approaches yield less
favorable results than later studies using the General Evaluation Estimator (GEE)
methodology and simulation techniques, and that more confidence can be placed in the
results produced by the recent studies. However, Dicks-Mireaux et al. (2000) claim that
there are considerable doubts about the validity of applications of the GEE that do not
rigorously test the underlying assumptions. Given the fact that the GEE framework has
many restrictive assumptions that are necessary to define the counterfactual and to
specify in a simple framework the main determinants of important endogenous
macroeconomic variables, a major shortcoming of most applications of the GEE is the
focus on the bottom line — the estimates of the effects of IMF support — with little or no
evaluation of the validity of the underlying model. At the same time, both Hutchison
(2001) and Barro and Lee (2002) argue that Heckman’s (1979) inverse Mills ratio (IMR)
approach does not adequately control for selection bias. Besides the methodological
problems discussed in previous literature, another possible limitation, resulting from the
short period of observation used in most of these studies, is the likelihood that the
reported results reflect business cycle effects rather than the growth effects of these
programs, or some combination of these two effects.
10
Goldstein and Montiel (1986) show that IMF programs have negative effects on
growth. Conway (1994) concludes that IMF programs have only favorable growth and
investment implications in the long run. Recent studies of the macroeconomic effects of
IMF lending often report adverse growth effects. A large number of studies (Barro 1998,
Stiglitz 2000, Bordo and Schwartz 2000, Przeworski and Vreeland 2000, Hutchison
2001, and Barro and Lee 2002) claim that IMF programs do more harm than good for the
recipient countries. Yet, there is still no consensus about the growth effects of IMF
lending. There are other studies reporting positive growth effects of IMF lending (Dicks-
Mireaux et al. 2000, Mercer-Blackman and Unigovskaya 2000). These conflicting results
may arise from several sources, including differences in the types of IMF programs that
are investigated; differences in the groups of countries that are investigated (e.g. poor
developing versus emerging market economies or transition economies); differences in
the methodologies that are employed; and, perhaps most importantly, controlling for
other determinants of economic growth.
Barro and Lee (2002) examine the growth effects of IMF lending using the
instrumental variables (IV) estimation technique. After controlling for a number of
determinants of growth, and the endogeneity of IMF lending, they find that IMF
stabilization programs (SBA and EFF) have a statistically significant negative effect on
growth in the subsequent five years but not on contemporaneous growth rates.
Similarly, Hutchison (2001) investigates the growth effects of IMF stabilization programs
using the GEE. He finds that participation in an IMF-program, regardless of whether a
currency or balance of payments crisis has recently occurred, “costs” about 0.6-0.8
percentage points of real GDP growth annually. Przeworski and Vreeland (2000)
estimate a growth model using all types of IMF programs. They divided the sample into
(IMF) program observations and non-program observations and also include the inverse
11
Mills ratio in their regressions to control for selection bias. They conclude that these
programs reduce growth while countries remain under the programs, and furthermore,
the programs do not return benefits that would compensate for the losses incurred once
countries leave the program. Bordo and James (2000), using all types of IMF programs,
conclude that turning to the IMF may be harmful to a country’s real economic
performance, once account is taken of the self-selection bias, and that this effect has
been amplified since the Mexican crisis. Finally, Easterly (2001), using both IMF and
Bank structural programs, finds no systematic effect of adjustment lending on growth.
Using the GEE methodology, Dicks-Mireaux et al. (2000) evaluate the effects of
IMF lending to low-income countries eligible for the IMF’s Enhanced Structural
Adjustment Facility (ESAF). They find significant, positive effects of IMF-support
programs on growth. However, they conclude that the diagnostic tests cast doubt on the
appropriateness of the restrictive assumptions underlying the GEE and, accordingly,
about the reliability of the results. This finding raises questions about whether there are
inherent problems estimating GEE models with panel data. At a minimum, it strongly
indicates that future applications of the GEE to other data sets need to incorporate
standard diagnostic tests to ascertain whether the GEE methodology is valid for the
sample under study. Similarly, Mercer-Blackman and Unigovskaya (2000) investigate
whether greater compliance with Fund structural policy conditionality is associated with
better growth performance. They find that after controlling for other factors, transition
economies that successfully implement Fund programs, measured as greater
compliance with performance criteria, have better records of sustained economic growth
(defined as three consecutive years of positive real GDP growth); in contrast, they could
find no significant relationship between compliance with Fund structural benchmarks and
economic growth.
12
Rodrik (1995) claims that multilateral lending is not expected to affect subsequent
growth directly. Instead, using actual flows of multilateral lending over the period 1970-
1990 and OLS, he investigates whether IFIs have had an informational advantage, due
to their close monitoring of government policies, in determining which countries have
superior growth potential. He finds that the estimated coefficient on lagged multilateral
lending is uniformly negative, and becomes significant in some versions of his
regressions. Disaggregating each category of lending further, he finds that the
coefficient on net lending by the World Bank is consistently positive. However, the
estimated effect of IMF lending is always negative.
A number of case studies that analyze individual country experiences with IMF
programs also report adverse effects from these programs. Zaki (2001) argues that
Egypt’s arrangements since 1991 have successfully met the program targets, but have
experienced less success meeting targeted growth rates. He thus concludes that the
experience of Egypt demonstrates that economic reforms and liberalization in the
absence of commensurate political and institutional reforms will not produce economic
growth and prosperity for the poor.
Similarly, Yeldan (2001) claims that Turkish authorities were clearly successful in
maintaining the 2000 Disinflation Program targets both in exchange rate administration
and monetary control, as well as attaining the fiscal targets. In this sense the outburst of
the November crisis (in 2000) -and the ultimate collapse of the program in February
2001– cannot be attributed to any divergence from the monetary targets. He claims that
the collapse of the Turkish program was the result of internal inconsistencies and errors
in design, and thus modest gains in disinflation were achieved at the expense of de-
stabilization of the Turkish economy, along with a worsening of its financial and external
balances.
