Interest groups, democracy, and policy volatility.
Empirical evidence suggests democratic polities produce more stable policy than do autocracies. For example, Henisz (2004) finds that democracy is associated with more stable fiscal policy, and Dutt and Mobarak (2007) find that democracy is associated with more stable fiscal and trade policy. The literature has focused attention on the decentralized decision-making that characterizes democracy as the causal factor linking political regime-type to policy stability. Checks and balances and the presence of veto-players in more democratic polities, in particular, are thought to constrain policy-makers and thereby produce policy stability (Dutt and Mobarak 2007; Lindblom 1959; Lindblom 1965; Tsebelis 1995).
In this paper, we explore an alternative source of the policy stability observed in democracies: special interest groups. As noted by Freedom House, a cornerstone of democracy is the granting of both political rights and civil liberties. Under their definition, the latter include associational and organizational rights, which promote civil society. However, such rights also enable the proliferation of special interest groups which may seek to influence policy. Indeed, studies have shown that distributional coalitions, or Olson-groups as they are sometimes called (based on Olson 1982), are positively correlated with the degree of democracy (Coates, Heckelman, and Wilson 2007; Murrell 1984).
According to Olson (1982), distributional coalitions impede a society's ability to redirect resources as needed in response to changing conditions. This occurs both in markets by erecting barriers to entry and stifling innovations, and in the public sphere by lobbying and influencing policy-makers. Mokyr (2000) and Parente and Zhao (2006) have studied the former, but the latter has not yet been directly tested. We aim here to bridge the gap by noting the connection between interest groups and policy stability.
Olson argued that special interest groups tend to form and accumulate in societies with stable institutions. In the aftermath of institutional upheaval, old groups may be destroyed and new groups may form. For a period of time, one might expect the group formation process to contribute to policy instability, as the power of old groups diminishes in a new institutional environment and new groups seeking special favors engage politicians. However, once the formation process is complete, existing groups are expected to use their privileged positions to preserve the status quo in order to protect their rents. We therefore hypothesize that the status quo bias of groups may manifest in a more limited range of policy options and changes over time. Thus, interest groups are expected to reduce policy volatility.
A finding consistent with our hypothesis contributes to the literature in several ways. In particular, our hypothesis identifies a previously unexamined mechanism to explain policy stability. Moreover, if our hypothesis is correct, then prior literature has likely overstated the role of decentralized decision-making (checks and balances and veto-players) in policy stability. The relationship between democracy and policy stability may have less to do with democracy's decentralized decision-making and more to do with the large number of interest groups that accumulate in democracies and the natural tendency of such groups toward preservation of the status quo.
The stability associated with democracy is commonly construed as beneficial, but it is important to draw a distinction between institutional stability and policy stability. In the case of policy, the normative implications of stability are ambiguous. In particular, to the extent that policy stability reduces uncertainty and risk and thereby spurs productive investment, its normative implications are positive. Such positive normative effects may occur irrespective of the mechanism that generates the stability--checks and balances and veto-players or interest groups. However, to the extent that policy stability implies that warranted change does not occur, as might be the case in the context of a shock or when existing policy favors certain interests unfairly, its normative implications are negative, again irrespective of the mechanism that generates it. Although we might like to believe that democratic institutions limit the tendency toward favoritism, the vast literature on rent-seeking (Tullock 1967) and bureaucratic capture (Stigler 1971) suggests otherwise. In any case, our focus here is not on determining the net normative effect of groups on policy stability, but on determining if the previously documented democracy-policy stability connection is direct, as previously presumed, or occurs only (or mainly) through democracies' tendency to foster interest groups.
We proceed by first establishing that the previously identified stabilizing impact of democracy on policy is evident in the data we examine. We then examine whether interest groups are also associated with policy stability. The findings reveal that groups are indeed associated with policy stability. Moreover, the findings suggest that interest groups mediate the stabilizing impact of democracy on policy. The relationship between interest groups and policy volatility is found to be robust with respect to some measures of policy. We also find that interest groups have a more consistent effect in Organization for Economic Cooperation and Development (OECD) economies than in developing nations. Finally, we find that the impact of interest groups on policy volatility depends on the degree of polarization or fractionalization in a society, suggesting that competition among groups may mitigate their impact.
