NEVER AROUND NOON: ON THE NATURE AND CAUSES OF THE TRANSITION SHADOW
Nauro F. CAMPOS * CERGE-EI, Prague, Centre for Economic Policy Research (CEPR), London and William Davidson Institute at the University of Michigan.
This version: February 2000 DRAFT: Please do no quote or cite without the author’s permission.
Abstract. The emerging conventional wisdom is that the principal cause of the informal sector in transition economies is the same as in developing and developed countries: excessive government interference. This paper challenges this wisdom, firstly, by articulating an alternative explanation: it claims instead that the main cause is instead the institutional vacuum left after the fall of socialism. Secondly, it presents econometric evidence that endorses the “institutional vacuum” as a more promising explanation.
Keywords: informal sector, transition economies. JEL classification: E23, H26, O17, O41, P20. * CERGE-EI, P.O. Box 882, Politických veznu 7, 111 21 Prague 1, Czech Republic. E-mail:
[email protected]
I thank Randall Filer, Byeongju Jeong, Dean Jolliffe, Inna Piven, Gerard Roland and seminar participants at the 9th Silvaplana Workshop on Political Economy (Switzerland) for comments on an earlier version. Aurelijus Dabusinskas and Rodica Cnobloch provided alacritous research assistance. The usual disclaimer applies.
1
I. INTRODUCTION
You don't know life. No one lives on wages alone. I remember in my youth we earned money by unloading railroad freight cars. So, what did we do? Three crates or bags unloaded and one for ourselves. That is how everybody lives in [our] country. Leonid Brezhnev, quoted in Kornai (1992)
The 1980s were the “lost decade” for Latin America and the Brazilian economy struggled. In all those summers, the Largo da Carioca, the main square in downtown Rio, was packed with a loud multitude of vendors. Forty-two degrees Celsius, and ten minutes to walk a block. You could buy anything on the street: from tomatoes to stereos, from pet parrots to life insurance. The best estimates put the size of the informal sector1 at slightly below 30% of the Brazilian GDP. The 1990s have been tough for the Ukraine: it is the only transition economy to post negative growth rates throughout. In a Fall in the late 1990s, the temperature was minus twelve degrees Celsius in the Khreschatik, Kiev’s main street. It also took ten minutes to walk a block. Because of snow not a loud multitude of vendors. To be sure, there were two people selling window98 CDs for USD 3.50 and eight babushkas selling cigarettes and vodka. Yet, the best estimates put the size of the Ukrainian shadow economy at slightly below 50% of GDP.2 How can a third of the Brazilian GDP be so conspicuous and half of the Ukrainian GDP so concealed? What can explain such differences in the workings of the informal sector? Why is the informal sector in transition so large, even when compared with one of the largest in the developing world? How did it grow so fast?
In this paper, “informal”, “underground”, “parallel”, “unofficial” and “shadow” economy are used interchangeably. See OECD (1997) for operational definitions. 2 The best available estimates for Brazil and Ukraine are from Enste and Schneider (1998). 1
2
Are the causes of the informal sector different in transition vis-à-vis developing countries? These are the questions that motivate this paper. The objective of this paper is to debate the prematurely emerging and benevolent conventional wisdom on the informal sector in the former centrally planned economies. This wisdom can be broken down in two elements: one is the claim that the cause of the “shadow” is the same in transition as in developing and developed countries. The second is that this cause is excessive government interference.3 An efficient way to bring debate into being is to assert that government interference is not the cause of the “transition shadow.” Why is excessive government interference not the cause? Because the period in which the transition shadow exploded in size coincides with the dismantling of state control of economic activity and the emergence of a state without a comparative advantage in delivering regulation (Roland and Verdier, 1999). It may well be that government interference was ineffective, capricious or misguided, but it is very unclear it was excessive. In addition, government interference as the cause squares badly with the evidence on a negative relationship between the share of the “transition shadow” in GDP and one measure of government interference.4 Is there any alternative explanation? This paper offers that the ultimate cause of the “transition shadow” is, instead, the institutional vacuum left after the fall of socialism. The first qualification is that the conventional wisdom is prematurely emerging. Although it at times recognizes that the nature of the transition shadow is unique, this wisdom overlooks the implications. This nature is unique because, while
Johnson, Kaufmann and Zoido-Lobaton (1998) claim that the cause of the informal sector is “political control of economic activity,” and it is the same everywhere, so they compare Brazil with Russia. Johnson, Kaufmann and Shleifer equate “political control of economic activity” to “excessive regulations and taxes” (1997, p. 161). 4 Johnson, McMillan and Woodruff discuss evidence that corporate taxes in Russia and Ukraine are lower than in Poland, Slovakia and Hungary (1999, pp. 31-36). As it can be seen from Table 1, Russia and Ukraine have much larger shadow economies as a share of GDP. 3
3
low capital intensity is a defining characteristic of informal sector activities in developed and developing countries, those in the transition shadow are characterized by high capital intensity. This makes it harder to accept that the causes and consequences of the transition shadow are the same as elsewhere. How does high capital intensity relate to the institutional vacuum? Institutional vacuum means that the state has lost political power or that the interests of fractions of society have been able to capture political initiative. There are few more obvious demonstrations of political power than the ability to engage in high capital intensity activities, report zero output, and receive no punishment.5 The second qualification is that the conventional wisdom is benevolent. In transition countries, government agencies rapidly accounted for the informal sector in their revised output figures. Policymakers equate it with the de novo private sector, dismissing claims that transition countries lack entrepreneurial talent. Academicians argue that the driving forces behind the explosion in size of the transition shadow are a legacy of the past (government interference), nourishing expectations that the shadow will shrink as transition inevitably progresses. As a result of all interested parties embracing such views, the costs of the “transition shadow” are downplayed and the benefits highlighted. As this paper’s title insinuates, it is not wise to measure the costs of the “shadow” at noon and its benefits at sunset. The paper is organized as follows. The next section establishes the different natures of the developing and transition shadows, disentangle their causes, and infer the sets of costs and benefits. It makes the case against excessive government interference and in favor of the institutional vacuum as the ultimate cause of the
Oil refining is an activity in which capital intensity tends to be high. The December 9, 1998 issue of the ??????????? Daily (“Kommerzant Daily,” Moscow) reports that the Ministry of Interior of the Russian Federation launched “Operation Oil,” aiming at combating widespread informal oil refining. 5
4
transition shadow. Section III presents the data and discusses difficulties in testing the two competing hypotheses (namely, “government interference” and “institutional vacuum”). The econometric evidence, presented in Section IV, does not support the conventional wisdom and broadly endorses the “institutional vacuum” as a promising alternative explanation. Section V summarizes and concludes.
