Historical Evidence on Determinants of Growth

That recovery has begun much earlier in some countries and still has not started in others, and that the rate of recent growth varies considerably among countries, speaks to the view that recovery and sustained growth are not an automatic or cyclical rebound. Several issues concerning the transition and growth are elaborated here:

• The applicability of conventional explanations of growth in market economies;

• The effect of special initial conditions in the transition;

• The contribution of government policies, in particular financial stabilization and structural reforms;

• The sustainability of growth; and

• The time pattern of reforms—gradual or more rapid, early or delayed.

A brief summary of the main factors that appear to explain growth is then provided.²⁴ This paper does not attempt to address all the complex causal relations explaining growth. Rather, it draws on a variety of sources using differing methodologies such as case histories, earlier empirical studies, some new analysis updating these studies in a simplified empirical framework.²⁵ and a summary presentation of some stylized facts of possible growth determinants.

The earliest and probably least controversial conclusion of empirical studies is that stabilization is a necessary condition for recovery of output (Fischer, Sahay, and Vegh, 1996, and 1998).²⁶ The apparent exceptions of Bulgaria and Romania fell into line when their growth and then stabilization reversed. Two exceptions exist at present—Belarus and Uzbekistan—and Box 5 discusses their similarity with the reversal cases. Second, and somewhat more controversial, emphasis was also put on the additional, necessary conditions to promote growth: liberalization and structural reforms. Whether the framework was a simple one relating only to growth and some index of structural reforms (Sachs, 1996; Selowsky and Martin, 1997; and Åslund, Boone, and Johnson, 1996), or a more sophisticated one reflecting effects of stabilization, initial conditions, conflicts, and so on (Fischer, Sahay, and Végh, 1996 and 1998; Hernández-Catá, 1997; De Melo and others, 1997; and Berg and others, 1999), the conclusion was firm: more reforms are associated with better growth performance. The results are not without exceptions, Belarus and Uzbekistan today being the key ones, and Bulgaria and Romania earlier. Åslund, Boone, and Johnson (1996) point to a dichotomy here: while empirical studies concluded that fast and early reforms result in early and strong recovery, theoretical work on transition has often shown that a gradual pace might lead to less early decline of output (Aghion and Blanchard, 1993). A third set of conclusions relates to other factors such as wars and initial conditions; these do have an effect that is country specific. Some studies have tried to determine the relative importance of adverse initial conditions versus the role of policies (De Melo and others, 1997; Berg and others, 1999), although they have not been specific about the trade-offs, such as how much a high share of industrial output impedes growth or how much of a compensation is needed from faster reforms.²⁷ A fourth set of conclusions relates to institutions and conditions such as rule of law, corruption, and fairness of the tax burden. These factors are even less easily measured than the degree of liberalization, hence not surprisingly the statistic used varies a great deal among studies. Nevertheless, the studies such as Brunetti, Kisunko, and Weder (1997a, b; 1998), Johnson, Kaufmann, and Sheifer (1997), and Olson, Sarna, and Swamy (1997) all concur that growth is higher where these elements show higher levels of institutional achievement. Fifth, the conventional factor input explanations of growth theory (investment, labor, human capital, and the like) are not considered in any of the empirical studies, reflecting the view noted in Section II that the short-run recovery from transitional recession is of a special nature. Only Wolf (1997) tests for the statistical importance of investment-to-GDP ratios, and finds a negative correlation.

In what follows, the paper provides some stylized facts that illustrate, confirm, and elaborate on these conclusions and permit some updating of the kind of empirical analysis noted in Annex III. Availability of data through 1997 makes it easier to distinguish the factors influencing decline from those inducing recovery and permits a considerable updating of previous empirical studies cited earlier.²⁸ Extension of the sample period even further to include preliminary estimates for 1998 for the main variables leaves the results of the empirical analysis unchanged, despite the general slowdown of growth in 1998.

First, it appears that the conventional growth determinants—investment, labor, and human capital expansion—played only a limited role in the recovery. As noted earlier (Section III), while there is evidence that economic recovery was accompanied by an upturn in investment, movements in labor productivity were a somewhat better predictor of the onset of recovery or output reversals.²⁹ For over–industrialized, distorted, and inefficient transition economies, recovery only comes after some elimination of the wasteful old production (see Hernández-Catá, 1997), and usually cannot be based on a large investment effort to build the new before the proper incentives for efficient resource use are in place. The impact of export growth is less clear. The group averages for export growth show a broad statistical association with growth (Table 6, panel E), although the empirical studies generally do not test for openness on exports (with the exception of Wolf, 1997, who finds no correlation). The explanation may be simply that almost all transition economies became relatively open very quickly, and those that achieved earlier restructuring were able to reorient their trade quickly to new markets (see Box 4). That is, exports may not so much have led to growth, but rather restructuring—which was necessary for recovery–tended to promote exports.