13
Rodrik and Kaplan (2001) argue that the appropriate counterfactual for Malaysian
capital controls in 1998 is the performance exhibited by the other countries subsequent
to their resort to IMF assistance. Using the “time-shifted difference-in-differences
methodology“ they find that the Malaysian controls produced better results than the
alternative on almost all dimensions. On the real side, the economic recovery was faster,
and employment and real wages did not decline as much as in comparable countries.
On the financial side, the stock market did better, interest rates fell more, and inflation
was lower than the comparison group.
The present study focuses on the impact of IMF and Bank lending on real
economic growth. As in Rodrik (1995), actual flows of IMF and Bank lending to
developing countries are used rather than the existence and/or number of programs per
se. Given that multilateral lending has many different objectives, depending in large part
on the circumstances of the members, and that IMF and Bank programs differ in size
and in their degree of emphasis on structural adjustment, Truman (2001, 3) claims that
“a better variable to try to capture the impact of such programs would be the size of the
program in (SDRs or dollars) per capita rather than just the number of programs
approved by the IMF and World Bank Executive Boards.” Moreover, using actual values
of multilateral lending makes more sense because there is a high level of noncompletion
in IMF programs and also because both IMF and Bank programs may not be
implemented as designed even if all loans are disbursed. It is also important to note that
since the primary focus is to assess the growth effects of IMF and Bank lending
programs, examining the separate impact of SAF/ESAF lending makes sense because
these lending programs are more likely to have growth effects. The data includes all
categories of IMF and Bank lending for 1970-1997. IMF lending includes stand-bys,
extended arrangements, structural adjustment facilities, and enhanced structural
14
adjustment facilities (recently renamed Poverty Reduction and Growth Facilities). World
Bank adjustment lending includes structural adjustment loans, sectoral structural
adjustment loans, and structural adjustment credits.
III. Model and Data
An empirical growth model is used to investigate the relationship between
international financial intervention and long-run growth. In general form, this model can
be characterized as
( ))(;,, ttttyt ZhkyF? = , (1)
where ?y t is country y’s per capita growth rate in period t, yt is initial GDP per capita, kt is
the physical capital stock per person, and ht is initial human capital per person.
Telephone mainlines per worker and life expectancy rates serve as proxies for the stock
of physical and human capital, respectively. Although the initial GDP per capita level is
employed to estimate the rate of conditional convergence, it is possible to interpret this
variable as a proxy for the stock of capital for a country. The variable Z represents a
vector of control and environmental variables, the majority of which are determined by
the decisions of governments or individuals. These variables are the value of IMF and
World Bank credits as a percent of GDP, the Gastil index, fertility rates, trade intensity
ratios, inflation rates, government consumption, war deaths, and three regional
dummies.
While GDP growth is calculated using the national accounts data come from
Easterly and Lu (GDNGD)6, initial GDP per capita levels are from Penn-World Table
6 They maintain the “Global Development Network Growth Database” on the World Bank web site.
15
5.6.7 Data for telephone mainlines per employee, life expectancy rates, fertility rates,
and inflation rates are from Easterly and Lu. The value of IMF and World Bank credits
as a percent of GDP, trade openness, government consumption, and gross domestic
investment data are from the World Development Indicators 1999 CDROM (WDI 1999).
Data for the Gastil index, which is a simple average of political rights and civil liberties,
used to measure the level of democracy in a country, are from Freedom House
database. The data have been normalized from zero to one, with one denoting the
greatest degree of freedom. Data for war deaths (WAR) are from Easterly (1999a). Use
of IMF credit indicates “repurchase obligations to the IMF for all uses of IMF resources
(excluding those resulting from drawings on the reserve tranche). These obligations,
shown for the end of the year specified, comprise purchases outstanding under the
credit tranches, including enlarged access resources, and all special facilities (the buffer
stock, compensatory financing, extended fund, and oil facilities), trust fund loans, and
operations under the structural adjustment and enhanced structural adjustment
facilities”. International Bank for Reconstruction and Development (IBRD) loans and
International Development Association (IDA) credits are extended by the World Bank
Group. The IBRD lends at market rates. Credits from the IDA are at preferential rates.
Cross-country regressions are estimated for a panel of one hundred developing
countries8 for the sample period 1970 to 1999. The number of countries is actually
limited by the availability of data. The system is a five-equation system estimated using
the seemingly unrelated regression technique (SUR) and/or three stages least squares
(3SLS) as in Barro (1997)9. The dependent variables are the average growth rates of
7Nuxoll (1994) and Summers and Heston (1991) discuss the reasons why the Summers and Heston data should be used for initial income levels while World Bank data should be used for income growth rates. 8 The list of countries is available from the authors. 9 See Greene (1997) for a detailed discussion of these techniques.
16
real per capita GDP over five periods: 1975-1979, 1980-1984, 1985-1989, 1990-1994
and 1995-1999.10
IV. Results
The estimates of the basic growth model are reported in Table 1. The model is
estimated as a five-equation panel using the seemingly unrelated regression technique.
Other than the constant terms, all coefficients are constrained to have the same value in
each equation. The range of reported R-squares spans the high and low values for the
five equations. As is the practice in the literature, the impacts of IMF and Bank credits
are estimated separately.
As expected, the proxies for human and physical capital have insignificant,
positive effects on growth, while the initial level of real GDP per capita and fertility rates
have significant negative effects. The estimated rate of convergence is consistent with
that found in several other studies. The Gastil index has the expected positive effect,
but is insignificant. Trade as a percentage of GDP (a measure of openness) has a
significant positive, albeit small, effect on growth.
Inflation, government consumption, and war all depress growth. The area-
specific dummies indicate below average growth in all three areas, but the effects are
significant only for Sub-Saharan African and Latin American nations.