II. WHEN SHADOWS MEET This section attempts to answer the following questions, first for developing and developed countries and then for the transition economies. What are the defining characteristics of the informal sector? What are its causes? And what are its benefits and costs? If these answers turn out to be different for transition than for developing and developed countries, the nature of the transition shadow is said to be unique. The difficulties in defining and measuring the informal sector are exacerbated when the objectives are international comparisons.6 Using data from seven developing countries,7 Mead and Morrisson (1996) find just three features recurring in the developing shadow: legality, firm size, and capital intensity. Evidence for developed (Nicolini, 1998) and developing countries (Mead and Morrisson, 1996) indicate that informal activities are carried out by small firms, concentrated on petty trade, services and small-scale manufacturing, and are characterized by their low capital intensity.8 The sector tends to employ recent migrants, with little formal education, without marketable (or urban) skills, in unregulated labor market
The vast literature on the informal sector in development economics emanates from the Harris and Todaro (1970) model. It emphasizes labor reallocation through rural-to-urban migration decisions and include, among others, Loayza (1996), Marcouiller and Young (1995), Mead and Morrisson (1996), Patel and Srivastava (1996), Rauch (1991), and Stark (1982). Enste and Schneider (1998) provide a most extensive and up-dated survey of the literature. 7 Algeria, Ecuador, Jamaica, Niger, Swaziland, Thailand, and Tunisia. 8 According to Mead and Morrisson, “in terms of fixed capital, measure either in absolute terms or by capital/labor ratios and reflecting different levels of mechanization or modernization ” (1996, p. 1612). 6
5
conditions (that translate into, e.g., the absence of job security and pensions) and attracts a disproportionate number of women.9 There is, however, broad consensus on the causes of the informal sector: “(i) The burden of direct and indirect taxation, both actual and perceived (…), (ii) The burden of regulation as proxy for all other state activities. (…) (iii) The ‘tax morality’ (citizen’s attitudes towards the state) which describes the readiness of individuals (at least partly) to leave their official occupations and enter the shadow economy” (Enste and Schneider, 1998, p. 15).10 The developing countries’ experience, and the different readings of it, has clarified the costs and the benefits of the informal sector. As for the costs, by not contributing to government revenues, a large shadow economy constrains the provision of public goods, reduces the government’s ability to regulate the economy, and misconstrues the basis upon which economic policy is formulated. A buoyant informal labor market fosters migration, increasing congestion over an often inadequate infrastructure (Loyaza, 1996). On the other hand, among the potential benefits from the informal sector the literature highlights the fact that it uses a technology that is more “appropriate” (i.e., labor intensive) for conditions in developing countries, thereby generating employment (in particular, by absorbing unskilled workers) and contributing to ameliorate urban poverty. It is sometimes
Analyses of the informal sector are seldom concerned with those activities that are clearly illegal, such as drug trafficking and prostitution because “illegal activities contribute a small percentage of the income generated in the underground sector” (Nicolini , 1998, p. 217). 10 The current consensus is that high levels of informality result from government-induced distortions. Yet Levenson and Maloney (1998) model informality as a decision not to participate in societal institutions. Hence, in the future, hypotheses should be formulated in terms of which component of the cause is more important: government distortions or institutional failure. This is a trivial issue only if government distortions are directly related to institutional failure. The new institutional economics has emphasized that the relationship between policies and institutions is more intricate than previously thought (Nugent, 1998, Campos, Khan and Tessendorf, 1999). See Åslund (1999) for an analysis of the transition failure in Russia in terms of the relationship between policies and institutions (rapid elimination of government-induced distortions followed by inattention to institutional reform). 9
6
thought of as providing some basic training, for example, with respect to “urban skills” (Mead and Morrisson, 1996). While the evidence for developed and developing countries establishes that informal activities are characterized by low capital intensity, in socialist countries they were concentrated in manufacturing (Dallago, 1987).11 Although, at the time of writing, there is no direct evidence on which sectors the transition shadow concentrate, anecdotal evidence is abundant and suggests that capital intensity is much higher in the transition than in the developing shadow.12 If this is indeed the defining characteristic of the transition shadow, what are its causes? And how different are its benefits and costs? Can they explain the extraordinary growth of the shadow in the early years of the transition? The argument in this paper is that the ultimate cause of the “transition shadow” is the institutional vacuum left after the fall of socialism (instead of government interference), while its proximate cause is macroeconomic instability. The effects of the institutional vacuum are compounded by the long absence of a rule of law tradition: under socialism, there was no law (but the Plan), there was no tradition (but the Party), and there was no rule, just discretion.13 The rule of law is meant to provide well-defined property rights (Lin and Nugent, 1995). Tradition matters because in order to ensure that property rights are protected, the rule of law must be applied consistently.14
See also Landau (1987) and Mars and Altman (1987). It certainly includes the previously mentioned December 9, 1998 issue of the ??????????? Daily (“Kommerzant Daily,” Moscow) which reports on the Ministry of Interior of the Russian Federation launching “Operation Oil” to combat “widespread informal oil refining.” 13 Various analysts, among them Dewatripont and Roland (1997), Ickes (1997), Kornai (1992) and Pistor (1999), note the repercussions of the long absence of a rule of law tradition. 14 A counter-argument is that it is just too soon to expect tradition in transition. A reply would be that tradition there is, but is one of discretion and inconsistency. This legacy of unchecked discretion and constant reform thwarts credible commitments by post-socialist governments (Murphy, Shleifer and Vishny, 1992). 11 12
7
In order to assess the relevance of these considerations about the institutional vacuum consider the costs and benefits of operating “informally” in a transition vis-àvis in a developing or developed country. The costs are not only the probability of being caught and the punishment (which for the sake of argument are assumed to be the same), but are also those of setting-up operation.15 Insider privatization, unbundling and asset stripping practices (Blanchard, 1997) may have contributed in reducing the latter.16 Even if they did not, these practices have certainly increased the benefits, as they allowed the transition shadow to operate under technological conditions that are unmet in their developing and developed counterparts (for example, by exploiting economies of scale). The institutional vacuum has not only helped shrinking the costs, but it has also broadened the pecuniary (private) benefits of operating “informally” in the early stages of the transition. The difference between excessive government interference and the institutional vacuum as the cause of the transition shadow can now be appreciated. If government interference is the main cause then tax cuts and fine-tuning of regulations would help contain the transition shadow. If the absence of a rule of law tradition is the main cause, disregard for institutional reform early on would help explain the explosion in size of the transition shadow. Note that this exceptional growth further undermines the rule of law with damaging consequences for economic performance. If this is correct, the transition shadow is not shallow,17 it is deeply entrenched and its effects on aggregate economic performance are to be felt over the long run.
