ALTERNATIVE STRATEGIES FOR FIRMS IN OPPRESSIVE AND CORRUPT STATES: INFORMALITY OR FORMALITY VIA BUSINESS ASSOCIATIONS?

I. INTRODUCTION

Firms in oppressive environments such as those of the transition economies face a variety of problems, including the high cost of complying with regulations, frequent changes in those regulations, and hence lack of knowledge about current regulations and policies; difficulties in gaining access to telephones, electricity, and infrastructure; inefficient and corrupt judicial systems; and cumbersome and nontransparent licensing, customs procedures, and rule making. As a result, it is common for firms in transition economies to make unofficial payments to officials to obtain needed services and avoid taxes and regulations (Djankov et al., 2002; Johnson, Kaufmann, and Schleifer, 1997).

While there are oppressive and corrupt regimes in many parts of the world, a survey by the European Bank for Reconstruction and Development (EBRD, 1999) showed the transition economies to have the highest scores for corruption among public officials of all country groupings. As recently as 2001, only 1 of the 25 transition countries included in this study (Slovenia) had a score for control of corruption of 5.0 or more on the widely used 0–10 scale of the Corruption Perception Index (CPI) of Transparency International.¹ Even after some improvement in corruption control or freedom from corruption in the region since then, in 2007 there were still only three transition countries with scores of 5.0 or more, and half of the rest had scores of 3.0 or below (Corruption Perception Index website). These scores are based on the perceptions of a wide variety of people.

Using data from the first year (2002) of the two rounds of the survey used in this study for 25 transition countries, Campos and Giovanoni (2008) assert that some 18% of the firms in the sample say that 10% or more of their revenues are spent on unofficial payments to officials. Based on a still earlier round (1999) of the same data set, Hellman, Jones, and Kaufmann (2000) find that the percentage of firms in each transition country admitting to ever having paid bribes ranged from 45% (Belarus and Slovenia) to 90% (Kyrgyzstan). With respect to the frequency of such payments, Tanzi and Tsibouris (2000) report that as much as 59.3% of firms (for Azerbaijan) admitted to making these unofficial payments “frequently or more often.” Somewhat surprisingly, even with improving economic conditions in transition economies and their success in attracting foreign direct investment (FDI), firms in transition economies with FDI are reported to be even more prone than other domestic firms to make unofficial payments (Anderson, 2005).

Tax evasion is another huge and endemic problem in the transition economies. Uslaner (forthcoming) cites numerous sources stating the extent of tax evasion to be extremely high in most transition countries, ranging up to 40% of gross domestic product (GDP). He also shows that “tax morale” (as measured by questions on the World Value Surveys concerning the acceptability of tax cheating) is typically much lower for people in transition countries than for those in other countries. (See also Hanousek and Palda, 2004.) When citizens deem their governments to be corrupt, implying that funds are diverted from their intended public purposes to the private ones of bureaucrats and their friends, they see nothing wrong with not paying their taxes. This makes these two distinguishing characteristics of transition countries, high corruption and tax evasion, interrelated. If judges are corrupt, firms know that the court system will not be fair. If tax officials are corrupt, tax enforcement is likely to be sporadic and inefficient. With corruption and stiflingly rigid regulations and tax rates, a common strategy for firms is to operate outside the official economy. Indeed, Johnson, Kaufmann, and Schleifer (1997) reported the shadow economies of the countries of the former Soviet Union to average 34% of their respective GDPs and corruption to be related to tax evasion. But, since this implies that firms would remain small and thereby be unable to take advantage of economies of scale and scope, discouraging market development and preventing governments from collecting the revenues to pay for needed infrastructure, such informality can be very costly to development.

At the same time, however, firms in transition economies have also been joining various business associations. In some countries and industries this is mandatory, but for the most part such memberships are voluntary. The rise in association memberships could perhaps be interpreted as a response to the void created by the radical departure from central planning that had previously orchestrated virtually all transactions in the economy (the void being nicely characterized by Blanchard and Kremer, 1997).² While in principle, these associations could help firms in dealing with the various problems confronting them, as noted in Section II below, they could also serve as but another vehicle for unofficial payments, thereby contributing further to corruption and tax evasion, and to an anticompetitive economic environment.

Given the wide variation in views about the usefulness of business associations in transition countries, quantitative assessments of the effects of these associations on unofficial payments to officials and tax reporting would seem important. Such assessments, however, are rather rare. Quite naturally, most attention has been devoted to the roles of regulatory policy, property rights, and trade policy in explaining these phenomena.