Equations 1 and 2 in Table 1 report the estimated effects of IMF lending (as a
percent of GDP). Equation 1 uses the contemporaneous value of IMF lending, and
equation 2 uses a five-year lagged value. In both estimates IMF lending has a small but
significant negative effect on real growth. Equations 3 and 4 are similar estimates for
World Bank lending. The impact of contemporaneous bank lending is also negative, but
10 Estimates of the model for decade intervals obtain similar results, available from the authors.
17
neither the contemporaneous nor lagged effect of bank lending is significant. For IMF
lending, a 1% increase in lending is estimated to reduce growth by approximately 0.1%
annually. For Bank lending, the impact is much smaller. A 1% increase in lending
reduces growth by 0.01% annually.
An alternative approach is to divide the sample by income level and by level of
democracy. The rationale for this approach is that there exists considerable evidence in
the aid literature suggesting that aid flows encourage growth only when operating in a
good-policy environment. Investigation of the data indicates that middle-income and
good-democracy countries have relatively good policy measures, such as more open
economies, lower budget deficits and lower inflation. The estimated effects of IMF and
World Bank lending segregating countries by income level and degree of democracy are
reported in Table 2.11 Following Burnside and Dollar (2000) the break for the level of
income is set at real GDP per capita of $1900 at both the beginning and end of the
sample. As the Gastil index is insignificant, an alternative approach is to break the
sample into countries above and below 0.5 on the Gastil scale. As shown in Table 2,
dividing the sample by income level or degree of democracy obtains similar effects for
IMF lending. In most cases, both current and lagged IMF lending coefficients indicate a
significant negative effect. Dividing the sample according to income levels and degree
of democracy obtains different implications for Bank lending. Now, the estimated
coefficient for lagged Bank lending to low-income developing countries is positive. The
impact of current Bank lending to good-democracy developing nations is significantly
negative, while all other estimated effects are insignificant.
Some studies report positive effects of Fund lending programs. The Fund’s
recent move into development lending may account for this finding. Table 2 also reports
11 The basic model is the same for all estimates. Only the effects of IMF and Bank lending are reported. Estimates of the entire model are available upon request.
18
the results of estimates that exclude 73 observations for 26 countries with SAF/ESAF
arrangements (data from Dicks-Mireaux et al. 2000) during the period 1986-1991.12 The
magnitude and significance of the estimated coefficients is unchanged in many cases.
In those cases where the estimates change, the magnitude of the negative growth effect
increases. This suggests that the SAF/ESAF programs have either less negative or
positive growth effects. These results may explain why many studies using IMF
stabilization program data obtain negative growth effects while studies using SAF/ESAF
program data obtain nonnegative growth effects of IMF lending. These results also
suggest that IFIs, especially the IMF, should direct their lending more toward long-term
financing, which likely has a more favorable effect on growth in developing countries.
Since the IMF lends often to countries in the midst of a crisis, it is possible that a
simultaneity problem exists for estimates of the effects of IMF lending, and possibly for
Bank lending as well. To test whether simultaneity affects the estimated results, one
alternative approach is to estimate the model using average lending over the five
previous years instead of contemporaneous values, as discussed above. While using
lagged values is not a perfect solution for this problem, this alternative estimation
technique is one way to investigate the simultaneity issue. The results reported in Tables
1 and 2 do not indicate that a simultaneity problem exists.
Another approach to correct for simultaneity is to estimate the model using an
instrumental variables technique. However, a major limitation with this approach is that it
is difficult to find good instruments that are correlated with the exogenous variables but
are not correlated with the error terms.
In order to find plausible instruments, IMF and Bank credits are regressed on a
number of variables commonly used in the literature as well as variables not previously
12 The recent change to the PRGF occurred after the end of the sample period used in this study.
19
used. Table 3 reports the estimated results for the determinants of IFI lending13. One
interesting result is that countries with a British legal system received more IMF and
Bank credits (statistically significant) relative to countries with a French legal system. A
possible explanation is that both institutions conduct their business in English. Another
possible explanation can be related to the experiences of former British and French
colonies. The institutional framework of the former colonies may affect their ability to
negotiate agreements with the IFIs. Years of secondary and higher education of the
female population are significantly and negatively (positively) related to the level of Bank
(IMF) credits. Significant, negative coefficients for lagged initial income per capita and
female education are consistent with idea that low-income countries are the primary
concern of the World Bank, and that these nations are the primary recipients of Bank
credits. Significant and positive coefficients for the Sub-Saharan African and Latin
American dummies indicate that the IFIs direct their lending toward these regions. A
significant coefficient for the dummy for lagged IMF lending programs implies that
recidivist nations frequently receive IMF credits, but the similar variable is not significant
in the Bank credits regression. The significant coefficient for lagged IMF credits in the
Bank credits equations indicates that Bank credits follow IMF credits, but the reverse is
not supported by the data. The estimated coefficients suggest that foreign investment
and IMF lending are substitutes, while domestic investment and Bank lending are
complements.
The variables in Table 3 as well as current or lagged values of the variables used
in Table 1 are used as instruments for three-stage least-squares estimates reported in
Table 4. The estimated coefficients for contemporaneous IMF and Bank lending are
reported along with the SUR estimates of the same coefficients for the sample of 263 or 13 Many other commonly used variables have been tried (such as the current account balance, GDP per capita growth, trade openness, political instability measures, external debt, inflation, government consumption) but none were statistically significant.
20
205 countries. The SUR and 3SLS estimates for both IMF and Bank lending are almost
identical. Because of data availability limitations, the SUR estimates in Table 4 are for a
sample of 263 countries, compared to 407 observations for the SUR estimates in Tables
1. The similarity of the estimated coefficients for lagged and contemporaneous values in
Tables 1 and 2, and the identical SUR and 3SLS coefficients in Table 4 indicate that
simultaneity is not a problem for this sample. These estimates imply that
contemporaneous IMF lending does reduce growth in borrowing nations.