A similar conclusion applies to foreign direct investment. While there exists a clear association between economic growth and cumulative foreign direct investment per capita (Table 6. panel H), at least for CEE and the Baltics, the relationship between growth and foreign direct investment may be mutually reinforcing (as discussed in Annexes II and III). In other words, while there is little doubt that foreign direct investment promotes growth, those factors that promote greater stabilization and reforms also attract foreign direct investment.³⁰

Second, initial conditions are not without importance, but their impact becomes less over time and can be offset by better policies, in particular slightly faster progress on structural reforms. It is not a straightforward matter to observe statistically the effect of initial conditions such as historical experience and attitudes, differences in degree of distortion under central planning, or levels of development. One reason is the difficulty in measuring initial conditions; thus, while there is broad consensus that before 1990 the countries of CEE had relatively less distorted economies than the U.S.S.R., this does not provide a measure of how much they differed. Another reason is that the impact of initial conditions may be indirect; that is, they affect growth through politics. The degree of commitment to reforms and their effective implementation may be partially explained by such differences in initial conditions. Three recent studies address this—De Melo and others (1997), Hernández-Catá (1997) and Wolf (1997)—generally finding that countries of CEE and the Baltics, because of fewer historical distortions, a shorter history of communism, and closer proximity to market economies were more likely to implement market-based policies early and quickly.

Although such difficulties will qualify any statistical analysis, it is still useful to undertake it, using measures of initial conditions compiled by De Melo and others (1997). Annex III details the results that illustrate the relative impact of initial conditions on growth. Initial overindustrialization lowers the rate of output growth (by about 0.4 percentage points for each additional 5 percentage points of the share of industry in total output). But this effect is offset by improvements in the growth rate arising from even modest progress on structural reforms; a move from a low level of reform of 0.2 (as measured by the reform index; see Annex IV for details) to 0.3 yields 1 to 2 percentage points of additional growth. Thus, moving from an intermediate stage of reforms, as indicated by an index value of 0.4, to a more advanced level of 0.7, would add 3 to 6 percentage points of economic growth. It is notable that in the analysis of Annex III. when government size is controlled for in growth regressions for CIS countries, the magnitude of the effect of initial conditions is markedly lower. Another method frequently used to capture initial conditions is the use of dummy variables differentiating groups of countries: former Soviet Union versus other; conflict cases; and the like. Earlier growth studies, as noted above, found these differences to be significant. Using data through 1998. however, this kind of differentiation shows no statistically significant results, suggesting that the impact of such effects diminishes over time. However, regression analysis done separately for CIS and other countries gives an interesting result: the impact on growth of a given “amount” of reform as measured by the type of index constructed by the European Bank for Reconstruction and Development (EBRD) is less for CIS countries.

Other initial conditions matter as well, such as very low income (including after civil strife) or resource availability. Thus in Albania, Armenia, Georgia, and the Kyrgyz Republic, substantial progress in reforms saw growth rates surge, reflecting a catch-up phenomenon, as well as available natural resources to be exploited (for instance gold in the case of the Kyrgyz Republic). But resource wealth can also be an impediment to reforms and therefore may provide only temporary, unsustainable (or at least weak) growth: Uzbekistan may be such a case (see Box 5). Civil strife is found by many analysts to affect growth negatively, but the effects seem to disappear quickly (see De Melo and others, 1997): the quick recoveries in Armenia, Georgia, and now Tajikistan are such examples. Thus, on balance, the evidence suggests that initial conditions have had an effect, but it should not be exaggerated because it declines over time and can be offset by stronger reform progress. Furthermore, the effect is possibly more on the political will and capacity to implement reforms than directly on growth. Hence the conclusion remains that where more reforms are done, more growth will occur.

Table 6.

Economic Performance by Country Group

Table 6.

Economic Performance by Country Group

		Central Europe	Baltics	Southeast Europe	CIS
A. Countries
	Consistent growth	Croatia Czech Rep Hungary Poland Slovak Rep. Slovenia	Estonia Latvia Lithuania	Macedonia. F.Y.R.	Armenia Azerbaijan Belarus Georgia Kyrgyz Rep. Uzbekistan
	Growth reversals			Albania Bulgaria Romania
	Little or no growth				Kazakhstan Moldova Russia Tajikistan Turkmenistan Ukraine
B. Average Annual Growth Rate Since Initial Recovery(Percent a year)
	Consistent growth	4.3	4.7	1.7	5.7
	Growth reversals			1.4
	Little or no growth				0.5
C. Average Index of Liberalization, 1995 Consistent growth excl.
	Belarus and Uzbekistan for CIS	0.79	0.72	0.64	0.54
	Belarus and Uzbekistan				0.54
	Growth reversals			0.62
	Little or no growth				0.50
D. Average Inflation During Years of Growth(Percent a year)
	Consistent growth	14	18	1	27
	Growth reversals			67
	Little Or no growth				37
E. Average Growth In Exports, 1994-97(Percent a year)
	Consistent growth	15.4	24.4	2.9	14.3
	Growth reversals			11.2
	Little or no growth				8.4
F. Share of Private Sector in GDP, Mid-1997(Percent) Consistent growth excl.
	Belarus and Uzbekistan for CIS	66	67	SO	53
	Belarus and Uzbekistan				33
	Growth reversals			48
	Little or no growth				44
G. Index of Fund Program Implementation, 1993-97
	Consistent growth	92.6	95.2	83.5	82.2
	Growth reversals			72.5
	Little or no growth				80.7
H. Foreign Direct Investment per Capita, 1990-96(U.S. dollars,)
	Consistent growth	74.1	68.7	5.6	7.5
	Growth reversals			9.8
	Little or no growth				15.9

View Table

Table 6.