The purposes of this paper were to (1)examine the relative importance of the aforementioned motives for membership in a business association, (2)assess the role of business associations and the functions they perform across the various transition countries with respect to unofficial payments to government officials and underreporting of income for tax purposes, and (3)derive implications for policies that would induce firms to reduce their unofficial payments to government officials and pay more of the taxes mandated by law.

These objectives are accomplished by taking advantage of cross-section data for a sizable sample of randomly selected firms in each of 25 transition economies for the years 2002 and 2005 taken from the Business Environment and Enterprise Performance Surveys (BEEPS) of the European Bank for Reconstruction and Development (EBRD) and the World Bank.³ We also make use of a panel formed by the subset of firms included in both the 2002 and 2005 surveys.

The remainder of the paper is organized as follows: Section II reviews relevant literature and outlines a simple analytical framework. Section III describes the data and methods used. Note that because membership in most associations is voluntary and self-selective, in estimation we need to deal with potential endogeneity. Section IV presents the empirical results, showing that the results are at least fairly robust to different assumptions and estimation strategies. Section V presents our conclusions, including policy implications.

II. LITERATURE AND ANALYTICAL FRAMEWORK

As noted above, there is very considerable disagreement in the existing literature on the role of business associations in transition economies. The dominant viewpoint, however, is negative. On the one hand, business associations are criticized for being relics of the old regime wherein industry was dominated by state enterprises and the associations were used for controlling businessmen. On the other hand, they are criticized for being mere lobbying organizations that serve primarily as vehicles for rent seeking, for example, to help favored firms obtain licenses and government favors of various sorts. Indeed, Olson (2000), one of the most important contributors to the theory of collective action and hence to business associations, has blamed business associations in Russia and other transition countries for contributing further to the noncompetitive, inefficient, and corrupt nature of these economies.

Yet, there is a strong and articulate minority, comprising Western nongovernmental organizations, donor organizations, and some scholars (Pyle, 2005, 2006, 2007), which sees hope in associations, arguing that under certain conditions they may be able to help firms deal with their problems. For example, Pyle (2006) draws on data for three transition countries to show that membership in a business association can help firms reduce transaction costs arising from commercial and other disputes especially those between firms located far from one another. Similarly, Pyle (2006) shows that business association membership helps to protect property rights, especially in the less democratic parts of Russia. Others point out that associations help firms get better information about foreign standards and markets. This raises the question of whether unofficial payments and informality on the one hand and business association memberships on the other are substitute or complementary strategies for resolving the problems that firms face in transition economies.

While there can be situations where conditions are sufficiently bad that neither strategy will be sufficient and that firms will want to use both methods, our hypothesis, based more on the literature of business associations in developing countries than on that in transition countries, is that, on balance at least, unofficial payments to officials and memberships in business associations are net substitutes for one another. In rather ideal conditions in which government regulations are moderate and policies are business friendly, neither association memberships nor unofficial payments to government officials may be needed. But, in other circumstances, we argue that firms are likely to choose one strategy over the other depending on the relevant circumstances. When associations do not provide the kind of services that firms need, and institutional conditions are characterized by excessive regulation and taxation, firms will choose to bribe officials and/or to remain informal to avoid these regulations and taxes. On the other hand, when associations are better able to provide the needed services, regulations are somewhat less onerous and firms have the capabilities that allow them to take advantage of the services provided, they will choose to become association members. Moreover, as suggested by Olson (1962, 1982) in his earlier and broader analysis of collective action theory, associations (especially broad ones) are bound to reflect more encompassing and less particular concerns than individual firms directly approaching state officials. As a result, efforts to affect government indirectly through associations are likely to be less distorting and narrowly rent seeking than those made directly by firms in the form of unofficial payments to officials.

An important reason for these widely differing perspectives on the role of business associations in transition countries, undoubtedly, is their considerable heterogeneity in type and function. Certainly, business associations exist that are relics of the communist era and designed to help state agents control firms. So, too, there exist business associations that, instead of being designed to control businesses, have the purpose of trying to control government through lobbying and corruption. Associations can differ in other important respects as well. In some cases, membership is mandatory, in other cases voluntary. Associations operating in different sectors; catering to firms of different size, organized on different scales (local, regional, or national); possessing memberships of different sizes; and operating in countries with different governance characteristics may also differ in their usefulness to firms. There has been much classificatory and analytic work of this sort on associations, mostly in developing countries, demonstrating how these differences may matter for firms’ behavior and performance (Doner and Schneider, 2000; Maxfield and Schneider, 1997; Schneider, 2005).

For transition countries, the literature assessing the effects of association memberships on either firm behavior or firm and/or economy level performance is less advanced. Much of that which exists is confined to descriptions of what specific associations do or to variations in memberships across countries. One aspect that has been examined more thoroughly in transition countries is (as indicated below) the influence of political conditions on the relative usefulness of one function of associations, namely lobbying.