The conventional methodology used to estimate empirical growth models
includes both SUR and 3SLS. Use of alternative estimation techniques obtains different
implications for IMF and World Bank lending. Estimates using first differences and fixed
effects models are reported in Table 5. In the first differences estimates, the dependent
variable is the growth rate of GDP from 1980 to 1984 minus that from 1975 to 1979, and
similarly for other periods. The independent variables are first differences of the
variables that appear in Table1. Fixed effects estimates using the panel data allow for
both country and time effects.
In the first-difference and fixed-effect estimates, the effect of IMF lending is
statistically insignificant, except for the case of (lagged) contemporaneous lending,
where the estimated coefficient indicates a significant and negative (positive) effect in
good democracy (middle-income) developing countries. The estimated coefficients for
World Bank lending now generally indicate positive, and frequently significant, effects on
growth. These results indicate that bank lending benefits poor-democracy and low-
income-developing countries.14
It is important to note that there are substantial discrepancies in the estimated
convergence rates between the SUR estimates and fixed effects estimations. While the
14 Again, similar results are obtained excluding nations with SAF/ESAF arrangements.
21
SUR estimates indicate 2 to 3 percent annual convergence rates, fixed effects
regressions imply a 5 percent annual converge rates. Barro (1997) claims that
elimination of the cross-sectional information in the fixed effects estimation might
exacerbate measurement error bias, which would lead to higher estimated convergence
coefficients.15
It is often argued that IMF programs require demand-restraining measures that
decease domestic investment rather than increase savings. To assess this claim, the
empirical growth model framework is used to estimate the effects of IMF and Bank
lending on aggregate, public, and private investment, for the complete sample, and by
the level of income and the level of democracy.
The estimates for aggregate investment ratios are reported in Table 6. In seven
of the ten reported cases for aggregate investment, IMF lending has a significant,
negative effect on domestic investment. The results suggest that IMF lending with or
without SAF/ESAF countries likely reduces growth through reduced aggregate
investment. When public and private investment ratios are used as the dependent
variable, the results for IMF lending, reported in Table 7, imply that current IMF lending
significantly reduces both public and private investment in low-income and poor-
democracy countries.
The results in Table 6 indicate that Bank lending to middle-income developing
nations has a significant effect on investment. A 1% increase in Bank lending increases
investment in these countries by 0.5% annually.16 When public and private investment
ratios are used as the dependent variable, the results for Bank lending, reported in Table
7, imply that lagged Bank lending significantly increases public investment in three of the 15 See Barro (1997) for a more complete discussion of the issue. 16 When countries with SAF/ESAF arrangements are eliminated from the sample, IMF lending has similar results. Without the SAF/ESAF observations, current and lagged Bank lending has a positive impact in good democracy nations. Otherwise, the estimates produce similar results.
22
four groups of countries. Surprisingly, Bank lending reduces private investment in low-
income countries but has a significant, positive effect on private investment in middle-
income countries. Overall, these results are consistent with the existing evidence even
though the estimates use actual flows of multilateral lending and a different methodology
(SUR) to assess the growth effects of IFI lending.
In sum, the implications of the reported results are that IMF and World Bank
lending have different effects on growth. IMF lending does not increase developing
nation growth rates, and appears to harm growth in low-income and poor-democracy
developing nations. Alternatively, World Bank lending appears to benefit low-income
developing nations, and increases public investment in all groups of nations.
V. Discussion
The negative impact of IMF lending on growth can be due to reverse causation.
However, the cases of statistically significant lagged coefficients for IMF lending and the
statistically significant 3SLS estimates do not support this idea. Instead these results
imply that IMF lending may retard growth in some instances. The estimated results for
Bank lending are more favorable. Estimates for samples divided according to countries’
income levels and democracy levels imply that Bank lending benefits low-income
developing nations. This finding is important because low-income developing countries
are intended to be the main focus of Bank aid.
The literature contains a number of possible explanations for the failure of IMF
lending programs. Easterly (1999b) shows that international financial institution lending
had been based on the “financing gap model.” Recently, it has been recognized that
funding often fails to attain the intended results unless borrowing nations adopt sound
policies and structural reforms. The formulation of the “Washington consensus” in the
mid-1990s focuses on the development of good institutions, transparency, and good
23
governance. Thus, the conditions attached to recent loans emphasize “good
governance” and “good institutions.”17 Goldstein (2000) claims that emerging nations
with stronger corporate governance were less affected by the recent Asian crisis.
Bordo and Schwartz (2000) argue that Fund crisis management in recent years
has created a moral hazard problem that exacerbates international capital flows. Fund
crisis lending through the early 1980s avoided closures but did not prevent major
investor losses. During more recent crises, commercial banks appear to have sustained
few losses. If Fund lending drains out to private lenders, these capital account reversals
will likely increase cyclical volatility (see Meltzer 2002).
Another aspect of the moral hazard problem is the reluctance or refusal of some
national governments to assume ownership of IMF and World Bank programs, and/or
failure to implement the reforms required by the lending agreement. Killick (1995),
Stewart and FitzGerald (1996), James (1998), Bordo and Schwartz (2000), Woods
(2000), and Krueger (2000) attribute at least some of the ineffectiveness of Fund and
Bank programs to the failure of national governments to implement the necessary
reforms. Bordo and Schwartz (2000) blame lack of ownership for the failure of some
nations to develop even with fifty years of IMF and World Bank support. The success of
the Turkish economy during the late 1980s and early 1990s, and the more recent
success of Egypt, both serve as examples where implementing the required conditions
(and sometimes going further) resulted in periods of rapid economic growth.