The development literature justifiably downplays this issue. It can be ignored when the activity in question is “to sell oranges in a traffic light.” It should not when “refining oil” is. 16 Stiglitz (1999) refers to these practices as “looting.” 17 Among others, Kaufmann and Kaliberda claim that “Since unofficial activities operate in very close proximity to official activities (often within a continuum in the same line of business), and since they respond to economic incentives largely driven by government policies, the unofficial economy is mostly of a shallow nature. Except for the entrenched, hardcore mafia, activities hover flexibly between officialdom and unofficialdom, largely motivated by government-induced incentives” (1996, p.4). 15
8
How does recognizing high capital intensity as the defining characteristic change the conclusions from the theoretical and empirical literatures on the transition shadow? Let first look at the implications for the theoretical literature, which includes Commander and Tolstopiatenko (1998) and Vostroknoutova (1998). The former studies a two-sector model in which transition is driven by labor shifts across sectors. The formal sector is thought of as composed by state firms, while the informal sector is equated to the de novo private sector. Tax evasion is the crucial differentiating characteristic in this set-up: while the informal sector does not pay taxes at all, the state sector pays taxes only for a fraction of the labor force it employs. Vostroknoutova puts forward a model in which our main argument is echoed: “inadequate institutional structures create space for the expansion of the informal sector” (1998, p. 3). This is a dynamic model in which the representative firm produces in the two sectors (official and unofficial) subject to a set of costs that include those of concealing output (or evading taxes). An interesting feature is that the “capacity of the government to tax” depreciates in standard fashion. In light of the previous discussion, these models share two limitations. The first is the absence of capital, which precludes the analysis of the consequences of differences in capital intensity, for instance, in terms of the benefits and costs entrepreneurs face. The second important limitation is that they depend on tax rates when, as noted before, the available evidence indicates that the relation between the share of the transition shadow in GDP and tax rates is negative. What are the implications for the empirical literature (of recognizing high capital intensity as the defining characteristic of the transition shadow)? The classic empirical analysis of the transition shadow is the paper by Johnson, Kaufmann and
9
Shleifer (1997).18 It provides a theoretical model (without capital and one in which high taxes are what drive firms to the informal sector), it measures the share of the transition shadow in GDP, and it explains the shadow share in terms of “political control of economic activity.” The latter corresponds to “excessive government interference” because these authors define “political control of economic activity” as highly distortionary “tax and regulatory policies” (p. 209) and because they explicitly differentiate
between
“depoliticization
and
institutional
building”
(p.207).19
Recognizing high capital intensity as the defining characteristic of the transition shadow would solve one of the largest puzzle Johnson, Kaufmann and Shleifer encounter, “the little difference in performance between the official and unofficial sectors” (1997, p. 236). This is puzzling because their model assumes that the shadow is less efficient than the official sector, which may not be the case if both sectors can exploit a technology that is capital intensive (or if both sectors use similar technologies). Finally, the case should be made that although the literature does not recognize high capital intensity as the defining characteristic of the transition shadow, it could. For instance, Johnson, McMillan and Woodruff (1999) consider four hypotheses to explain the transition shadow without noting that most of these are
Although there are very few empirical studies of the transition shadow in comparative perspective, there are a number for individual transition countries. See, for example, Kaufmann and Kaliberda (1996) for the Ukraine, Gaddy and Ickes (1998a, 1998b) and Kolev (1998) for Russia, Bedi (1998) for Poland, and Anderson (1998) for Mongolia. 19 Johnson, Kaufmann and Shleifer acknowledge that they can not distinguish empirically between a model driven by depoliticization from the one driven by institutional reform (1997, p.181). Moreover, the variables chosen to reflect depoliticization seem to be better proxies for the (lack of) institutional reform. For instance, they regress on the share of unofficial economy, “measures of state control,” among which “tax fairness,” “crime and corruption,” and “regulation” (pg. 189), and also “legal environment indicators,” including “rule of law,” “legal effectiveness,” and “legal extensiveness” (p. 193). At first sight, these variables are perfect to implement the hypothesis favored in this paper (“institutional vacuum”). Yet, the reason for not doing so is that they are available only for 1995-1997. 18
10
idiosyncratic.20 These are tested on survey data from about 1,500 manufacturing firms. The conclusion they reach is that managers in Russia and Ukraine face higher effective tax rates, worse official corruption, greater incidence of Mafia protection and have less faith in the court system than managers in Poland, Slovakia and Romania. Clearly, the data seem to allow the identification neither of the most important cause nor of the relationships among them. This outcome could be different had they differentiated firms by level of capital intensity. This could be done because, for instance, close to half of the Romanian firms answering their questionnaire (Appendix Table 1) have more than 250 employees and are concentrated in “metal parts and construction materials”.
III. DATA, HYPOTHESES, AND METHODOLOGY This section describes the data set and the methodology used in this paper. An important constraint to investigate the causes of the informal sector in transition economies is the availability of time series data on the share of the shadow economy in GDP. Although estimates for a single country exist (but tend not to be comparable), the best panel of consistent estimates is the one by Johnson, Kaufmann and Shleifer (1997). These figures are constructed for the period 1989-1995,21 and for the following transition economies: Czech Republic, Hungary, Poland and Slovakia (the “Visegrad” group), Bulgaria, Romania and Moldova
(the “Balkan” group),
Estonia, Latvia and Lithuania (the “Baltics” group), Belarus, Ukraine and Russia
Because only one out of the four is in the Enste and Schneider (1998) set of causes discussed above. The four hypotheses Johnson, McMillan and Woodruff (1999) consider are (a) high taxes and onerous regulations, (b) predatory behavior (bureaucratic corruption), (c) extortion by criminal gangs, and (d) inadequacy of the institutional environment (with emphasis on contract enforcement). 21 The reason the empirical analysis ends in 1995 is that governments, after this date, have started to incorporate the shadow economy into their National Product estimates. I thank Daniel Kaufmann (personal communication) for this point. 20
11