Gehlbach (2006) uses the earlier 1999 round of BEEPS to examine the relationship among association membership, public goods supply by government, and underreporting of income for tax purposes to support an innovative incomplete contracts model of collective action. Firm and industry characteristics are shown to be related to underreporting of income and then from income underreporting and association membership to public goods supply. The relationship between association membership and underreporting of income, however, is ignored.

As noted above, using a smaller sample of firms in five transition countries, Johnson et al. (2000) show that corruption is positively correlated with underreporting of income for tax purposes. But, they do not bring in the role of association membership which, as we demonstrate below, appears to be rather essential to the understanding of this relation.

Based on both a cross-country analysis of the 25 countries covered by BEEPS (1999) and qualitative data on business association formation in two sectors of the Russian economy, Duvanova (2007) also finds a positive effect of bureaucratic corruption on membership levels in business associations. According to her analysis, regulatory pressures on business and bureaucratic corruption may induce firms to join business associations as a means of protecting themselves from predatory state behavior. She does not, however, examine either the inverse effect, that is, the effect of association membership on bureaucratic corruption or distinguish among the different functions that associations perform in these relationships. She also does not consider (let alone treat) either (a) possible endogeneity of bureaucratic corruption or (b) the related and equally important effects of association membership on either the underreporting of income for tax purposes or the informality of firms. Duvanova (2006) examines the effect of firm and sector characteristics on business association membership in Russia, but again does not relate these to either unofficial payments or underreporting of taxes.

The studies of greatest relevance to the present one are Campos and Giovannoni (2007, 2008). Like us, these authors examine the determinants and effects of two different strategies–lobbying and corruption payments–of firms in transition economies to obtain favor for their businesses from the authorities. Contrary to the widely presumed complementarity between the two strategies both studies show these activities to be substitutes and the choice between them and the relative effectiveness of each in influencing policy to depend on political conditions. The first study uses BEEPS (1999). The latter goes into much more depth as far as the political variables are concerned and examines the effects of each on performance and makes use of BEEPS (2002). As in much of the literature on business associations in transition countries, membership is identified exclusively with lobbying, implying that all other functions of such associations, for example, in settling disputes and in obtaining information about customers, suppliers, technical standards and government policies, are ignored. Unlike Duvanova (2007), Campos and Giovannoni (2007) find rather a negative and statistically significant effect of bureaucratic corruption on membership in business association, even though both studies use the same data set, BEEPS 1999. In their 2008 study, using BEEPS (2002), they find no such effect (though positive) to be significant, regardless of whether lobby membership is used to explain bureaucratic corruption or bureaucratic corruption is used to explain lobby membership.⁴ By contrast, in this study, we find a strong, negative effect of association membership on making unofficial payments in three different data sets, BEEPS (2002, 2005, and the smaller 2002–2005 panel). Our results are robust to the use of different specifications and whether or not we account for the possible endogeneity of such variables. Furthermore, none of these studies examines the effects of lobbying on income underreporting for tax purposes, one of the objectives in the present study.

Another issue that has been neglected in these previous studies but which is an objective of ours is the relative effectiveness of the various other functions that associations perform besides lobbying. We demonstrate that some of these neglected functions of associations may be more important than lobbying in determining firm behavior. Since the 2002 and 2005 rounds of the survey are larger and more detailed than that for 1999, and since a subset of the firms in these two rounds form a panel, our attention is focused on the more recent 2002 and 2005 rounds of BEEPS.

We focus on the issues overlooked in these earlier studies. Specifically, our results suggest that business association memberships seem to provide sufficient incentives for firms to become formal, reducing the underreporting of income for tax purposes and unofficial payments. Likewise, unlike the results obtained in Duvanova (2007), we find rather a negative and significant effect of membership in business association on making unofficial payments.

For an analytical framework, let us assume that firms have four different options in dealing with their common problems: (1) to become members of a business association and benefit from the services it provides, and (2) to make unofficial payments to corrupt government officials to get around regulations and underreport profits so as to reduce their taxes, (3) to do both (1) and (2), and (4) if the problems are not sufficiently great, to do neither (1) nor (2).

During transition, with inefficient or corrupt judicial systems and governments that are unable to enforce their myriad regulations and laws, there is an institutional vacuum that needs to be filled by Options (1), (2), or (3). In Option (1), the association provides a public good to member firms, for example, by promoting interfirm cooperation through information about member profiles and activities. It helps enforce contracts by informing each other about contractual histories of firms—both good and bad. On the other hand, because of the weakness of government and its institutions, firms may be obliged to choose Option (2), reducing costs by avoiding regulations and taxes. Joining a business association⁵ is likely to be the more formal alternative but requires payment of a membership fee to receive the association’s services.