Economic Performance by Country Group

		Central Europe	Baltics	Southeast Europe	CIS
A. Countries
	Consistent growth	Croatia Czech Rep Hungary Poland Slovak Rep. Slovenia	Estonia Latvia Lithuania	Macedonia. F.Y.R.	Armenia Azerbaijan Belarus Georgia Kyrgyz Rep. Uzbekistan
	Growth reversals			Albania Bulgaria Romania
	Little or no growth				Kazakhstan Moldova Russia Tajikistan Turkmenistan Ukraine
B. Average Annual Growth Rate Since Initial Recovery(Percent a year)
	Consistent growth	4.3	4.7	1.7	5.7
	Growth reversals			1.4
	Little or no growth				0.5
C. Average Index of Liberalization, 1995 Consistent growth excl.
	Belarus and Uzbekistan for CIS	0.79	0.72	0.64	0.54
	Belarus and Uzbekistan				0.54
	Growth reversals			0.62
	Little or no growth				0.50
D. Average Inflation During Years of Growth(Percent a year)
	Consistent growth	14	18	1	27
	Growth reversals			67
	Little Or no growth				37
E. Average Growth In Exports, 1994-97(Percent a year)
	Consistent growth	15.4	24.4	2.9	14.3
	Growth reversals			11.2
	Little or no growth				8.4
F. Share of Private Sector in GDP, Mid-1997(Percent) Consistent growth excl.
	Belarus and Uzbekistan for CIS	66	67	SO	53
	Belarus and Uzbekistan				33
	Growth reversals			48
	Little or no growth				44
G. Index of Fund Program Implementation, 1993-97
	Consistent growth	92.6	95.2	83.5	82.2
	Growth reversals			72.5
	Little or no growth				80.7
H. Foreign Direct Investment per Capita, 1990-96(U.S. dollars,)
	Consistent growth	74.1	68.7	5.6	7.5
	Growth reversals			9.8
	Little or no growth				15.9

Third, macroeconomic stabilization and general progress on market-oriented reforms are dominant determinants of recovery in the transition period, as both Table 6 (panels C and D) and Annex III show. As noted, the role of stabilization is well established in the growth literature as perhaps the first virtually necessary condition for recovery, and hyperinflation conditions in many of the BRO in the early to mid-1990s were without question inimical to growth. A recent paper on disinflation in transition countries confirms this relationship and demonstrates further that while recovery may begin once inflation has declined to double digit levels, sustained growth requires sustained disinflation (Cottarelli and Doyle, 1999). But, although stabilization appears, with few exceptions,³¹ to be a necessary condition for achieving growth, it is not a sufficient condition; note in Table 6, panel D, that several of the “little or no growth” cases—Kazakhstan, Turkmenistan, Ukraine, and, until recently, Russia—also recently have had relatively low inflation rates. However one may view the “necessary but not sufficient” issue, it is clear that high levels of inflation have generally been detrimental to growth, and that stabilization, along with structural reforms, has helped growth to resume.

Indeed, accompanying structural reforms as reflected in the index constructed by the EBRD, were equally necessary to achieve growth (see Annex III). It is surely not purely coincidental that Poland, which was among the earliest to begin and sustain recovery, was also among the earliest to put in place reforms, which were reflected in an increase of the rescaled EBRD index from 0.68 in 1990 to 0.83 in 1994; growth in the Billies, beginning in 1994–95. was associated with improvements in their reform indices from around 0.30 in 1991 to between 0,70 and 0.85 in 1994; and the six CIS countries which have thus far achieved only little or no growth have an average reform index value well below most others (Tables 6 and 7).

Table 7.

Economic Performance of Selected Country Groups

¹For reversal cases, average during years of growth (in parentheses, average including reversal years).

²Average EBRD Index; for reversal cases it is the index in the last year of growth.

³Average EBRD Index for 1997.

Table 7.

Economic Performance of Selected Country Groups

	Growth Since Initial Recovery1	Annual Average Inflation During Growth	Average Index of Reforms During Growth2	Index of Reforms in I9973	Percent Output Decline Until Initial Recovery
Country groups
Consistent growth (excluding Belarus and Uzbekistan)	4.2	15	0.74	0.73	33.0
Belarus and Uzbekistan	5.0	46	0.48	0.45	26.0
Growth reversals	5.2 (1.4)	67	0.63	0.65	25.8
Little or no growth	0.5	37	0.61	0.56	49.6

¹For reversal cases, average during years of growth (in parentheses, average including reversal years).

²Average EBRD Index; for reversal cases it is the index in the last year of growth.

³Average EBRD Index for 1997.

View Table

Table 7.

Economic Performance of Selected Country Groups

	Growth Since Initial Recovery1	Annual Average Inflation During Growth	Average Index of Reforms During Growth2	Index of Reforms in I9973	Percent Output Decline Until Initial Recovery
Country groups
Consistent growth (excluding Belarus and Uzbekistan)	4.2	15	0.74	0.73	33.0
Belarus and Uzbekistan	5.0	46	0.48	0.45	26.0
Growth reversals	5.2 (1.4)	67	0.63	0.65	25.8
Little or no growth	0.5	37	0.61	0.56	49.6

¹For reversal cases, average during years of growth (in parentheses, average including reversal years).

²Average EBRD Index; for reversal cases it is the index in the last year of growth.

³Average EBRD Index for 1997.