(the “BUR” group), and Georgia, Azerbaijan, Kazakhstan and Uzbekistan (the “Asia” group).22 Measuring the hidden economy is (understandably) difficult, and there are a number of methods that have been tried.23 Johnson, Kaufmann and Shleifer (1997) correctly choose the “macro-electric approach,” which is based on total electricity consumption as an overall measure of economic activity. Correctly because this approach is very appropriate for an informal sector characterized by high capital intensity. Based on an “electricity-to-GDP” elasticity of 1,24 and assuming a constant ratio of electricity consumption to GDP, the difference between the resulting overall level of activity and official GDP growth figures provides the measure of the growth of the unofficial economy.25 Table 1 shows these estimates and simple averages for the Commonwealth of Independent States (CIS) and for the Central and Eastern European and Baltic (CEEB) countries. The average share across these 17 countries over 1989-1995 is 25.3 percent of GDP, which is well above the share of the unofficial economy in OECD countries.26 These values range from 5.8 (for Slovakia in 1995) to 63.5 percent (Georgia in 1994), implying wide disparities in terms of initial shares and speeds of growth in the first years of the transition. Finally, the data show that while the
The groups referred to in parenthesis are arbitrary and used solely for simplifying the presentation in the Figures below. 23 Enste and Schneider (1998) discuss nine different methods. They report (Table 3.4.1, p. 26) that, for five countries in which all nine methods were used, the estimates from the “macroelectric approach” (which they call “physical input (electricity) approach”) are the closest to the average (over the results of the nine methods). 24 Johnson, Kaufmann and Shleifer differentiate among transition countries according to hypothesized elasticities, as “energy efficient” (.95), “energy neutral” (1) and “energy inefficient” (1.15). The Central European countries are in the first category, the Baltics in the middle, and the other former Soviet Union countries in the latter. Changes in these values do not affect the results in any way they deem meaningful. 25 Various individual country studies (quoted in Johnson, Kaufmann and Shleifer, 1997, esp. p. 175), provide the initial estimate for this share in 1989. 26 In the late 1970s, these shares in the OECD countries range from 4 to 13 percent of GDP (Frey and Weck-Haneman, 1984). In the early 1990s, Enste and Schneider (1998) estimate this share in the OECD countries to be around 15.4%. 22
12
“shadow share” in the CIS countries grew steadily between 1989 and 1995, in the CEEB countries it reached its peak in 1992, and has declined since. Although severe measurement problems make the argument of “too high a shadow share” in transition vis-à-vis developing countries difficult to substantiate,27 the claim of high speeds of growth in comparative perspective seems more difficult to ignore. Figure 1 shows group averages weighted by real GDP per capita, while Figure 2 shows these same group averages weighted by population.28 Note the extraordinarily rapidly growing shadow share for three groups (“Asia,” “BUR” and “Balkan”). For the “Visegrad,” “Baltic,” and “Asia” groups, the shadow share reaches its peak in 1992 (in 1994 for “Asia”) and declines since. Before examining the variables chosen to reflect the hypotheses articulated in the previous section, a caveat about data quality and comparability needs to be raised. These data problems are well known and have been discussed in detail by, for example, Bartholdy (1997). Since the focus is on international comparisons, it was decided to not use national sources, in the belief that effort has been put in ensuring comparability by the international agencies responsible for the collection and publication of these data. The data set is to contrast the two competing hypotheses. On the one hand, the literature stresses that government interference is the ultimate cause of the informal sector in developed, developing and transition countries alike. The competing hypotheses, favored here, is that the ultimate cause of the transition shadow is the institutional vacuum left after the fall of socialism, while its proximate (identifiable)
The “macro electric approach” is the correct one for an informal sector characterized by high capital intensity. Because this is not its defining characteristic in the rest of the world, available estimates using this method are rare outside transition. 28 The composition of the groups of countries was discussed above. The figures present group averages weighted by real GDP per capita (in PPP 1987 dollars) and by population. Choosing other variables as weights (in the first case, initial income per capita or equal weights, and in the second case, GDP) generates similar conclusions (these are available upon request). 27
13
cause is macroeconomic instability. Table 2 shows basic statistics. The variable called Shadow share is the one just described, and is taken from Johnson, Kaufmann and Shleifer (1997). A major source for the remaining variables is the EBRD Transition Report 1998. Two variables represent the conventional wisdom’s hypothesis (hereafter, CW). The literature often equates “government interference” to the “size of government” because measuring government size is simpler than measuring the extent and effectiveness of government intervention. Accordingly, larger governments are expected to be associated with worse economic performance.29 “Government interference”
is
proxied
by
“government
consumption”
and
“government
expenditures.” The share of general government consumption in GDP measures government consumption. It is also a component of general government expenditures, which includes, inter alia, capital expenditures, interest payments and social expenditures.30 If the conventional wisdom holds, one should expect that government consumption and expenditures are positively and causally related to the shadow share in transition. For the institutional vacuum hypothesis (hereafter, IVC), the following four variables were selected. Inflation is the annual average of the inflation rate.31 Investment is gross domestic fixed investment as a percentage of GDP. Unemployment See Durlauf and Quah (1998) for an overview of this literature and, as influential examples of it, Barro (1991) and Commander, Davoodi and Lee (1997). Dissent is found in La Porta, Lopez-de-Silanes, Shleifer and Vishny, (1998) and, focusing on transition, Campos (1999). 30 Note the different sources for government consumption and expenditures: the former is taken from World Bank (WDR, WDI), and the later from the EBRD. Although the definitions in the two sources are the same, coverage of government expenditures is more complete. 31 Boone and Horder have established the link between the institutional vacuum and macroeconomic instability. For instance, they argue that the institutional vacuum is at the very root of the inflation problems experienced by the transition countries in the early 1990s: “[the breakdown of the one-party system] meant that many of the checks and balances on political decision were lost… In the vaccum that followed the political breakdown, the old elites and rent seekers captured the political initiative in these countries. To sustain their powers, and sequester incomes, they issued credits and maintained distortionary policies and, as a result, acquired enormous assets” (1998, p.43). 29
14