Some differences between these two options can also be expected. For example, the activities of business associations but not those provided by corrupt bureaucrats would be subject to increasing returns to scale. For example, payments to an association are entirely used for public goods production, whereas some portion of those to corrupt bureaucrats would be diverted to a private good consumed by the official.

Various firm characteristics can be expected to influence the choice between Options (1) and (2). For example, larger firms may pay higher fees than smaller firms but are likely to benefit from the larger variety of services provided by associations. At the same time, larger firms would be more easily detected making unofficial payments and hence less likely to choose Option (2). Similarly, firms that are sufficiently experienced and capable as to be able to make use of the services provided by the association, could also be expected to choose Option (1) but less likely to choose Option (2) because the illegality of the payments may undermine the renewal of the implicit contracts. (The services provided by business associations may also change over time in response to the collective opinions of the firms. But, in the short run, we assume that the services are given.) Smaller firms, on the other hand, are likely to save by making unofficial payments to corrupt bureaucrats and underreporting taxes only on those specific occasions when certain specific types of protection are needed.⁶

But external conditions also matter. When the firms’ problems are sufficiently adverse, multifaceted, or volatile, firms may feel the “need” for Option (3), that is, to join an association as well as to make unofficial payments. On the other hand, when the external conditions are favorable, Option (4) may apply. Option (4) may also apply to state-owned enterprises and influential firms (see footnote 5). Among the relevant external conditions are reforms. For example, serious efforts to combat corruption by punishing those detected making or receiving unofficial payments would discourage Option (2). Better enforcement of property rights might encourage association membership but reduce unofficial payments and increase willingness to pay taxes. In subsequent sections, we attempt to construct ways of bringing empirical evidence to bear on these hypotheses.

III. DATA AND METHODS

The firm level data from the 2002 and 2005 BEEPS employed in this study are for 25 transition economies.⁷ The samples for each year and country range from about 120 firms to over 800, amounting to 6667 firms for 2002 and 9655 for 2005. Since a substantial subset of the firms (about 2200 firms) is surveyed in both years, we also make use of panel data methods for estimating the relationships. Given that conditions in the transition countries were changing quite substantially between 2002 and 2005, we deem it important to look for changes in the results concerning the determinants and effects of association membership between the 2 years.

For measures of the firm’s participation in a business association and its functions, we make use of the firm’s responses to following questions: (1) “Is your firm a member of a business association or chamber of commerce?” and (2) “What services do you receive from the association or associations to which you belong, and what is the value of these services to your firm”? Responses were obtained from each firm on each of the following six functions or services: (a) lobbying government, (b) resolving disputes, (c) providing information and/or contacts on domestic product and input markets, (d) providing information and/or contacts on international product and input markets, (e) obtaining accrediting standards or other methods assuring product quality, and (f) providing information on government regulations. The value of each of these services to the firm was evaluated on a 0–4 scale.

Several questions in the survey deal with firm behavior with respect to unofficial payments and taxes. Given that firms are understandably reluctant to explicitly admit to making illicit payments and not paying taxes, the questions used for this purpose both by ourselves and others are: (1) “On average, what percent of total annual sales do firms like yours typically pay in unofficial payments/gifts to public officials?” and (2) “Recognizing the difficulties that many firms face in fully complying with taxes and regulations, what percent of total annual sales would you estimate the typical firm in your area of business reports for tax purposes?” Also included is a question regarding the perceived impact on the firm’s business of unofficial payments.⁸ (As mentioned below, we use this measure as an instrument for unofficial payments.)

Table 1 provides descriptive statistics for the aforementioned variables based on both the 2002 and 2005 surveys as well as those for a number of explanatory and control variables. The firm level data (on size, years in business, organizational form, ownership type, and assessments of the importance of different problems [dealing with labor regulations, licensing, customs procedures, titling of land and anti-competitive behavior]) and even including the firm’s assessment of macrolevel variables such as the property rights enforcement⁹ are all taken directly from the BEEPS surveys. The macrolevel variables identified in Table 1 are taken from other sources. Specifically, the index of control of corruption is taken from the International Country Risk Guide; the extent of (a) privatization and financial sector reform and (b) the internal liberalization of prices and markets, including the extent to which competition exists in the economy, are constructed by the same method as in De Melo, Denizer, and Gelb (1996), the data for 1996–2000 constructed by Fischer and Sahay (2000) and extended to 2000–2004 by Godoy and Stiglitz (2006) and Sukiassyan (2007) based on the EBRD Transition Reports 1999–2003. As can be seen from the heading of the table, the respective sample sizes for the BEEPS (2002 and 2005) were 6153 and 7156, respectively, 2231 of which were common to the two surveys constituting a panel. Because of some missing observations for some variables in either the BEEPS data or the macrolevel data sources in subsequent analysis, the sample sizes are reduced, in some cases quite substantially.