Extensive recidivism and quasi-permanent involvement of the IMF with low-
income developing nations remains a continuing problem (Bird 1996a and 2001,
Easterly 2001). Having multiple arrangements with both the IMF and the World Bank
proves to be the rule rather than the exception. These facts demonstrate another facet
17 Woods (2000) discusses the governance conditions attached to IMF and World Bank lending.
24
of the moral hazard problem. Many developing nations experience “stop-go” cycles.
IMF aid and conditions improve both external and internal balance. However,
governments respond to the improved economic situation with monetary and fiscal ease,
which eventually result in inflation and deterioration of the balance of payments,
necessitating another IMF loan.18
VI. Conclusion
The two major international financial institutions, the IMF and the World Bank, are
important sources of capital funds for developing economies. The IMF now provides
developmental as well as “crisis” and adjustment funding, while the World Bank focuses
on developmental loans. An important question is whether the lending programs of
these IFIs have a positive impact on the growth rates of developing economies. In
particular, frequent, strenuous criticism of IMF lending raises questions about the
efficacy of its loan programs. In this study, an empirical growth model is used to analyze
the growth effects of IMF and Bank lending.
Empirical analysis indicates that contemporaneous IMF lending negatively affects
growth, probably by reducing domestic investment. However, estimates of lagged
effects and estimates obtained using alternative techniques indicate no significant
impact of IMF lending. The sum of the evidence regarding IMF lending is ambiguous.
Bank lending does appear to help low-income emerging economies, primarily by
stimulating public investment.
Structural problems in borrowing nations and/or various moral hazard problems
may account for the apparent failure of IMF lending programs. However, Fund programs
may still be more about resolving crises than promoting growth (Goldstein 2000). Also,
18 Reasons why the IFIs continue to lend to recidivist nations include failure to recognize the importance of ownership, failure to understand the internal politics of borrowing nations, and political pressure from the industrial nations to continue supporting the recidivists.
25
as emphasized by Gavin and Rodrik (1995), Rodrik (1995), Krueger (1998), Bordo and
James (2000), and Fischer (2000b), Fund lending involves more than promoting growth.
These other roles include provision and analysis of statistical data, training, and policy
advice and technical assistance. Thus, overall evaluation of both Fund and Bank
lending programs weigh the benefits of the entirety of services provided against the
costs. Unfortunately, to date, stimulating economic growth is not a benefit of IMF
lending, while a more favorable conclusion can be drawn regarding World Bank lending
programs.
26
Table 1
Estimates of the Impact of IMF and World Bank Lending in a Growth Model
SUR Panel Growth Regressions for Developing CountriesTime Period: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variable: Real GDP per capita growth ratesRegression No. 1 2 3 4Observations 407 406 407 406Initial GDP per capita -2.75** -2.40** -2.17* -1.61
(2.75) (2.35) (2.00) (1.52)Life Expectancy (lagged) 2.30 2.57 2.16 3.35
(0.56) (0.61) (0.49) (0.77)Telephone mainlines per employee 0.77 0.77 0.74 0.57
(1.24) (1.22) (1.13) (0.87)Gastil Index 0.95 0.91 0.92 0.76
(1.65) (1.55) (1.54) (1.26)Fertility Rates (lagged) -6.72** -6.48** -6.08** -6.57**
(4.17) (3.93) (3.57) (3.84)Trade (% of GDP) 0.018** 0.017** 0.016** 0.014**
(3.70) (3.56) (3.27) (2.92)Inflation Rates -0.0015** -0.0015** -0.0015** -0.0015**
(5.22) (5.16) (5.21) (5.22)Government Consumption -0.10** -0.11** -0.11** -0.10**
(3.63) (3.78) (3.62) (3.28)War Deaths -1.00* -0.89* -0.95* -0.92*
(2.14) (1.87) (1.95) (1.88)Sub-Saharan African Dummy -1.18* -1.20* -1.26* -1.24*
(2.27) (2.25) (2.28) (2.23)Latin American Dummy -2.00** -2.06** -2.11** -2.11**
(4.17) (4.22) (4.17) (4.14)East Asian Dummy -0.27 -0.19 -0.11 -0.10
(0.46) (0.31) (0.18) (0.16)IMF credits (% of GDP) -0.10**
(3.55)Lagged IMF credits -0.07*
(2.11)WB credits (% of GDP) -0.013
(0.82)Lagged WB credits 0.017
(0.96)R-square (low-high) .18 -.50 .15 -.51 .15 -.49 .15 -.49The system has 5 equations, where the dependent variables are the GDP per capita growthrate over each five-year period. Each equation has a different constant term.Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.** Significant at the 1 percent-level. * Significant at the 5 percent-level.
27
Table 2
Effects of IMF and Bank Lending by Country Type
SUR Panel Growth Regressions for Developing CountriesTime Period: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variable: Real GDP per capita growth rates
Low-income developing coun. Middle-income developing coun.Observations 271 198* 136 136*IMF credits (% GDP) -0.10** -0.10** -0.14* -0.14*
(3.20) (2.72) (1.77) (1.77)Lagged IMF credits -0.09** -0.13** -0.03 -0.03
(2.38) (2.52) (0.39) (0.39)IBRD credits (%GDP) 0.005 0.006 -0.08 -0.08
(0.25) (0.24) (1.10) (1.10)Lagged IBRD credits 0.04* 0.03 -0.07 -0.07
(1.92) (1.08) (1.12) (1.12)
Poor-democracy developing coun. Good-democracy developing coun.Observations 237 179* 170 155*IMF credits (% GDP) -0.10** -0.10** -0.19** -0.29**
(2.58) (2.80) (3.97) (4.40)Lagged IMF credits -0.08* -0.10* -0.09* -0.17*
(1.72) (2.28) (1.79) (2.26)IBRD credits (%GDP) 0.008 -0.011 -0.13** -0.14**
(0.37) (0.41) (3.81) (2.96)Lagged IBRD credits 0.034 -0.002 -0.05 -0.08
(1.43) (0.07) (1.41) (1.52)The system has 5 equations, where the dependent variables are the GDP per capita growthrate over each five-year period. Each equation has a different constant term.Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.** Significant at the 1 percent-level. * Significant at the 5 percent-level.* These columns exclude observations for the countries that had SAF/ESAF arrangements at some time during the period considered.