refers to the registered unemployment rate, and exchange rates are annual averages, national currencies to the U.S. dollar. If the “institutional vacuum hypothesis” holds, one should expect that unemployment, inflation and the exchange rates are causally and positively related and investment is causally and negatively related to the shadow share in transition.32 Table 3 displays the correlation matrix (notice that the inflation and exchange rates are in logarithms). Although most of the pair-wise correlations are low, they do reveal a number of interesting features. The share of the informal sector is negatively correlated with investment and positively with inflation. Somewhat surprisingly, it is also negatively correlated with unemployment, the exchange rate, and with government expenditures and consumption. The advantages and limitations of the Granger-causality framework, the method chosen for testing these hypotheses, must also be discussed. The framework has endured the test of time because of its elegance and strong intuitive appeal: the notion that an event in the future cannot cause one in the past.33 Consider two time series, x t and y t. Series x t is said to Granger-cause series y t if, in a regression of y t on lagged y’s and lagged x’s, the coefficients of the lagged x’s are jointly significantly different from zero.34 There are some important issues in performing Granger causality tests. One concerns the length and frequency of the time lags. On their length, Granger
The hypothesis was previously formulated in terms of macroeconomic instability. The use of first-differences below should emend reservations regarding this choice of proxies. 33 Notice that in Johnson, Kaufmann and Shleifer (1997), the exogenous variables (“rule of law” included) are measured over 1995-1997, while the endogenous variable is measured over 1995 or averaged over 1989-1995. The consequences of this seem not to have escaped one of the commentators: “can one conclude anything about causality here from running a lot of regressions?” (Weitzman, 1997, p. 232). 34 Granger remarks that “causation is a non-symmetric relationship, and there are various ways in which asymmetry can be introduced, the most important of which are controllability, a relevant theory, outside knowledge, and temporal priority” (1987, 49.) 32
15
admonishes that “using data measured over intervals much wider than actual causal lags can destroy causal interpretation” (Granger, 1987, p.49). Data availability dictates the use of a lag length of one year. As for their frequency, because of the very short panel I kept the number of lags fixed (at one). Another issue refers to the information set problem. The test depends on the assumption that the cause contains unique information about the effect, in that this information is exhaustive and unavailable elsewhere. If the information set underlying the test is composed solely of two series, both affected by a third, then the test is rendered useless. The Granger-causality results that follow are enlarged by a variable that can potentially play such disruptive role (the level of income per capita). Finally, the presence of the lagged dependent variable brings in a critical econometric issue: the dynamic panel data problem.35 It is well established that the lagged dependent variable is correlated with the error term by construction, rendering the OLS estimator biased and inconsistent. To rectify this problem, this paper use the fixed-effects estimator pioneered by Anderson and Hsiao (1982) and enhanced by Arellano (1989).36 Because the argument in the previous section was formulated in terms of the rates of growth of the shadow share in transition, equal weight is given to conceptual and econometric considerations. This ultimately results in an unorthodox use of standard econometric methodology, suggesting “quasi-structural Granger causality” as a good description for the exercise carried out below.37
See Hsiao (1986) and Baltagi (1995). Notice that econometric research is yet to reach a consensus on the appropriateness of competing estimators in the dynamic panel context (Judson and Oven, 1999). 36 The latter involves the use, as instruments, of lagged levels not lagged differences. 37 The imposition of theoretical restrictions to a vector autoregressive specification is referred to in the literature as “structural” (SVAR, after Blanchard and Quah, 1989). Yet the meeker is preferred for the results presented in this paper. 35
16
IV. RESULTS This section presents the results of the tests for the two hypotheses outlined above (CW and IVC). Table 4 shows the first set of results that investigate the causes of the annual increases in the shadow share in GDP. For the whole sample (row labeled “ALL”) as well as for the two sub-samples (Commonwealth of Independent States, CIS, and Central and Eastern Europe and Baltics, CEEB), only the annual changes in the rate of inflation is found to be related in a causal fashion to the annual changes of the shadow share. A higher rate of inflation Granger-causes a higher annual increase in the shadow share. For the complete sample, lower investment and lower government expenditures, and a higher exchange rate cause increases in the shadow share. The only result that obtains for the CIS is regarding inflation, while for the CEEB the coefficients on investment and government expenditures are also statistically significant.38 Table 5 shows “first-differences results”. For the complete sample, lower investment and lower government expenditures and higher inflation Granger-cause increases in the shadow share.39 Although there seem to be no identifiable determinants for the CIS shadow, for the CEEB countries lower investment and higher inflation cause increases in the shadow share. In order to assess the robustness of these results, the issue of the information set was addressed by controlling for the level of income per capita. For the “OLS/firstdifferences” results, the two major resulting changes are that the coefficient on government consumption for the CEEB group becomes statistically significant and it
Keeping with the “quasi-structural approach,” reverse causality results are not reported (they are available upon request). Yet, robust evidence for reverse causality was only found for government expenditures, further damaging CW. For the sake of completeness, there are no statistically significant results for the relationship between the shadow share in GDP and economic growth. 39 Notice that these results also contradict CW and support IVC. 38
17
is positive 40 and the coefficient on inflation for the CIS becomes statistically insignificant (Table 6). Finally, there are two resulting changes for the “IV/firstdifferences” results (Table 7). First, the coefficients for inflation (for “ALL” and “CEEB”) become statistically insignificant and, second, the coefficient on government consumption for the CEEB group becomes statistically significant (and it is positive). These results are discussed in the next section.
V. CONCLUSIONS The conventional wisdom on the informal sector in the former centrally planned economies is that its cause is the same in transition as in developing and developed countries, and that this cause is excessive government interference. Because of its high capital intensity, we argue that the transition shadow is unique. This uniqueness raises the possibility that the causes are not the same, and prompted a search for an alternative. Instead of government interference, a case was made that the ultimate cause of the transition shadow is the institutional vacuum left after the fall of socialism. In order to test this empirically, a second search took place for a proximate cause. Drawing from the literature, we postulate that macroeconomic instability is the best available candidate. The results presented in the previous section show that the support for the conventional wisdom is feeble. There seem to be little econometric evidence supporting the argument that a larger government is associated with a larger shadow economy. For the two proxies for excessive government interference, the results are somewhat robust for only one, but still disserving the conventional wisdom: lower government expenditures Granger-causes a greater shadow share. One seems to have two options: to accept these results at face value or recognize their preliminary 40
Notice that this is the only result favorable to CW.