As indicated above, the popular perception of business associations in transition economies is that they are lobbying organizations involved in rent seeking and state capture. However, from the perspective of the firms in the BEEPS sample (as reported in the survey), on average, in both 2002 and 2005 sample firms assigned the “lobbying government” function the lowest value in usefulness to their firms of the six functions identified. In 2002, the highest value was assigned to the “information and/or contacts on international markets” function followed by the “information on and/or contacts on domestic markets,”“information on government regulations” and “accreditation or quality standards” functions.¹⁰ In 2005, the most highly rated function was “information on domestic markets” with minor changes in the rank order of the remaining functions. For both years, the evaluation shows the average rating on the highest rated function to be about double that of the lobbying function. Since these functions are largely determined by the individual association, we assume that they are in practice given to the firm, though in cases in which a given firm might have several associations to choose from, the individual firm could be expected to join an association that it believes best supplies the service which the firm feels is most important.

Table 1 provides evidence of rather serious underreporting of sales for tax purposes. In 2002, only 73% of sales were reported. Unofficial payments amounted to about 1.5% of total sales. Undoubtedly, both underreporting of sales and unofficial payments as a percent of sales are substantially underestimated as respondents would be reluctant to self-incriminate. What is likely to be more reliable is that both underreporting of sales and the magnitude of unofficial payments improved between 2002 and 2005. So, too, the decline in the firm’s assessment of the direct impact of unofficial payments on its business is probably also realistic. On the corruption front, therefore, the business environment was improving somewhat over the 2002–2005 period. This improvement may at least partially reflect the widely publicized campaigns in some transition countries to reduce corruption, either by leaders (new or old) or in other cases (e.g., Bulgaria) by grass roots organizations (Anderson and Gray, 2006).

As shown in the rows for aggregate indicators of institutional reform, the indicators for property rights enforcement, degree of privatization, extent of internal liberalization, and control of corruption were also increasing quite substantially between 2002 and 2005. Consistent with the increase on the privatization front were substantial declines in the share of sample firms that were state owned and large. Other indications in Table 1 of the improving institutional conditions were the declining need for courts to enforce business disputes,¹¹ importance of customs and trade regulations, business licensing and permits, judicial inefficiency, and even telecommunications as obstacles to business.

If, as we hypothesize, firms would be more likely to join associations and pay their membership fees the more strongly association services are “needed,” these gradually improving institutional conditions may also have contributed to the slight decline in the likelihood of association membership among firms in the samples. But, the changing composition of the sample in favor of smaller firms may also have contributed to this slight decline.¹²

Our estimates of the determinants of the selection decision to join a business association and the evaluation of its effects on unofficial payments and tax reporting are based on the following two equation model:

(1)

(2)

where S_ijt represents either the decision to join an association or the firm’s evaluation of the usefulness of its services by firm i in country j at time t, B_ijkt represents behavioral indicator k (unofficial payments or tax reporting) for firm i in country j and time t, X_it is a vector of time-varying firm characteristics, Z_jt is a set of time-varying country characteristics and the two error terms in each equation represent country fixed effects and random errors.

As shown in the descriptive statistics of Table 1, firms that are members of an association seem to make better use of, and evaluate more highly, some services provided by the association than others. A high score for a certain function by a certain firm could reflect either that this particular function offered by the association was especially useful to the firm or that the association(s) to which that firm belonged put more emphasis on that particular function. Since the individual associations are not distinguished in the survey, in practice, a high score would likely reflect a combination of these factors.

Other variables listed in Table 1 represent the various firm characteristics which we have hypothesized may induce firms to either join an association or make unofficial payments to remain informal and hidden from the authorities. Chief among these are large size, years in business, whether the firm is private (either a privatized state firm, an originally private firm, or another private firm), all of which would be expected to be positively related to association membership and tax payments but negatively related to unofficial payments. Some of the measures listed in the table, however, could be positively related to both association membership and unofficial payments. Examples of these would be the measures of the importance of various regulations such as those on labor, customs, licensing, judicial inefficiency, and the importance of anticompetitive behavior on the part of other firms. Finally, as indicated by condition (5) at the end of the previous section, we would expect favorable institutional developments such as the perception by firms that property rights are strong, or that the government has made efforts to control corruption or liberalize regulations of various sorts would encourage association membership and lower the need for unofficial payments.