It is also noteworthy that in 1997 the share of the private sector in (officially recorded) GDP tended to be higher in countries with likely sustained growth than for countries with growth reversals or little or no growth (Table 6, panel F).³² In early discussions of transition, privatization was seen as the key mechanism for achieving efficiency and new growth by enforcing a hard budget discipline and providing profit incentives. This would lead first to restructuring, enabling enterprises to reach commercial viability through a variety of measures: labor shedding, increases in sales, closure or sale of unprofitable units, creation of a retail network and marketing, development of new products, seeking contracts with new partners, renovations to improve productivity, a quality control process, and a reduction in barter transactions. Viable firms would increase their productivity and market share, attract more resources, and increase production—hence new growth. Nonviable firms would shrink or close down altogether; their assets (both human and capital) would be reallocated to alternative production.

After nearly a decade of privatization, one should ask two questions: first, has privatization given these expected results and, second, does the form of privatization (vouchers, auction, insider buyout, and so on) make a difference? The evidence, including that adduced in the literature, generally finds privatized firms perform better than state-owned ones. How much the form of privatization affects this is not clear. Many studies show insider privatization gives better results, but some find little or no difference (see Box 3). It is noteworthy that most studies are based on survey data not more recent than 1995, by which date little time had passed after privatization, and indeed many countries (especially in the CIS) were only beginning privatization. Given the limited time period of postprivatization experience analyzed in most of the studies, any conclusion is tentative, but the bulk of the research suggests that privatization is superior to continued state ownership, and³³ there is some evidence of a preferred ordering roughly as follows: de novo and foreign-partner-privatized firms tend to do better than those with local outsider or insider privatization; and within the latter, manager control may be superior to employee control; state-owned firms perform least well in the transition. The literature is only beginning to address the role of the accompanying or background policy environment—that is, hard budget³⁴ and open competition—demonstrating that the stronger these are, the better the performance after privatization. Still unanswered is the question whether, with the right policy environment, the initial mode of privatization matters.

The structural reform index is of course a composite of several dimensions. While some efforts have been made to isolate the subcomponents, it has generally not been possible to isolate one element as more important than others. Such a decomposition is, however, found in Fischer, Sahay, and Végh (1996 and 1998) and De Melo and others (1997). as well as in the background study to this paper summarized in Annex III. From the statistical analysis in Annex III, it can be seen that the subcomponents—price liberalization, privatization and enterprise reform, trade and exchange liberalization, and legal reforms—are all closely correlated to general reform. In a statistical test, none is important separately from the general index, although each of them alone “substitutes” for the general index as an explanatory factor of growth. Only price liberalization stands out as unique in its effects over time, as described in the discussion of the timing and pace of reforms below.³⁵

The role of an effective legal framework in promoting growth in transition economies has taken on increased importance, as many analysts of the process began to point to the practical barriers in exercising de jure freedom of economic action (see World Bank, 1996a; and EBRD, 1997). The finding of some empirical growth studies (see Section II) that the institutional “climate” of property rights matters is illustrated by the statistical results described in Annex III: the legal framework index (estimated by EBRD since 1995) is strongly associated with growth performance. The substantive meaning of “legal climate” is explained in Box 6. Recent papers (Havrylyshyn and van Rooden, 1999; and Kaufmann, Kraag, and Lobaton, 1999) show that indicators of institutional development—legal, political, and economic—are highly statistically significant in econometric analysis of growth in transition (and other) countries.

There are two groups of counterexamples to the conclusion that growth follows reforms: Belarus and Uzbekistan show strong positive growth despite limited reforms (rescaled EBRD index figures in 1997 of 0.37 and 0.54, respectively); and Albania. Bulgaria, and Romania achieved early recovery only to see it reversed. Box 5 and Table 7 suggest that the two groups are in fact similar and provide an object lesson about the fourth issue, the sustainability of growth. The three cases of reversal had achieved greater reform progress than the two “puzzle” cases, but financial stabilization was incomplete, and the elimination of soft budgets was generally not achieved. Albania, as a result of social upheaval triggered by its pyramid schemes, and Bulgaria and Romania, through continued directed credits and energy price supports to state enterprises, experienced a reversal of stabilization with annual inflation reaching an average of 67 percent and with growth reversing. Belarus and Uzbekistan, with limited reform, have seen a return to low-cost directed credits, which have fueled the recovery but also have given rise to a return of high inflation (about 50–60 percent in 1997).³⁶ This picture raises a concern about the sustainability of growth in these two cases.

Fourth, the timing and pace of reforms have perhaps been the most provocative issues in debate about growth in transition economies. Often, this has been reduced to a too-simplistic debate about shock therapy versus gradualism. It may be better to think of it in more nuanced terms—that is, whether an earlier start or more discrete sharp moves forward in those reforms that can be done quickly (for example, price liberalization) provide a better growth performance. Further, the choice is made more complex by the possibility that it is a complementary package of reforms rather than some “golden key” that works best. A separate issue is whether, once the conclusion is accepted that moving faster gives higher growth, the political will or a political environment exists enabling policymakers to choose to inflict pain up front, knowing that improvements will come only later. In a sense, it is still early to judge which approach has worked better, even with eight years of experience, since some countries may still show late but strong growth surges, while still others may see reversal (as was already witnessed in southeast Europe). Nevertheless, some illustration of the different effects of pace and intensity of reforms can be provided.