18
character, the fact that it can only be remedied at enormous cost, and indicate the alternative explanation as a more fruitful avenue for future research. The results show that investment and inflation are important causes of the shadow economy in the early years of the transition, but also that exchange rates and unemployment are related to the transition shadow in a causal manner. Declining investment and higher inflation rates do Granger-cause an increasing shadow share in GDP. And they do so in much more robust fashion then any of the conventional wisdom’s variables there are data for. In light of these results, what are the suggestions for future research? First, notice that until this point the words barter, arrears, evasion, corruption, or Mafia have not been mentioned.41 The reason is simple: despite their apparently obvious connections, there seems to be no single contribution that articulates their relationships in transition. This should receive top priority. Second, more empirical research is needed on which sectors or activities do the transition shadow concentrate. For economists working in Transition, the notion that high capital intensity is the defining characteristic of the shadow economy does not seem to raise objections.42 Yet, development economists will surely demand hard evidence, and time to grasp. Third, the lack of a formal theoretical structure guiding the econometric exercises above is somewhat of a hindrance. A formal model, in which firms choose the share of output to report and in which capital intensity is explicitly incorporated, would be a major contribution. Notice also that such a model to be empirically
On these issues, see Anderson (1995), Bardhan (1997), Clifford and Ickes (1998a, 1998b), Commander and Mumssen (1998), and Pirttila (1999). 42 In personal communications, Simon Johnson, Arye Hillman, Gur Ofer, Gerard Roland and Heinrich Ursprung, all agreed with this characterization. 41
19
verifiable, in a cost-effective manner, should not depend mostly or exclusively upon tax rates. Fourth, within the limits of the econometric exercise above, there are many of the variables used in this paper that could be measured in a better way. For instance, the available data is for registered unemployment and one can not distinguish the separate roles of public and private investment. The econometrics methodology can also be improved, for example, by experimenting with other estimators (Judson and Owen, 1999). Notice, however, the latter requires larger panels (in both dimensions: more years and more countries), so one should mind the costs. Finally, an important task is the construction of a panel data set of institutional measures that would allow a direct testing of the relationships between the “institutional vacuum” and macroeconomic instability, and between the latter two and the share in GDP of the transition shadow.43
The empirical literature on the economic impact of institutions has draw much comfort from the proposition that institutions do not change, or that they change so slowly that crosssectional data provides an appropriate representation. Campos and Nugent (1999) critically assess this proposition. Aron (1998) surveys this literature to find only one panel data study. For an empirical analysis of the impact of institutions in transition, see Adelman and Vujovic (1998). Campos (1999) provides a panel data set of institutional measures but adverts that those are preliminary and their consistency is still to be checked, independently. 43
20
REFERENCES Adelman, Irma and Dusan Vujovic, “Institutional and Policy Aspects of the Transition: An Empirical Analysis”, in A. Levy-Livermore (ed) Handbook on the Globalization of the World Economy, Cheltenham: Edward Elgar, 1998. Anderson, A., “The Red Mafia: A Legacy of Communism,” in E. Lazear (ed) Economic Transition in Eastern Europe and Russia: Real Ties of Reform, Stanford: Hoover Press, 1995. Anderson, James, “The Size, Origins, and Character of Mongolia’s Informal Sector during the Transition,” Washington: World Bank, WPS 1916, May 1998. Anderson, T.W. and Cheng Hsiao, “Formulation and Estimation of Dynamic Models using Panel Data,” Journal of Econometrics 18, 47-82, 1982. Arellano, M., “A Note on the Anderson-Hsiao Estimator for Panel Data,” Economic Letters 31, 337-341, 1989. Aron, Janine, “Political, Economic and Social Institutions:A Review of Growth Evidence,” Oxford: Oxford University, Institute of Economics and Statistics Working Papers Series No. 98-4, 1998. Åslund, Anders, “Why Has Russia’s Economic Transformation Been So Arduous?” Paper presented at Annual World Bank Conference on Development Economics, Washington, D.C., April 1999. Baltagi, Badi, Econometric Analysis of Panel Data. New York: John Wiley & Sons, 1995. Bardhan, Pranab, “Corruption and Development,” Journal of Economic Literature, XXXV, 1320-1346, 1997. Barro, Robert, “Economic Growth in a Cross Section of Countries”, Quarterly Journal of Economics 106, 407-444, 1991. Bartholdy, Kasper, “Old and New Problems in the Estimation of National Accounts in Transition Economies”, Economics of Transition 5, 131-146, 1997. Bedi, Arjun, “Sector Choice, Multiple Job Holding and Wage Differencials: Evidence from Poland,” Journal of Development Studies 35, 162-179, 1998. Blanchard, Olivier, The Economics of Post-Communist Transition, Oxford: Clarendon Press, 1997. Blanchard, Olivier and Danny Quah, ‘The Dynamic Effects of Aggregate Demand and Supply Disturbances,” American Economic Review 79, 655-73, 1989.
21
Boone, Peter and Jakob Horder, “Inflation: Causes, Consequences, and Cures”, in Peter Boone, Stanislaw Gomulka and Richard Layard (eds) Emerging from Communism: Lessons from Russia, China, and Eastern Europe, Cambridge: MIT Press, 1998. Campos, Nauro, “Back to the Future: The Growth Prospects of Transition Economies Reconsidered,” Ann Arbor: William Davidson Institute Working Paper No. 229, 1999. Campos, Nauro, “Context is Everything: Measuring Institutional Change in Transition Economies,” World Bank: Robert McNamara Fellow Working Paper, 1999. Campos, Nauro and Jeffrey Nugent, “Development Performance and the Institutions of Governance: Evidence from East Asia and Latin America,” World Development 27, 439-452, 1999. Campos, Nauro, Khan, Feisal and Jennifer Tessendorf, “Can Good Institutions Substitute for Bad Policies? Some Econometric Evidence from Pakistan,” Los Angeles: USC Economics Department Working Paper No. 9905, 1999. Commander, Simmon and Andrei Tolstopiatenko, “A Model of the Informal Economy in the Transition Setting”, Ann Arbor: William Davidson Institute Working Paper No. 122, 1997. Commander, Simon, Davoodi, Hamid and Une Lee, “The Causes of Government and the Consequences for Growth and Well Being,” Washington, D.C.: World Bank WPS 1785, 1997. Commander, Simon and Christian Mumssen, “Understanding Barter in Russia”, London: European Bank for Reconstruction and Development Working Paper No.37, December 1998. Dallago, Bruno, “The Underground Economy in the West and in the East: A Comparative Approach,” in S. Alessandrini and B. Dallago (eds), The Unofficial Economy, London: Gower Publishing, 1987. Dewatripont, Mathias and Gerard Roland, “Transition as a Process of Large Scale Institutional Change,” in D. Kreps and K. Wallis (eds), Advances in Economics and Econometrics: Theory and Applications (Volume II), Cambridge: Cambridge University Press, 1997. Durlauf, Stephen and Danny Quah, “The New Empirics of Economic Growth,” London: LSE Centre for Economic Performance Discussion Paper No. 384, 1998. Enste, Dominik and Friedrich Schneider, “Increasing Shadow Economies All Over the World — Fiction or Reality? A Survey of the Global Evidence of their Size and of their Impact from 1970 to 1995,” Bonn: Institute for the Study of Labor (IZA) Working Paper No. 26, 1998. European Bank for Reconstruction and Development [EBRD], Transition Report, London: European Bank for Reconstruction and Development, 1998.