Since as noted above, we believe it important to mitigate the problems arising from endogeneity or self-selection with respect to the influences of association membership and unofficial payments on tax payments. In identifying such instruments, the primary criterion we use is to find variables that could be considered a determinant of the variable to be instrumented (association membership or unofficial payment) but at the same time unlikely to exert any direct influence on the variable to be analyzed (e.g., in the first instance unofficial payments and later tax payments). With this criterion in mind, for instruments for association membership, we choose “membership in a business association is voluntary,”“experience with courts resolving disputes with customers,” and a dummy representing “a firm which has retained customers accounting for more than 20% of sales for the last three years.”¹³ Similarly for unofficial payments, we make use of the importance of customs, licensing, telecommunications, and contract violations. While making no claim that our instruments are ideal, they are at least sufficiently plausible and we verify the validity of our choice using the Amemiya-Newey-Lee test of overidentifying restrictions, the results of which are provided in the relevant tables below.

Due to the differing nature of the dependent variables and different degrees of controlling for potentially endogenous variables, several different estimation techniques are employed. As a dichotomous variable, the decision to join an association is estimated as a probit equation, while those for unofficial payments and tax reporting (both continuous but bounded) are estimated by either Tobit or IV Tobit. In each case, separate equations are estimated for 2002, 2005 and the smaller 2002–2005 panel, in the latter case with panel techniques.

IV. EMPIRICAL RESULTS

The first step in our analysis is to examine the determinants of membership in a business association. The variables included in our analysis are designed to capture the aforementioned five general conditions that would raise the firm’s demand for membership identified above. For example, reflecting the relative “needs” of firms for association membership are perceptions of problems on the part of firms, such as the firm’s assessment of the relative importance of various obstacles to its operations and growth (anticompetitive behavior of other firms, customs, licensing, obtaining titles or leasing land, and labor regulations) and judicial inefficiency. To reflect firm capabilities to take advantage of association services and to pay the membership dues, we include years in business, size (measured by a large firm dummy), the way the firm has been established (privatized, originally private, and other private, each relative to the excluded reference category state owned), and ability to retain its most important customers over a 3-yr period (reflecting reputation effects in the weak contract enforcement environment of most transition economies). To capture efforts to control corruption, we include the Index of Control of Corruption.

The probit estimates of the relevant parameters of Equation (1) are presented in Table 2. Estimates are provided for each of the three samples, the 2002 and 2005 BEEPS cross-sections and the smaller 2002–2005 panel, in columns (1)–(3) of the table. Those presented in column (4) are again for the panel but in this case estimated with country fixed effects.

Note that “years in business” and the large firm dummy—both measures of firm capability and inability to hide in informality with unofficial payments—are positively and significantly related to association membership in the estimates of all four columns. So too is the dummy for firm reputation (retaining customers accounting for at least 20% of its sales for at least 3 yr) for the year 2002 though not for either 2005 or the panel. Relative to the well-established state-owned enterprises (the excluded ownership type), those types with greater “needs,” namely, originally private, privatized, and other private firms, are as expected positively and generally significantly related to association membership.

As expected also, the coefficient of voluntary association membership is negative and significant in all four columns. Also, as expected, the coefficients estimated for favorable institutional characteristics like progress in privatization, internal liberalization, and corruption control at the macroeconomic level, and property rights enforcement at the firm level are all positive.

Note also that the coefficients of both the importance of anticompetitive behavior on the part of other firms and of judicial inefficiency are estimated to be positive and significant in all four columns of the table, while that of customs and trade regulations is positive and significant in all columns except column (3), the panel estimates. These results support the hypothesis that in these conditions firms would feel a stronger “need”—challenged either by monopolistic behavior on the part of other large firms or a judiciary that cannot be counted upon—for the relevant services of business associations. The same logic would apply to the importance of labor regulations, the importance of title or leasing, and the importance of business licensing and permits though in the first case the coefficient is positive and significant only in the first column of the table; in the second, the coefficient is negative and significant in column (2) for 2005; and in the third case, the coefficients remain insignificant in all columns. These findings could suggest that title or land-leasing problems could prompt firms to rely on unofficial payments as a more efficient alternative than association membership.

By and large therefore, the various probit estimates of Equation (1) for association membership support items (1), (2) and (5) of the rationale provided at the end of Section II. The results are also generally quite robust to the use of different data sets and (although not reflected in the table) to various experiments with different specifications of the explanatory variables.

Our second step is to present the Tobit estimates of our model for unofficial payments based on Equation (2). This equation includes as explanatory variables the association dummy as well as the country- and firm-level variables identified in Table 3. Panel (b) gives the results for the simple specification of the equation while panel (a) presents the results for the complete specification with and without instruments for member of business association variable. Given our hypothesis that association membership and unofficial payments are basically alternative strategies that firms choose between in trying to get what they need, the explanatory variables are virtually identical to those used for association membership in Table 2.