28
Table 3
Determinants of IFI lending
SUR Panel IFI Regressions for Developing CountriesTime Period: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variables: IMF credits or WB credits (as percent of GDP)
IMF Credits WB CreditsObservations 275 200Years of secondary&high schooling 0.69* Years of secondary&high schooling -2.09**in the female population (2.07) in the female population (3.46)Landlocked dummy -2.15** Landlocked dummy -1.79*
(4.09) (2.06)Real exchange rate index (lagged) 0.007** Real exchange rate index (lagged) 0.005*
(4.49) (2.05)Aid (% of GDP, lagged) 0.11** Aid (% of GDP, lagged) 0.40**
(4.11) (6.77)Sub-Saharan African Dummy 1.46** Sub-Saharan African Dummy 3.61**
(2.71) (4.19)Latin American Dummy 1.10* Latin American Dummy 2.85**
(2.09) (3.56)East Asian Dummy -0.87 East Asian Dummy 0.77
(1.36) (0.82)British legal system 1.83** French legal system -1.49*
(4.72) (2.24)Telephone mainlines per employee -1.08** Gross domestic inv. (%GDP, lagged) 0.09*
(2.51) (2.43)Foreign direct inv.(% of GDP, lagged) -0.18* Initial GDP per capita, (lagged) -3.21*
(2.24) (2.32)IMF program dummy (lagged) 1.13** Budget Surplus (% of GDP, lagged) -0.13*
(3.39) (2.25)IMF credits, lagged 0.49**
(4.67)Mining % of GDP -8.04*
(2.42)R-square .24 -.55 R-square .30 -.76The system has 5 equations, where the dependent variables are five-year averages of IMF or WB credits. Each equation has a different constant term.Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.** Significant at the 1 percent-level; * Significant at the 5 percent-level.Gross domestic investment data are from WDI 1999; data for female education are from Barro and Lee (2000); mining data are from Hall and Jones (1999); all the other data are from Easterly and Lu.
29
Table 4
3SLS Estimates of the Effect of IMF and Bank Lending
3SLS/SUR Panel Growth Regressions for Developing CountriesTime Period: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variable: Real GDP per capita growth rates
# of obs # of obsSUR R2 3SLS R2
IMF credits (% GDP) -0.14** 263 -0.14** 263(2.79) .18 -.54 (2.45) .16 -.55
WB credits (%GDP) -0.04 205 -0.04 205(1.34) .25 -.74 (1.52) .26 -.74
The system has 5 equations, where the dependent variables are the GDP per capitagrowth rate over each five-year period. Each equation has a different constant term.Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.** Significant at the 1 percent-level; * Significant at the 5 percent-level.Use of 3SLS to estimate growth regressions reduces the country sample to 263 (or 205)observations from 407. The loss of observations precludes county-type estimates.
30
Table 5
Fixed Effect and First Difference Estimates of the Effects of IMF and Bank Lending
Panel Growth Regressions for Developing CountriesTime Period: First differences of five five-year periods/five periodsDependent Variable: Real GDP per capita growth rates
First # of obs Fixed # of obsDifferences(SUR) R2 Effects R2
IMF credits (% GDP) -0.03 321 -0.04 86(0.65) .13 -.44 (0.69) .60
Lagged IMF credits 0.003 320 0.03 86(0.06) .12 -.45 (0.66) .60
WB credits (%GDP) -0.003 321 0.02 86(0.13) .13 -.45 (0.93) .60
Lagged WB credits 0.03 320 0.05* 86(1.41) .13 -.44 (2.05) .60
Low-income Developing Countries Middle-income Developing CountriesRegression No. First Differences Fixed Effects First Differences Fixed EffectsObservations 213 58 108 28
IMF credits (% GDP) 0.01 -0.007 -0.06 -0.09(0.21) (0.12) (0.65) (0.82)
Lagged IMF credits -0.04 0.04 0.19* 0.12(0.63) (0.67) (1.70) (1.14)
WB credits (%GDP) 0.01 0.05* 0.12 0.13(0.46) (1.98) (1.08) (1.21)
Lagged WB credits 0.05* 0.09** 0.008 0.11(1.91) (2.91) (0.07) (0.99)
Poor-Democracy Developing Coun. Good-Democracy Developing Coun.Regression No. First Differences Fixed Effects First Differences Fixed EffectsObservations 187 50 134 36
IMF credits (% GDP) 0.07 0.04 -0.08 -0.12*(0.98) (0.57) (1.38) (1.82)
Lagged IMF credits 0.04 0.09 0.02 0.004(0.45) (1.24) (0.37) (0.07)
WB credits (%GDP) 0.03 0.06* -0.08* -0.08(0.94) (2.16) (1.82) (1.53)
Lagged WB credits 0.06* 0.09** 0.009 0.02(1.80) (2.76) (0.19) (0.44)
Other regressors include the variables reported in Table 1 excluding time-invarient variables.Fixed effects estimations also includes country and time dummies (not reported).