The Asian experience, especially of China, is often considered the best example of gradualism. Box 7 elaborates on these experiences, leading to more balanced conclusions. First, the process is better characterized as “selective” reform—especially in China—moving rapidly in one sector or region (agriculture, exports in coastal “enclaves”) during any given period. Second, the substantial initial difference of very low industrialization, with a very large surplus of labor in agriculture, puts into question the applicability of the Asian approach to most of the 25 transition countries in this study. In this latter group, there are clearly instances both of early or rapid reformers doing quite well (Poland, Hungary, Croatia; later Armenia and Georgia), and of slow reformers performing better than intermediate ones (Belarus and Uzbekistan recently; Bulgaria and Romania in the mid-1990s). As noted earlier, the last two slow reformers have experienced reversal; while the first two show similar signs of unsustainable growth, this is still to be tested in time. A comparison with the Asian experience is useful, however, to the extent that it underlines the importance of privatization and allowing markets to work in the agricultural sector. This provides an important lesson, in particular, for the CIS countries, many of which have a relatively sizable agricultural sector. Owing to a lack of significant progress in these areas, notably in countries such as Russia and Ukraine, but others as well, where government interference in the agricultural sector is still predominant and privatization of land has hardly progressed, large segments of the population have yet to reap the fruits of a transition process in their sector, in a way seen in China and other Asian transition countries.

A useful way of thinking about the time pattern of reforms and resulting output performance is the “creative destruction” framework developed in the second part of Section II. At the outset, rapid reforms (including price liberalization) have a greater destructive than creative effect (Hernández-Catá, 1997, also emphasizes this). Indeed, for the early period 1990–93, the regression analysis suggests that greater price liberalization results in more decline (Annex III: see negative coefficient of LIP in equation B2). The tables and charts of Section III tell a similar story in the U-shaped relationship between growth performance and the degree of progress in reforms in 1990–93 (Figure 5, top panel based on the fitted values from the regression analysis in Annex III). Were the analysis to stop at that point, the conclusion could be drawn—as it often has been—that less reform is better than intermediate reforms. Moreover, since most of those at the right end of the “U” in 1993 were central European economies, their performance has often been attributed to geography and history (that is, better initial conditions).

Impact of Reform on Growth: Fitted Regression Values¹

Sources: IMF staff estimates; and EBRD (1997).¹ The fitted values are obtained from the multivariate panel data regression B1 and C1 (see Annex III). A polynomial trend line has been added to the fitted regression values.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Impact of Reform on Growth: Fitted Regression Values¹

Sources: IMF staff estimates; and EBRD (1997).¹ The fitted values are obtained from the multivariate panel data regression B1 and C1 (see Annex III). A polynomial trend line has been added to the fitted regression values.

• Download Figure
• Download figure as PowerPoint slide

Impact of Reform on Growth: Fitted Regression Values¹

Sources: IMF staff estimates; and EBRD (1997).¹ The fitted values are obtained from the multivariate panel data regression B1 and C1 (see Annex III). A polynomial trend line has been added to the fitted regression values.

• Download Figure
• Download figure as PowerPoint slide

But with additional time to observe the process, one in fact sees very much what the analytical framework suggests: as reform progress continues, in the advanced group of countries the negative, destructive effects of reform come to be strongly outweighed by the positive, creative, and growth-inducing effects. In the bottom panel of Figure 5, for the recovery period, this is now reflected in a very clear positive relation between the degree of reform and growth performance. In the formal regression analysis, the price liberalization variable becomes strongly positive in the second period (see equation C2 in Annex III); indeed, the size of the coefficients for all policy variables tends to be much higher. Of course, a few exceptions remain where limited reform and recovery are associated—Belarus and Uzbekistan again—but the overwhelming part of the story is that many of those who started early (or started later but moved rapidly) may have paid an initial price in sharper output declines, but they are now reaping the benefits through higher positive growth. In contrast, those with limited or very moderate reform continue to suffer decline, or at best experience only nascent and fragile recoveries.

One can best summarize the implications of the above evidence by asking if there are some particular aspects of structural reforms that are more important determinants of growth, and that therefore deserve more attention from policymakers. The answer is a straightforward, but not simple, “no and yes.” The evidence described so far suggests, if anything, that what is needed is a package of across-the-board policy measures and that there is no magic single measure for growth success. In the words of Harberger (1998) in a recent article on growth, it is the “thousand and one” little things that hard-working and innovative managers do that create new output, A more formal demonstration of the need for a comprehensive package is found in Aziz and Wescott (1997), In the early studies of growth in transition, many different policy measures have been found to be important: stabilization; general progress on structural reforms (including price liberalization, privatization); reduction of the government budget deficit; a strong legal climate of transparency and fairness (or good rule of law); openness to external trade; and a high degree of competitiveness. Having said this, it is still possible to point to some areas of structural reform needing more attention, since the actual implementation of policy has not been even and across the board. In some instances, beyond financial stabilization and initial (often still partial) price or exchange rate liberalization (or both), structural reforms are not very far advanced (Ukraine), and even liberalization itself has been delayed in countries such as Belarus, Turkmenistan, and Uzbekistan.