If as we suggest the two strategies are net substitutes with respect to many of these characteristics, we should expect (1) the coefficient of association member to be negative and those of many of the other explanatory variables in Table 3 to have signs opposite to those reported in Table 2 for association membership. Although from the first row of Table 3, we can see that the magnitudes of the estimated coefficients for association membership are quite sensitive to the data set and estimation procedure, qualitatively at least, they are quite consistent. Indeed, without exception, the estimated coefficients are all negative, and in all but two cases, they are also significant at least at the 5% level. One can also see several instances in which the signs of other coefficients in Table 3 are opposite to those in Table 2. This is true for years in business, large size firm, importance of titling, and importance of labor regulations (but only for 2002). The negative coefficients of some of these variables in Table 3 suggest that the stronger or more important are these influences the lower is the need for unofficial payments, while the positive coefficients of others suggest problems or characteristics that seem to induce unofficial payments. In most of these cases, the coefficients with opposite signs are also each statistically significant, implying that the comparable pairs of estimates in the two tables are also statistically different from each other.

Notice that, after controlling for the negative effect of association membership on unofficial payments, the coefficients of the variable representing establishment as an originally private firm are positive and significant in all three samples. Those for other private are positive and significant only in the 2002 sample and those for privatized firms positive and significant in 2002 and the panel but not in 2005. Comparing these findings with those for association membership in Table 2, originally private firms seem to choose unofficial payments over association memberships as their preferred strategy, while other private and privatized firms may use both strategies, though in some cases the dominant choice differing from one year to another. Whereas state-owned firms receive state protection virtually by definition and have little need for either strategy, it is other types of firms that have to solve their problems associated with the oppressive institutional and regulatory environment by joining associations, making unofficial payments or both.

By contrast to these various variables with opposite effects on association membership and unofficial payments, many of the variables representing perceived problems of the firms have effects of the same signs and significance in Tables 2 and 3. This is especially true for the importance of anticompetitive behavior of other firms and judicial inefficiency where the positive coefficients indicate that the problems are sufficiently deep seated that firms see the need to use both strategies at the same time.

Our final step is to present the estimates of the other variant of Equation (2), the Tobit estimates of the percent of sales reported for tax purposes, that is, an inverse measure of tax evasion. The specifications used in estimating this variant of Equation (2) are somewhat more inclusive since in general we allow for variables to enter that have been included in both Tables 2 and 3 as well as some additional variables deemed relevant to the aforementioned “tax morale” in previous empirical work on the subject (Anderson and Gray, 2006; Gordon, 1989). These include “the extent of the direct impact of unofficial payments on the firm’s business,” importance of judicial inefficiency, and property rights enforcement. The first two of these would be expected to make people feel negatively about the state and therefore to have a negative influence on tax reporting. The third would reflect positively on the state and hence on firms’ willingness to pay their taxes.

Once again, panel (b) of Table 4 reports the simple estimations results and panel (a) the results with the full set of controls, with and without instrumenting the two potentially endogenous variables, namely, member of business association and unofficial payments. Since with the full specification there is much greater potential for collinearity problems, if one were to neglect the endogeneity issue, in this case the simple specification results in panel (b) may deserve more attention. According to the results in panel (b) unofficial payments, as hypothesized have negative and significant effects on tax reporting in all three samples, while association membership has positive effects in all three, but statistically significant at the 10% level in only one, the 2005 sample. The negative effect of unofficial payments is confirmed in all columns of panel (a) of the table as well, though in several of the cases the coefficients are not statistically significant.

On the basis of the hypothesis that unofficial payment and association membership strategies are net substitutes, we might also expect some matching of the signs of parameter estimates in panel (a) of this table with those in either Table 2 or Table 3. Note that there are at least a few cases that support the positive link between those for association and tax reporting (Tables 2 and 4) and negative links between those in Tables 3 and 4. For example, the coefficients of Property Rights Enforcement are generally positive and significant in both Table 2 (for association member) and Table 4 (for tax reporting) but negative and significant in Table 3 (for unofficial payments). There is a similar pattern in the case of the importance of Labor Regulations but the results are less consistent and less significant. The effects of Control of Corruption are somewhat similar, though again much weaker in the case of tax reporting.