31
Table 6
Impact of IFI Lending on Aggregate Investment
SUR Panel Investment Regressions for Developing CountriesTime Period: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variable: Total Investment (% of GDP)Observations 401 400 401 400IMF credits (% GDP) -0.27**
(4.01)Lagged IMF credits -0.11
(1.37)WB credits (%GDP) 0.008
(0.20)Lagged WB credits 0.004
(0.10)R-square .32 -.58 .31 -.55 .32 -.55 .32 -.55
Low-income developing coun. Middle-income developing coun.Observations 266 135IMF credits (% GDP) -0.30** -0.33*
(4.19) (2.18)Lagged IMF credits -0.25** 0.17
(2.92) (0.93)WB credits (%GDP) 0.03 -0.07
(0.61) (0.38)Lagged WB credits -0.005 0.51**
(0.09) (2.79)
Poor-democracy developing coun. Good-democracy developing coun.Observations 233 168IMF credits (% GDP) -0.34** -0.18*
(3.63) (1.86)Lagged IMF credits -0.21* -0.04
(1.85) (0.36)WB credits (%GDP) 0.03 0.12
(0.69) (1.38)Lagged WB credits 0.009 0.08
(0.16) (0.90)The system has 5 equations, where the dependent variables are the investment shares over each five-year period. Each equation has a different constant term.Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.Regressors are same as in the growth model.** Significant at the 1 percent-level; * Significant at the 5 percent-level.
32
Table 7
Effects of IFI Lending on Public and Private Investment
SUR Panel Investment Regressions for Developing CountriesTime Dimension: Five five-year periods, 1975 -1979 to 1995 -1999 Dependent Variable: Public/Private Investment (% of GDP)
Public Investments Private InvestmentsObservations 315 315 315 315 309 309 309 309IMF credits (% GDP) -0.10* -0.17**
(2.05) (3.08)Lagged IMF credits -0.02 -0.099
(0.33) (1.27)IBRD credits (%GDP) 0.09** -0.07*
(3.61) (2.23)Lagged WB credits 0.08* -0.07***
(2.37) (1.89)R-square -.01 -.37 .01 -.37 .19 -.43 .11 -.42 .40 -.61 .38 -.61 .39 -.61 .38 -.62
Low-income developing coun. Middle-income developing coun. Public Private Public Private
Observations 218 213 97 96IMF credits (% GDP) -0.16** -0.19** -0.07 -0.10
(2.71) (3.18) (0.85) (0.79)Lagged IMF credits -0.11 -0.14*** 0.05 0.13
(1.49) (1.89) (0.57) (0.99)IBRD credits (%GDP) 0.11** -0.06*** 0.36** 0.10
(3.98) (1.73) (3.70) (0.66)Lagged IBRD credits 0.08* -0.07*** 0.50** 0.36*
(2.19) (1.75) (4.82) (2.07)
Poor-democracy developing coun. Good-democracy developing coun. Public Private Public Private
Observations 188 183 127 126IMF credits (% GDP) -0.19** -0.14*** 0.06 -0.14
(2.71) (1.84) (0.96) (1.65)Lagged IMF credits -0.07 -0.05 0.13*** -0.03
(0.84) (0.59) (1.77) (0.25)IBRD credits (%GDP) 0.14** -0.04 0.13** -0.04
(4.31) (1.27) (2.54) (0.58)Lagged IBRD credits 0.12** -0.07 0.10 0.03
(2.76 (1.60) (1.63) (0.33)The system has 5 equations, where the dependent variables are the investment shares over each five-year period. Each equation has a different constant term. Other coefficients are restricted to be the same for all periods. t-statistics are in parentheses.* significant at the 5 percent level** significant at the 1 percent level*** significant at the 10 percent level
33
VII. References
Alesina, Alberto, David Dollar. (2000). “Who gives foreign aid to whom and why?”, Journal of Economic Growth, 5, 33–63.
Barro, Robert J. (1997). Determinants of economic growth: A cross-country empirical study. Cambridge and London: MIT Press.
Barro, Robert J. (1988). “The IMF doesn’t put out fires, it starts them,” Business Week, 12/07/98 Issue 3607, p18, 1p, 1c.
Barro, Robert J. Jong-Wha Lee. (2000). "International data on educational attainment: updates and implications," manuscript, Harvard University, February 2000.
Barro, Robert J., Jong-Wha Lee. (2002). “IMF programs: Who is chosen and what are the effects?”, NBER Working Paper Series No. 8951.
Bird, Graham. (1996a). “Borrowing from the IMF: The policy implications of recent empirical research”, World Development, 24(11), 1753-1760.
Bird, Graham. (1996b). “The International Monetary Fund and developing countries: a review of the evidence and policy options”, International Organization 50(3), 477-511.
Bird, Graham. (2001). “IMF programs: Do they work? Can they be made to work better?”, World Development, 29(11), 1849-1865.
Birdsall, Nancy, Juan Louis Londono. (1997). “Asset inequality matters: An assessment of the World Bank’s approach to poverty reduction”, American Economic Review, 87(2), 32-37.
Boone, Peter. (1996). “Politics and the effectiveness of foreign aid”, European Economic Review, 40, 289–329.
Bordo, Michael D., Harold James. (2000). “The International Monetary Fund: Its present role in historical perspective”, NBER Working Paper Series No. 7724.
Bordo, Michael D, Anna J. Schwartz. (2000). “Measuring real economic effects of bailouts: Historical perspectives on how countries in financial distress have fared with and without bailouts”, Carnegie-Rochester Conference on Public Policy, 53, 81-167.
Bretton Woods Commission, (1994). Bretton Woods: Looking to the Future, Bretton Woods Commission, Washington, D.C.
Burnside, Craig, David Dollar. (2000). “Aid, policies, and growth”, American Economic Review, 90(4), 847-868.
Camdessus, Michel. (1994). “International cooperation for high-quality growth: The role of the IMF at 50”, Washington DC, International Monetary Fund, http://www.imf.org/external/np/sec/mds/1995/MDS9521.HTM.
Camdessus, Michel. (2000), “Development and Poverty Reduction: A Multilateral Approach”, Washington DC, International Monetary Fund, http://www.imf.org/external/np/speeches/2000/021300.htm.
Conway, Patrick. (1994). “IMF lending programs: Participation and impact”, Journal of Development Economics, 45, 365-391.
34
Dicks-Mireaux, Louis, Mauro Mecagni, Susan Schadler. (2000). “Evaluating the effect of IMF lending to low-income countries”, Journal of Development Economics, 61, 495-526.