For many transition countries, in both CEE and the BRO, privatization is well advanced, but the associated conditions of a hard budget constraint and open competition (see Box 3) are by no means fully in place—though admittedly the degree varies from the advanced transition cases such as Croatia, Hungary, Poland, Slovenia, and (nearly so) the Baltics; through moderately advanced cases such as Armenia, Kazakhstan, Kyrgyz Republic, and in some ways Russia; to the least advanced reformers previously noted. Finally, for almost all countries, but again to varying degree, the establishment and effective implementation of strong legal institutions supporting property rights is a key item of the future agenda. Some of the critically needed policy actions concern items that, by their nature, take rather more time to do (strengthening of legal institutions, privatization of natural monopolies, and the strengthening of banking systems and more generally the system of financial intermediation, in part through improving prudential and supervision mechanisms). Others are simply items that should have been but were not done earlier (delayed privatization, incomplete price liberalization, fiscal reforms), and perhaps a few are items whose “bottleneck” importance was not well enough understood earlier (simple business registration and entry regulations, agricultural land collateral systems, competition in supply and marketing networks related to agriculture, and effective and quick contract law settlement).

In addition to the proximate causes of growth discussed so far, it is useful to consider some political economy factors that, by affecting the nature and effective implementation of economic reforms, indirectly affect growth. At the start of transition, political support for transformation may have been quite widespread, but it was not by any means universal. Opposition of varying degree was to be found among the managerial elite of the Soviet period, the bureaucracy, and many but not all of the communist party elite. For most countries of CEE and the Baltics, the prospects of eventual accession to the EU acted as a strong beacon showing the way to market reforms. There, as well as in a few cases of early reformers farther east—Armenia and the Kyrgyz Republic—strong and continued progress on reforms created a demonstration effect of potential benefits: service sector growth, new opportunities for small entrepreneurs, and eventually general recovery. Of course, an early start on comprehensive reforms also created opposition to further reforms, particularly because of unemployment and the removal of privileges. But in many of these countries, the initial goodwill to reform was enough to carry reforms politically until the new beneficiaries added their voices to the support for continued reforms. It is also notable that, while a wave of elections in the mid-1990s in central Europe brought many left-oriented governments to power, the main direction of economic policy was little changed.

In countries where reforms did not have an early start, or were very partial—for example, considerable privatization, but continued soft budgets and inadequate development of rule-of-law institutions—the policy process was captured by new vested interests, especially in the energy, banking, and heavy-industry sectors. These vested interests often began by accumulating wealth through rent-seeking profits from large price distortions in energy and raw materials, and by borrowing from central bank credits during inflationary years. Mass privatization was often captured by such insider vested interests, who understandably favored such a form of privatization. Once their private ownership of former state assets was assured, they would support inflation control because this could help to stimulate recovery and the growth of the newly owned firms, but would strongly oppose the full liberalization of market opportunities, which would bring new domestic competitors or foreign competition. With transformation to the market thus frozen in mid-process, the aims of the new vested interests, those of the old surviving state enterprises, and a political establishment concerned about employment broadly coincided in a status quo. One manifestation of such a political economy equilibrium is the so-called virtual economy (see in particular Gaddy and Ickes. 1998) where noncash operations—barter, arrears, in-kind wages, tax offsets—help to maintain the status quo and provide a cover of nontransparency for economic operations that are of low or even negative productivity, are not market based, and allow for rent seeking and corruption. A potential virtuous circle of reforms and growth is replaced by a vicious circle of suspended reforms and stagnation.

Role of IMF-Supported Programs in the Transition

In some countries of CEE. and virtually all of the BRO,³⁷ the IMF’s Systemic Transformation Facility (STF)—which involved relatively weak conditionality but was accompanied by a large amount of policy advice and technical assistance—provided the IMF with considerable influence on the nascent stabilization and reform efforts of recipient countries. As the transition progressed, this initial provision of policy advice(in many cases, to officials who had little or no previous exposure to market-type economics) evolved into an ongoing policy dialogue aimed at achieving and then consolidating macroeconomic stabilization, and accelerating structural reform. With respect to the latter area, IMF staff efforts, along with those of staff of the World Bank, in the transition economies in many respects broke new ground, both in advising on why structural reforms were essential and how they could be carried out.

The IMF’s technical assistance program for the transition economies has played a key role in helping the authorities adapt their monetary, foreign exchange, fiscal, and statistical systems to the requirements of the market economy. This effort, which absorbed 40 percent of the IMF’s technical assistance resources as recently as 1995,³⁸ has also been complemented by the training programs of the IMF Institute as well as the establishment of the Joint Vienna Institute in 1993. The main areas of technical assistance, carried out mainly by the IMF’s Fiscal Affairs Department, Legal Department, Monetary and Exchange Affairs Department, and Statistics Department, have been in tax policy; tax administration; establishment of treasury operations; budget and expenditure control; monetary and foreign exchange operations; central bank accounting; bank supervision; and improvements in monetary, fiscal, balance of payments, price, and national income statistics. Training and technical assistance have also contributed to help institution building as well as to improve transparency and general good governance.