Finally, Table 5 provides separate estimates for the effects of each of the six different functions of association on the unofficial payments and tax reporting variables. Since the Tobit estimation procedure is applied to each of these functions, 0 values are assigned to all firms that are not association members. Two sets of specifications are again employed, one with the set of control variables listed in Table 2 and, the other without any controls.¹⁴

Turning first to the results for unofficial payments, the results are consistently negative and significant across the different association functions for 2002 and for the panel but, though still negative, are usually insignificant for the 2005 sample. Among the different functions, the magnitudes are in most cases higher for information on government regulations and in the panel sample also for the dispute resolution function than for the lobbying function. In the case of tax reporting, the dispute resolution and information on government regulations functions seem to have a more decisive edge over lobbying than in the case of unofficial payments. These results generally complement the finding of Table 1, where the lobbying function was shown to be rated lowest of all the functions. In these regression results, the effects of the lobbying functions are somewhat stronger and in line with those of association membership in general in Table 4. Even in this case, the effects are generally weaker than for the dispute resolution and information on government regulations functions.

V. CONCLUSIONS

In general, the empirical results from transition economies support several of the hypothesized relationships: (1) The positive link between unofficial payments and tax evasion (implicit in a negative relation between unofficial payments and tax reporting), (2) the negative relation between association membership and unofficial payments, (3) that association membership and unofficial payments are, on balance, substitute rather than complementary strategies for firms in dealing with the problems facing them in transition economies, and (4) the positive relationships between favorable institutional characteristics and tax reporting. Moreover, qualitatively at least, these results are reasonably robust to the use of different samples, specifications, and estimation techniques. Indeed, we have experimented with other specifications (not reported here for reasons of space) that also yield similar findings. Our empirical results also confirm the results set forth in the theoretical model that there are increasing returns to scale in terms of firm size to association membership and decreasing returns to scale to making unofficial payments. Our empirical results confirm that larger firms are more likely to take up association membership, while smaller ones are more inclined to be making unofficial payments and evading taxes.

On the basis of these and other findings, we go on to suggest some possible policies for reducing unofficial payments and increasing tax compliance. First and foremost, we suggest that in the relatively oppressive conditions for businesses in transition economies, encouraging business associations of the type that can help firms deal with the problems facing them may be one important and virtually totally overlooked means of accomplishing these objectives. In this way, firms can achieve greater formality and the benefits thereof than in the case of unofficial payments. At the same time, governments also benefit from this strategy by being able to obtain greater tax revenues than when firms deliberately stay small and informal to avoid taxes. In democratic countries, this can also help government leaders to be reelected. As our theoretical and empirical findings suggest, membership in business associations is more productive than making unofficial payments in that the latter strategy induces tax underreporting and diverts the payments made from needed public goods.¹⁵

Second, and as a complement to the first recommendation, we have shown that the informational and dispute-reducing functions of associations are at least as highly valued and effective in reducing unofficial payments and raising tax reporting than the lobbying function with which associations in transition countries have been most closely identified. Hence, it should be quite clear that when associations can be induced to concentrate on supplying the specific services that their members value most highly, associations could play an even more positive role in affecting firm behavior in these respects than have associations to date.

Third, from Table 3, the degree of internal liberalization has the expected negative effect on Unofficial Payments but not the expected positive effect on Tax Reporting from Table 4. The index of control of corruption has a negative significant effect on making unofficial payments in Table 3 but a positive effect on tax reporting for 2005 and the panel in Table 4. Hence, depending on which problem (tax underreporting or unofficial payments) appears to be more serious, priorities in liberalization and control of corruption may be identified.

Fourth, as suggested by the Tax Morale literature, transition governments may boost tax compliance by providing institutions, policies, and infrastructure that firms appreciate and which reflect favorably on the effectiveness of the state, such as enhancing judicial efficiency, enforcing property rights, and providing appropriate infrastructure.

Since many of the same conditions identified here, such as sudden changes in political conditions giving rise to sharp changes in policies and institutions, apply to developing countries, some of these same recommendations would seem to apply to developing countries. For example, concerning business associations, Donner and Schneider (2000) and Schneider (2005) point out some of the characteristics and circumstances in which associations can be especially useful. For example, associations in both Chile and Costa Rica have played a very positive role in establishing broad-based government-business councils that provide rather continuous dialogue between business and government and which have proven to maintain more stable economic policies in the face of sometimes rather abrupt political changes.

However, some important caveats are in order. First, since as we have seen conditions vary sharply across the 25 transition countries in our sample and have been changing relatively rapidly, further research would be needed to validate these recommendations at the present time in individual countries. A longer and larger panel would be needed to reduce remaining possible sources of bias due, for example, to relations between time-varying omitted variables and the included explanatory variables. Most importantly, it would be highly desirable to complement the BEEPS with a survey of the specific associations with which sample firms are affiliated so as to explain the choices of specific association memberships and differentiate them by the strength of the services that each provides.