The next section of the paper describes the data and methods. Results are presented in Section III. Concluding remarks are offered in Section IV.
II. DATA AND METHODS
The dataset is taken from Heckelman and Wilson (2014). (1) It consists of a maximum 791 observations on an unbalanced panel of 159 countries over six 5-year periods, 1973-1977, 1980-1984, 1985-1989, 1995-1999, 1999-2003, 2002-2006. (2) All countries/years for which data are available are included in the analysis. Descriptions and sources for all variables are provided in the Appendix.
The dependent variables are three measures of policy volatility. Given the scope for endogeneity bias as well as the need for a time series long enough to compute a measure of volatility, we are limited to measures of policy that are available for a large cross-section of countries over a long period of time. We therefore examine the standard deviation of the share of government purchases in gross domestic product (GDP), as a measure of fiscal policy volatility, as well as the standard deviations of the share of imports in GDP and of the share of exports in GDP, as measures of trade policy volatility. Dutt and Mobarak (2007) examine similar measures of fiscal and trade policy. Moreover, fiscal and trade policy are likely of interest and importance to distributional coalitions.
In the case of fiscal policy, the expanded size and scope of government in the post-WWII era is commonly believed to have produced ample opportunity for pursuit of privileged transfers and tax treatment by rent-seeking groups (see, e.g., Mueller 2003). With respect to trade policy, the large "protection for sale" literature, initiated by Grossman and Helpman (1994), indicates that trade policy is subject to manipulation by special interests. Because interest groups may have different effects on exports and imports, we examine volatility in the share of exports in GDP and in the share of imports in GDP separately, rather than examining the sum of the shares. For example, the mercantilist nature of popular discussion about trade policy suggests that some special interests may favor policy that encourages exports, but limits imports. Moreover, while it is rare for a domestic interest group to oppose exports, distributional coalitions may have countervailing interests vis-a-vis imports. In particular, while unions and producers of final goods may favor import restrictions, producers who utilize intermediate products may oppose them. Thus, special interest concerns regarding exports and imports may not only be distinct, the extent to which interests within each of those categories are aligned or opposed may differ.
We estimate a baseline model and an extended model. The baseline model is parsimonious, and supposes that the extent of democracy, the degree of political instability, and the influence of distributional coalitions are the primary determinants of policy volatility. We use the annual average of the Freedom Flouse index of political rights to capture the extent of democracy. (3) (As is common, we invert the index so that it is increasing in the extent of democracy.) Traditionally, political instability has been measured by variables such as the number of revolutions and coups, as in Barro (1991). We follow Dutt and Mitra (2008) and construct a measure of movements in and out of democracy that reflects changes in regime-type in order to capture political instability. We use the initial value of the natural log of the number of interest groups in a country to capture the extent of influence of distributional coalitions. (4) Descriptive statistics are presented in Table 1 for our main variables of interest.