Easterly, William. (1999a). “Life during growth”, Journal of Economic Growth, 4, 239-275.
Easterly, William. (1999b) “The ghost of financing gap: Evaluating the growth model of the international financial institutions,” Journal of Development Economics, 60, 423-438.
Easterly, William. (2001). “IMF and World Bank structural adjustment programs and poverty”, World Bank, April 2001.
Feinberg, Richard E. (1988). “The changing relationship between the World Bank and the International Monetary Fund”, International Organization 42(3), 545-560.
Feldstein, Martin. (1998). “Refocusing the IMF”, Foreign Affairs, March/April 1998, pp. 20-33.
Fischer, Stanley. (1997). “Applied economics in action: IMF programs”, American Economic Review, 87(2), 23-27.
Fischer, Stanley. (2000a). “Presentation to the International Financial Institution Advisory Commission”, Washington DC, http://www.imf.org/external/np/speeches/2000/020200.htm
Fischer, Stanley. (2000b). “Remarks by Stanley Fisher, Acting Managing Director”, Remarks to the Bretton Woods Committee Meeting on April 4, 2000, International Monetary Fund, http://www.imf.org/external/np/speeches/2000/040400.htm
Gastil, Raymond D. (1988). Freedom in the World: Political Rights and Civil Liberties 1987-1988. New York: Freedom House.
Gavin, Michael, Dani Rodrik. (1995). “The World Bank in historical perspective”, American Economic Review, 85(2), 329-334.
Greene, William H. (1997). Econometric Analysis. New Jersey: Prentice Hall.
Goldstein, Morris, Peter Montiel. (1986). “Evaluating Fund stabilization programs with multicountry data: Some methodological pitfalls”, International Monetary Fund Staff Papers, 33, 304-344.
Goldstein, Morris. (2000). “IMF structural programs”, NBER Woodstock Paper, (December).
Hall, Robert C., Charles Jones. (1999) “Why do some countries produce so much more output per worker than others?” Quarterly Journal of Economics, 113, 83-116.
Haque, Nadeem, Mohsin S. Khan. (1998) “Do IMF supported programs work? A survey of cross-country empirical evidence.” IMF Working Paper No. 169.
Heckman, J. (1979). “Sample selection bias as a specification error”, Econometrica, 47), 153-61.
Hutchison, Michael M. (2001). “A cure worse than the disease? Currency crises and the output costs of IMF-supported stabilization programs”, NBER Working Paper Series No. 8305.
35
International Financial Institution Advisory Commission. (2000). “Report of the International Financial Institution Advisory Commission”, Allan H. Meltzer, Chairman, Washington DC, March.
International Monetary Fund and World Bank (2000). “The IMF and World Bank Group: An enhanced partnership for sustainable growth and poverty reduction,” Washington, D.C., International Monetary Fund, http://www.imf.org/external/np/omd/2000/part.htm.
James, Harold. (1998). “From grandmotherliness to governance: The evolution of IMF conditionality”, Finance and Development, 35(4), http://www.imf.org/external/pubs/ft/fandd/1998/12/james.htm.
Kaplan, Ethan, Dani Rodrik. (2001). “Did the Malaysian capital controls work? NBER Working Paper Series No. 8142.
Killick, Tony. (1995). IMF Programmes in Developing Countries: Design and Impact, London, Routledge.
Knight, Malcolm, Julio A. Santanella. (1997). “Economic determinants of IMF financial arrangements”, Journal of Development Economics, 54, 405-436.
Krueger, O. Anne. (1998). “Wither the World Bank and the IMF?”, Journal of Economic Literature, 36(4), 1983-2020.
Krueger, O. Anne. (2000). “IMF stabilization programs”, NBER Woodstock Paper, Version:2, (November).
Meltzer, Allan H. (2002). “Back to bailouts,” The Wall Street Journal, August 7, 2002, A14.
Mercer-Blackman, Valerie, Anna Unigovskaya. (2000). “Compliance with IMF program indicators and growth in transition economies”, IMF Working Paper No. 47.
Mussa, Michael, Miguel Savastano. (2000). “The IMF approach to economic stabilization” NBER Macroeconomics Annual, 1999. V. 14, 79-122, Cambridge and London: MIT Press.
Nuxoll, Daniel A. (1994). “Differences in relative prices and international differences in growth rates”, American Economic Review, 84, 1423-1436.
Przeworski, Adam, James Vreeland. (2000) "The effect of IMF programs on economic growth", Journal of Development Economics, 62, 385-421.
Rodrik, Dani. (1995). “Why is there multilateral lending?” NBER Working Paper Series, No. 5160.
Rowlands, Dane. (1996). “New lending to less developed countries: The effects of the IMF”, Canadian Journal of Economics, 2, S443-S447.
Stewart, Frances, Valpy FitzGerald. (1996). “The IMF and the global economy: Implications for developing countries”, Quenn Elizabeth House Working Paper Series, No:3.
Stiglitz, J. (2000). “What I learned at the world economic crisis”, The New Republic, April 17.
Thacker C. Strom. (1999). “The high politics of IMF lending”, World Politics, 52(1), 38-75.
36
Truman, Edwin M. (2001). “Comment on IMF and World Bank structural adjustment programs and poverty by William Easterly”, NBER Conference on Management of Currency Crises March 28-31, 2001.
Woods, Ngaire. (2000). “The challenge of good governance of the IMF and the World Bank themselves”, World Development, 28(5), 823-841.
Yeldan, Erinc. (2001). “On the IMF-directed disinflation program in Turkey: A program for stabilization and austerity or a recipe for impoverishment and financial chaos?”, Bilkent University Working paper.
Zaki, Mokhklis Y. (2001). “IMF-supported stabilization programs and their critics: Evidence from the recent experience of Egypt”, World Development, 29(11), 1867-1883.