All but two of the transition countries analyzed here have had IMF-supported programs (generally more than one); Slovenia and Turkmenistan, at opposite ends of the reform and performance spectrum, are the exceptions. The scope of IMF financial support to the transition economies—in the higher credit tranches—is summarized in Table 16 in Annex V. In addition to the SDR 24.6 billion made available during 1990-97 to 23 transition countries under Stand-By Arrangements and Extended Fund Facility programs. SDR 4.0 billion was provided under the STF facility, SDR 0.5 billion was provided under the Enhanced Structural Adjustment Facility (ESAF), and SDR 0.1 billion under other facilities.³⁹ Analysis of average cumulative purchases from the IMF, in terms of quota, for groups of transition countries (Table 8) suggests the superior initial conditions of the CEE countries, and consequently the cumulative smaller purchases required in terms of quota, compared with the BRO.

Table 8.

Average Cumulative Purchase from IMF as Percent of IMF Quota

¹SEE. Southeast Europe.

Table 8.

Average Cumulative Purchase from IMF as Percent of IMF Quota

	Central Europe	Baltics	SEE1	CIS
Consistent growth	61.1	137.5	51.6	9.9
Growth reversals			168.4
Little or no growth				122.3

¹SEE. Southeast Europe.

View Table

Table 8.

Average Cumulative Purchase from IMF as Percent of IMF Quota

	Central Europe	Baltics	SEE1	CIS
Consistent growth	61.1	137.5	51.6	9.9
Growth reversals			168.4
Little or no growth				122.3

¹SEE. Southeast Europe.

An analysis of the IMF-supported programs with the transition countries (in terms of program design and implementation)⁴⁰ suggests that program design has generally differed very little across countries with respect to macroeconomic performance,⁴¹ but somewhat more for structural reform elements; and that one finds a strong statistical association between relatively successful implementation of IMF-supported programs with respect to macroeconomic performance criteria, and the achievement of sustained stabilization and economic recovery.

Table 9 shows values for an index of implementation of performance criteria, compiled using the IMF’s Monitoring of Fund Arrangements database (MONA),⁴² Despite an apparent upward bias in this index for countries not performing well under IMF-supported programs,⁴³ the values are higher for countries with better growth performance, particularly those in CEE and the Baltics. But the interpretation of this statistical association should be made with caution. This positive correlation could probably best viewed as reflecting the high level of the authorities’ policy commitment—typically to both macroeconomic stabilization and reform—in those countries that have managed to achieve significant disinflation and the beginning of sustained economic growth. Annex III also shows a positive statistical relation between growth and the index of implementation using a broader regression analysis framework.

Table 9.

Index of IMF Program Implementation, 1993–97¹

(Percent; by region and growth performance)

¹Index ranges from 0 to 100 percent, where 100 percent implies that the country complied with all performance criteria under all IMF programs without modifications or waivers.

The Index of Implementation (∥) is defined for each country and each program (sum for all test dates). If there are n performance criteria in each test date, and T test dates in each program, define the index of implementation (∥) as follows:

$//=\frac{\underset{t=1}{\overset{T}{\mathrm{\Sigma }}}\underset{t=1}{\overset{n}{\mathrm{\Sigma }}}p{c}_{it}}{10T_n}$

where pc_it t refers to the result value for performance criteria i in test date t and takes values determined by compliance as follows: met = 10; waived = 5; met after modification = 5; waived after modification = 3; not met after modification = 0; not met = 0. Note that ∥ should have a value of 0 in the (presumably abstract) case in which all performance criteria at all test dates are not met—or not met after modification.

²No program as of end-1997.

Table 9.

Index of IMF Program Implementation, 1993–97¹

(Percent; by region and growth performance)

	Central Europe	Baltics	Southeast Europe	CIS
Consistent growth	Croatia (80) Czech Rep.(96) Hungary (98.1) Poland (96.4) Slovak Rep.(N/A) Slovenia2	Estonia (100) Latvia (93.1) Lithuania (92.6)	Macedonia (83.5)	Armenia (94.4) Aierbajan (93.4) Belarus (62.2) Georgia (94.6) Kyrgyi Rep.(78) Uzbekistan (70.6)
	Average (92.6)	Average (95.2)		Average (82.2)
Growth reversals			Albania (90.9) Bulgaria (50) Romania (76.7)
		Average (72.5)
Little or no growth			Kazakhstan (75.4) Moldova (89.5) Russia (88.5) Tajikistan (N/A) Turkmenistan2 Ukraine (69.1)
			Average (80.6)

¹Index ranges from 0 to 100 percent, where 100 percent implies that the country complied with all performance criteria under all IMF programs without modifications or waivers.

The Index of Implementation (∥) is defined for each country and each program (sum for all test dates). If there are n performance criteria in each test date, and T test dates in each program, define the index of implementation (∥) as follows:

$//=\frac{\underset{t=1}{\overset{T}{\mathrm{\Sigma }}}\underset{t=1}{\overset{n}{\mathrm{\Sigma }}}p{c}_{it}}{10T_n}$

where pc_it t refers to the result value for performance criteria i in test date t and takes values determined by compliance as follows: met = 10; waived = 5; met after modification = 5; waived after modification = 3; not met after modification = 0; not met = 0. Note that ∥ should have a value of 0 in the (presumably abstract) case in which all performance criteria at all test dates are not met—or not met after modification.

²No program as of end-1997.

View Table

Table 9.