The extended model includes variables selected primarily in order to control for the possibility of spurious correlation between groups and policy volatility. First, GDP per capita and population are included as controls for a country's level of development and size. Neither of these variables is generally understood to be a direct determinant of policy volatility. However, both are highly correlated with the number of groups observed in nations (Bischoff 2003; Coates, Heckelman, and Wilson 2007; Murrell 1984), and it may be that they are correlated with unobserved determinants of policy volatility. It thus seems important to examine whether any relation between groups and policy volatility is independent of any relation between policy volatility and a country's level of development and size. Second, the first two moments of each country's growth process are also included, average growth in GDP per capita and the standard deviation of growth in GDP per capita. The literature on the impact of groups on economic activity suggests that these moments of the growth process may, in part, be caused by groups (Coates, Heckelman, and Wilson 2011; Heckelman and Wilson 2014). They may also be associated with policy volatility, necessitating their inclusion as controls in order to identify whether the direct relation between groups and policy volatility that we have hypothesized exists. Third, we control for diversity in society. Diversity has been associated with group formation (Coates, Heckelman, and Wilson 2007), and while earlier studies have not examined it as a source of policy volatility, it has been associated with the kinds of policies a country adopts (e.g., Alesina, Baqir, and Easterly 1999; Alesina et al. 2003; La Porta et al. 1999). Fourth, the extended specification includes the average level of the policy measure under consideration. The first moment of a policy measure may be a determinant of the second (policy volatility). Moreover, if the second moment is determined by groups the first may be as well, necessitating a control for the first moment to determine whether a direct relation between groups and policy volatility exists. Finally, all specifications include a constant and time period dummies. Transition economy and oil exporter dummies are also included, to account for the possibility that policy volatility may systematically differ in such countries.
The estimated models, both the baseline and extended versions, contain potentially endogenous and weakly exogenous explanatory variables. In particular, democracy, GDP growth, and GDP growth volatility are measured over the entire span of each time period. Moreover, each of these variables may not only cause, but also be caused by, policy volatility, and may therefore be endogenous. The number of interest groups and GDP per capita are both predetermined variables, measured as of the first year of each time period. While endogeneity within a time period is thus ruled out by construction, these variables may only be weakly exogenous across time periods. We assume the remaining variables (political instability, population, and the level of policy) are exogenous.
In addition to multiple non-exogenous variables, our data also consists of a small time dimension (6) and a reasonably large cross-sectional dimension (159). The Arellano-Bover/Blundell-Bond System generalized method of moments (GMM) estimator is particularly useful in this kind of context (Roodman 2009a), and we therefore employ it to estimate both our baseline and extended specification. We employ heteroskedasticity-corrected, country-clustered, Windmeijer-corrected standard errors. Orthogonal deviations are used to transform the data to remove fixed effects. Finally, weighted estimation is used, with real GDP per capita serving as the weighting series, since poor data quality in poorer countries may be associated with error that increases measured policy volatility and decreases measured interest group counts in such countries relative to their true values.
System GMM estimation is especially helpful in a context such as ours with multiple endogenous regressors as well as predetermined but not strictly exogenous regressors, as internal instruments are available with this method that are not available with the method of instrumental variables. Although external instruments are not required, they are allowed, and we include several. Specifically, we include the legal origin variables of La Porta et al. (1998, 1999) and the Muslim share of the population as excluded instruments for democracy. We include the date of independence and a dummy variable for OECD membership prior to 1973 as excluded instruments for interest groups. La Porta et al. (1998) show a strong correlation exists between a country's legal origin and its allowance for political rights. Mobarak (2005) suggests the use of the share of the Muslim population as an instrument for democracy in the context of an analysis of growth volatility, and offers several possible reasons to think democracy and the share of the Muslim population may be related. The use of the year of a country's independence and an indicator of early OECD membership as instruments for interest groups are both motivated by Olson (1982). According to Olson, institutional stability will over time lead to the accumulation of groups. If Olson is correct, and if the length of time since independence as well as early membership in the OECD are indicative of institutional stability, these variables should serve as valid instruments for groups during the time period over which we measure policy volatility--1973-2006.
If instruments (internal and external) are to be valid, they must be both relevant and exogenous. Exogeneity is assessed with the typical System GMM Hansen and AR(2) tests. A drawback of System GMM estimation is that while tests of the exogeneity of instruments are available, a single test that evaluates the joint strength of the instrument set is not. We therefore assess whether the instruments employed are weak using statistics based on the first stage of an estimation of the model using the method of instrumental variables.