22
Frey, Bruno and Hannelore Weck-Haneman, “The Hiden Economy as a ‘Unonbserved’ Variable,” European Economic Review 26, 33-54, 1984. Gaddy, Clifford and Barry Ickes, “A Simple Four Sector Model of Russia’s ‘Virtual’ Economy”, Penn State University, mimeo, 1998a. Gaddy, Clifford and Barry Ickes, “To Restructure or Not to Restructure: Informal Activities and Enterprise Behavior in Transition,” mimeo, paper presented at the CEPR Transition Conference, Prague, 1998b. Granger, C.W., “Causal Inference,” The New Palgrave: Econometrics, New York: W.W. Norton, 1987. Ickes, Barry, “Dimensions of Transition in Russia”, in B. Granville and P. Oppenheimer (eds) The Russian Economy in the 1990s, Oxford: Oxford University Press, forthcoming. Harris, John and Michael Todaro, “Migration, Unemployment and Development: A Two-sector Analysis,” American Economic Review 60, 126-142, 1970. Hsiao, Cheng, Analysis of Panel Data, Cambridge: Cambridge University Press, 1986. Johnson, Simon, Kaufmann, Daniel and Andrei Shleifer, “The Unofficial Economy in Transition,” Brookings Papers on Economic Activity 2, 159-221, 1997. Johnson, Simon, Kaufmann, Daniel and Pablo Zoido-Lobaton, “Regulatory Discretion and the Unofficial Economy”, American Economic Review Papers and Proceedings 88, 387-392, 1998. Johnson, Simon, McMillan, John and Christopher Woodruff, “Why do Firms Hide? Bribes and Unofficial Activity After Communism,” London: CEPR Discussion Paper No. 2105, 1999. Judson, Ruth and Ann Owen, “Estimating Dynamic Panel Data Models: A Guide for Macroeconomists,” Economic Letters 65, 9-15, 1999. Kaufmann, Daniel and Aleksander Kaliberda, “An ‘Unofficial’ Analysis of Economies in Transition: An Empirical Framework and Lessons for Policy,” Cambridge: HIID Discussion Paper No. 558, 1996. Kolev, Alexandre, “Labor Supply in the Informal Economy in Russia during the Transition,” London: CEPR Discussion Paper No. 2024, 1998. Kommerzant Daily (??????????? Daily), Moscow: December 9 issue, 1998. See http://www.rbc.ru/news/archiv/econom/1998/12/09/news19981209100613.htm Kornai, Janos, The Socialist System: The Political Economy of Communism, Princeton: Princeton University Press, 1992.
23
Landau, Zbigniew, “Select Problems of Unofficial Economy in Poland,” in S. Alessandrini and B. Dallago (eds), The Unofficial Economy, London: Gower Publishing, 1987. La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and R. Vishny, “The Quality of Government,” Cambridge: NBER Working Paper No. 6727, 1998. Levenson, Alec and William Maloney, “The Informal Sector, Firm Dynamics and Institutional Participation,” Washington: World Bank WPS 1988, July 1998. Lin, Justin and Jeffrey B. Nugent, “Institutions and Economic Development”, in J. Behrman and T.N.Srinivasan (eds) Handbook of Development Economics: Volume 3A, Amsterdam: North-Holland, 1995. Loayza, Norman, “The Economics of the Informal Sector: A Simple Model and Some Empirical Evidence from Latin America,” Carnegie Rochester Conference Series on Public Policy 45, 129-62, 1996. Marcouiller, Douglas and Leslie Young, “The Black Hole of Graft: The Predatory State and The Informal Sector,” American Economic Review 85, 630-646, 1995. Mars, Gerald and Yochanan Altman, “Case Studies in Second Economy Production and Transportation in Soviet Georgia,” in S. Alessandrini and B. Dallago (eds), The Unofficial Economy, London: Gower Publishing, 1987. Mead, Donald and Christian Morrisson, “The Informal Sector Elephant,” World Development 24, 1611-19, 1996. Nicolini, Juan Pablo, “Tax Evasion and The Optimal Inflation Tax,” Journal of Development Economics 55, 215-232, 1998. Nugent, Jeffrey, “Institutions, Markets and Developmental Outcomes,” in R. Picciotto and E. Wiesner (eds) Evaluation and Development: The Institutional Dimension, New Brunswick: Transaction Publishers, 1998. OECD, Framework for The Measurement of Unrecorded Economic Activities In Transition Economies, Paris: OECD, 1997. Patel, Urjit and Pradeep Srivastava, “Macroeconomic Policy and Output Comovement: The Formal and Informal Sectors in India,” World Development, 24, 1915-23, 1996. Pirttila, Jukka, “Tax Evasion and Economies in Transition: Lessons from Tax Theory,” Helsinki: Bank of Finland BOFIT Discussion Paper 0299, 1999. Pistor, Katharina, “The Evolution of Legal Institutions and Economic Regime Change,” presented at the Annual World Bank Conference on Development Economics in Europe, Paris, June 1999. Rauch, James, “Modeling the Informal Sector Formally,” Journal of Development Economics 35, 33-47, 1991.
24
Roland, Gerard and Thierry Verdier, “Law Enforcement and Transition,” Brussels and Paris: ECARE and DELTA, mimeo, 1999. Stark, Oded, “On Modeling the Informal Sector,” World Development 10, 413-16, 1982. Stiglitz, Joseph, “Whither Reform? Ten Years of the Transition,” presented at the Annual World Bank Conference on Development Economics, Washington D.C.: World Bank, 1999. Weitzman, Martin, “Comments on Johnson, Kaufmann and Shleifer’s The Unofficial Economy in Transition,” Brookings Papers on Economic Activity 2, 230-236, 1997. World Bank, World Development Report [WDR], Washington D.C.: World Bank, various issues. World Bank, World Development Indicators [WDI], Washington D.C.: World Bank, various issues. Vostroknoutova, Katherina, “Informal Sector in Transition: The Case of Russia,” St. Petersburg State University, mimeo, June 1998.
25
Table 1 The Share of the Shadow Economy in the GDP of 17 Transition Countries, 1989-1995 CIS Azerbaijan Belarus Georgia Kazakhstan Russia Ukraine Uzbekistan AVERAGE CEEB Bulgaria Czech Rep. Estonia Hungary Latvia Lithuania Moldova Poland Romania Slovakia AVERAGE
1989 12 12 12 12 12 12 12 12
1990 21.9 15.4 24.9 17 14.7 16.3 11.4 17.37
1991 22.7 16.6 36 19.7 23.5 25.6 7.8 21.7
1992 39.2 13.2 52.3 24.9 32.8 33.6 11.7 29.67
1993 51.2 11 61 27.2 36.7 38 10.1 33.6
1994 58 18.9 63.5 34.1 40.3 45.7 9.5 38.57
1995 60.6 19.3 62.6 34.3 41.6 48.9 6.5 39.12
22.8 6 12 27 12 12 12 15.7 22.3 6 14.78
1990 25.1 6.7 19.9 28 12.8 11.3 18.1 19.6 13.7 7.7 16.29
1991 23.9 12.9 26.2 32.9 19 21.8 27.1 23.5 15.7 15.1 21.81
1992 25 16.9 25.4 30.6 34.3 39.2 37.3 19.7 18 17.6 26.4
1993 29.9 16.9 24.1 28.5 31 31.7 34 18.5 16.4 16.2 24.72
1994 29.1 17.6 25.1 27.7 34.2 28.7 39.7 15.2 17.4 14.6 24.93
1995 36.2 11.3 11.8 29 35.3 21.6 35.7 12.6 19.1 5.8 21.84
26
Figure 1. Share of the shadow economy in GDP Averages weighted by real per capita GDP 40.00 35.00 30.00 25.00 20.00 15.00 10.00 90
91
92
asia
balkan
93 baltic
94 bur
95
visegrad
Figure 2. Share of the shadow economy in GDP Averages weighted by population 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 90
91 asia
92 balkan
93 baltic
94 bur
95
visegrad
27
Table 2 Basic Statistics 17 Transition countries, 1989-1995 Mean
Standard Deviation Economic growth -6.50533 9.32633 Investment 21.60114 7.20660 Gov’t consumption 18.31163 4.49570 Gov’t expenditures 43.56000 10.94807 Shadow share 25.30533 12.40341 Log Inflation (annual) 4.92790 1.78893 Unemployment 6.00400 5.15294 Exchange rate 2824.107 17388.29 NOTE: See text for sources and definitions.