Index of IMF Program Implementation, 1993–97¹

(Percent; by region and growth performance)

	Central Europe	Baltics	Southeast Europe	CIS
Consistent growth	Croatia (80) Czech Rep.(96) Hungary (98.1) Poland (96.4) Slovak Rep.(N/A) Slovenia2	Estonia (100) Latvia (93.1) Lithuania (92.6)	Macedonia (83.5)	Armenia (94.4) Aierbajan (93.4) Belarus (62.2) Georgia (94.6) Kyrgyi Rep.(78) Uzbekistan (70.6)
	Average (92.6)	Average (95.2)		Average (82.2)
Growth reversals			Albania (90.9) Bulgaria (50) Romania (76.7)
		Average (72.5)
Little or no growth			Kazakhstan (75.4) Moldova (89.5) Russia (88.5) Tajikistan (N/A) Turkmenistan2 Ukraine (69.1)
			Average (80.6)

¹Index ranges from 0 to 100 percent, where 100 percent implies that the country complied with all performance criteria under all IMF programs without modifications or waivers.

The Index of Implementation (∥) is defined for each country and each program (sum for all test dates). If there are n performance criteria in each test date, and T test dates in each program, define the index of implementation (∥) as follows:

$//=\frac{\underset{t=1}{\overset{T}{\mathrm{\Sigma }}}\underset{t=1}{\overset{n}{\mathrm{\Sigma }}}p{c}_{it}}{10T_n}$

where pc_it t refers to the result value for performance criteria i in test date t and takes values determined by compliance as follows: met = 10; waived = 5; met after modification = 5; waived after modification = 3; not met after modification = 0; not met = 0. Note that ∥ should have a value of 0 in the (presumably abstract) case in which all performance criteria at all test dates are not met—or not met after modification.

²No program as of end-1997.

Unfortunately, the data on implementation of structural reforms were insufficient to construct a simple statistical test. An overview of structural benchmarks, by group of transition countries and by type of action, however, is provided in Table 10. There is a very broad similarity in two respects: all countries appear to have been subject to a large number of structural benchmarks (see Table 17 in Annex V for details); and the distribution by policy area is roughly the same, although programs with countries that have achieved consistent growth have had a larger share of fiscal reform benchmarks.

Table 10.

Structural Benchmarks in IMF Programs

(Percent)

Source: IMF, MONA database.

¹Numbers indicate the following structural benchmark groups: I, trade/exchange systems; 2, pricing and marketing; 3, public enterprise; 4, tax/expenditure reform; 5, financial sector; 6, systemic/ownership reform; and 7, other.

Table 10.

Structural Benchmarks in IMF Programs

(Percent)

Country Group	Structural Bench marks1							Total	Average Number of Benchmarks per Program
	1	2	3	4	5	6	7
Consistent growth	7	8	7	29	18	15	16	100	15
Growth reversals	8	5	15	16	23	17	17	100	16
Little or no growth	15	6	5	24	16	24	9	100	12

Source: IMF, MONA database.

¹Numbers indicate the following structural benchmark groups: I, trade/exchange systems; 2, pricing and marketing; 3, public enterprise; 4, tax/expenditure reform; 5, financial sector; 6, systemic/ownership reform; and 7, other.

View Table

Table 10.

Structural Benchmarks in IMF Programs

(Percent)

Country Group	Structural Bench marks1							Total	Average Number of Benchmarks per Program
	1	2	3	4	5	6	7
Consistent growth	7	8	7	29	18	15	16	100	15
Growth reversals	8	5	15	16	23	17	17	100	16
Little or no growth	15	6	5	24	16	24	9	100	12

Source: IMF, MONA database.

¹Numbers indicate the following structural benchmark groups: I, trade/exchange systems; 2, pricing and marketing; 3, public enterprise; 4, tax/expenditure reform; 5, financial sector; 6, systemic/ownership reform; and 7, other.

While the design and implementation of programs supported by IMF arrangements provided the framework for the IMF’s policy advice, the actual delivery of such advice involved intensive contacts between the IMF staff and the authorities, probably to a greater extent than in other parts of the world. First, resident representatives were in place in all countries that made use of IMF resources, and also in some, but not all, countries before or after using IMF resources. Second, staff missions were frequent, at least quarterly to most countries with IMF arrangements (including ESAF countries, where formal reviews by the IMF’s Executive Board were only biannual). The teams working on transition economies also tended to be somewhat larger than in other parts of the world. Although this partly reflected the difficulty of the task (poor data, need to use interpreters in many countries, and so on), it also permitted a more intensive dialogue. As a result, the IMF staff, especially the resident representatives, have often been closely involved in day-today policy advice in transition countries. This relatively intensive effort has allowed the IMF staff to help the authorities keep programs on track to a greater extent than would have been possible with less contact.

While the IMF has been ready to make a strong effort in all transition countries, especially the largest ones, its impact has not been the same across all countries. In general, the impact has been greatest in countries that were most willing to pursue policies of macroeconomic stabilization and structural reform; and that were open to advice from international financial institutions (especially likely in smaller countries that sought to build up quickly links with the international economy to replace links of the former CMEA and the former Soviet Union). In the countries least prepared to reform, the IMF has naturally not had a major impact, although, even there, policymaking is probably better as a result of the dialogue with the IMF, especially in the area of macroeconomic stabilization. Similarly, in countries least ready to accept advice from international financial institutions, there are many signs that the policies being pursued have benefited from the dialogue with the IMF. In the other countries, which form the majority of transition countries, the impact of the IMF’s policy advice has been an important factor in the successful—sometimes faster, sometimes slower—progress they have made toward market economies and sustainable growth.