Minimum
Maximum
-37.70 0.150 6.45936 12.300 5.800 2.20827 0.00 0.000600
7.10 43.39648 26.18489 65.90 63.50 9.75678 16.40 147307.0
Table 3 Correlation matrix 17 Transition countries, 1989-1995
Economic growth Investment Gov’t Consumption Gov’t expenditures Shadow share Log Inflation Unemployment Exchange rate CIS dummy
grow 1 -0.12 0.027 0.193 -0.43 -0.61 0.401 -0.03 -0.36
inv
gov_c
gov_e
shadow
Infl
un
exch
1 0.43 0.22 -0.49 -0.14 -0.11 0.292 -0.21
1 0.109 -0.518 -0.107 -0.153 -0.255 0.024
1 -0.34 -0.47 0.487 0.240 -0.36
1 0.371 -0.097 -0.267 0.360
1 -0.485 -0.029 0.731
1 0.0516 -0.5572
1 -0.16
28
Table 4 Granger-causality tests OLS, first differences 17 Transition countries, 1989-1995 Investment ALL
CEEB
CIS
-.268127 [.012] 65 .208617 -.466794 [.009] 36 .219231 -.076161 [.556] 29 .179980
Unemployment -.097664 [.663] 59 .047467 -.106942 [.674] 33 -.031796 .210699 [.717] 26 .074448
Exchange rate 1.01353 [.018] 50 .123638 1.10217 [.317] 31 .027748 .737571 [.148] 19 -.055730
Inflation (annual avg) 1.39011 [.001] 68 .220584 1.56602 [.019] 36 .146902 1.18409 [.037] 32 .142199
Government Consumption .274869 [.134] 63 .146090 .369991 [.225] 36 .052519 .225884 [.303] 27 .166193
Government Expenditures -.142579 [.066] 60 .110602 -.287259 [.080] 36 .100918 -.117377 [.132] 24 .115585
Note: Each cell contains summary information of a regression of the share of the shadow economy in GDP on the variable in that column. In each cell, the first row contain the value of the coefficient, the second row contain its p-value (in brackets), the third row contain the number of observations, and the fourth an last row contains the Adjusted R2 of that regression.
Table 5 Granger-causality tests IV, first differences 17 Transition countries, 1989-1995 Investment ALL
CEEB
CIS
-.261797 [.013] 65 .207299 -.454434 [.014] 36 .215333 -.071339 [.580] 29 .179834
Unemployment -.184126 [.480] 59 .048050 -.158658 [.566] 33 -.032231 -.200381 [.766] 26 .068267
Exchange rate 2.53089 [.264] 50 .00650476 1.67847 [.380] 31 .012735 -.131420 [.897] 19 -.052701
Inflation (annual avg) 1.22366 [.057] 68 .215313 1.46219 [.090] 36 .146572 .919964 [.191] 32 .135432
Government Consumption .276387 [.128] 63 .146042 .380306 [.210] 36 .047102 .233441 [.281] 27 .165678
Government Expenditures -.143201 [.061] 60 .110019 -.278828 [.121] 36 .102170 -.122587 [.107] 24 .114119
Note: Each cell contains summary information of a regression of the share of the shadow economy in GDP on the variable in that column. In each cell, the first row contain the value of the coefficient, the second row contain its p-value (in brackets), the third row contain the number of observations, and the fourth an last row contains the Adjusted R2 of that regression.
29
Table 6 Granger-causality tests OLS, first differences 17 Transition countries, 1989-1995 Investment ALL
CEEB
CIS
-.261863 [.016] 64 .185844 -.462356 [.011] 36 .184836 -.015578 [.910] 28 .152489
Unemployment -.179642 [.454] 59 .041324 -.065074 [.816] 33 -.066187 -.131705 [.814] 26 .096655
Exchange rate 1.01788 [.024] 50 .104179 1.35629 [.240] 31 -.026473 .081708 [.869] 19 .107561
Inflation (annual avg) 1.37481 [.002] 67 .212959 1.63897 [.023] 36 .128970 .867883 [.136] 31 .144982
Government Consumption .271623 [.170] 63 .131792 .586411 [.095] 36 .052239 .150975 [.497] 27 .156718
Government Expenditures -.141847 [.072] 60 .095358 -.285933 [.085] 36 .066596 -.090783 [.269] 24 .079445
Note: Each cell contains summary information of a regression of the share of the shadow economy in GDP on the variable in that column. In each cell, the first row contain the value of the coefficient, the second row contain its p-value (in brackets), the third row contain the number of observations, and the fourth an last row contains the Adjusted R2 of that regression.
Table 7 Granger-causality tests IV, first differences 17 Transition countries, 1989-1995 Investment ALL
CEEB
CIS
-.227902 [.077] 64 .146609 -.462132 [.014] 36 .184828 -.025328 [.893] 28 .152556
Unemployment -.190035 [.450] 59 .038422 -.108107 [.712] 33 -.066624 -.110360 [.865] 26 .093157
Exchange rate 6.43497 [.697] 50 -.037230 1.37589 [.443] 31 -.026055 .545823 [.559] 19 -.042792
Inflation (annual avg) 1.88669 [.429] 67 .139519 1.03339 [.430] 36 .070035 1.36693 [.190] 31 -.024717
Government Consumption .317103 [.299] 63 .130138 .740743 [.057] 36 .034678 .160998 [.569] 27 .157206
Government Expenditures -.139039 [.079] 60 .079923 -.237444 [.211] 36 .045733 -.077390 [.497] 24 .066204
Note: Each cell contains summary information of a regression of the share of the shadow economy in GDP on the variable in that column. In each cell, the first row contain the value of the coefficient, the second row contain its p-value (in brackets), the third row contain the number of observations, and the fourth an last row contains the Adjusted R2 of that regression.
30