In Table 2, the findings for the parsimonious specifications in columns (1), (3), and (5) indicate that more democratic countries experience less volatility in the measures of both fiscal and trade policy, consistent with the findings of Dutt and Mobarak (2007) and with the notion that the decentralized decision-making that characterizes democracy produces stability in policy. In columns (2), (4), and (6), we add the control for interest groups. (5) The findings indicate that countries with more groups experience less policy volatility, consistent with the notion that groups favor the status quo and thus seek stability. In addition, the findings indicate that the relation between democracy and policy volatility is not robust to the inclusion of the control for interest groups, suggesting that interest groups mediate the relation and may be a key source of the policy stability observed in more democratic countries. (6)
Table 3 contains results for specifications that include the additional controls, to explore the robustness of the relation between interest groups and policy volatility. The findings indicate a robust relation between interest groups and both fiscal and import volatility. However, the relation between interest groups and export volatility is not robust.
Distributional groups may have more divergent interests vis-a-vis imports than exports. For example, imports may damage domestic producers in import-competing final goods sectors yet benefit domestic producers that rely on imported intermediate goods. Exports though, except through general equilibrium effects on factor prices, are unlikely to harm existing industries. As such, there may be greater pressure to protect the status quo vis-a-vis imports than exports, and it is perhaps not surprising that the findings reveal a relation between groups and import volatility but not a (robust) relation between groups and export volatility.
In the lower section of Tables 2 and 3, we report findings that address the validity of the instruments used in the analysis. The results of Hansen tests are consistent with instrument exogeneity. In the case of fiscal volatility though, the Arellano-Bond AR(2) tests reveal evidence of autocorrelation that is not consistent with instrument validity. We assess the strength of the instruments used in the levels equation of the System GMM estimation by first estimating the levels equation using instrumental variables. The System GMM estimator uses both lagged differences as well as excluded instruments as instruments for the levels equation. As such, we use the first lagged-difference as an instrument as well as the excluded instruments listed in the Appendix and discussed above. We then examine the Cragg-Donald statistic and the Kleibergen-Paap rk statistic, from the first stage of the estimation. The Cragg-Donald statistic is the multivariate analog to the F statistic in the case of multiple endogenous variables. The KleibergenPaap statistic is the robust analog in the case of non-i.i.d. errors. Critical values do not exist for the Kleibergen-Paap statistic. Baum, Schaffer, and Stillman (2007) suggest using the Stock and Yogo (2005) critical values tabulated for the Cragg-Donald statistic as well as the Staiger and Stock (1997) "rule of thumb" value of 10. The statistics are reported in the lower portion of Table 2. The statistics are all well above the "rule of thumb" value of 10. In the case of export volatility (columns (3) and (4)), both statistics also exceed the Stock and Yogo critical values at the 5% level. In the case of fiscal volatility and import volatility (columns (1) and (2) and (5) and (6)), the Cragg-Donald statistic exceeds the Stock and Yogo critical value at the 5% level, but the Kleigergen-Paap rk statistic exceeds at only better than the 10% level.
In Table 4, we report results of additional sensitivity analysis. Roodman (2009b) points out that instrument proliferation is a common problem with System GMM estimation. Although the number of instruments used in the Table 3 specifications is well below the number of countries in the panel, suggesting our specification does not suffer from "too many" instruments, we nonetheless re-estimate with a reduced number of instruments. Findings are reported in panel A of Table 4, and indicate that the interest groups result is robust to this change. We also examine whether the interest groups result changes substantively if we exclude the countries with the most groups from the sample. These countries tend to be democracies with relatively stable policy. They also tend to be rich countries, which are weighted more heavily in the analysis due to concerns about measurement error in the data for poorer countries. These observations could be driving the results in Table 3, but the findings in panel B of Table 4 indicate this is not the case. The relation between interest groups and policy volatility is robust to the exclusion of these observations. We also examine whether the findings are robust to the exclusion of two time periods--the 1980-1984 period and the 1999-2003 period. The groups counts in 1980 are systematically larger than counts in other periods, due to the inclusion of "local" groups in this edition of the primary source of the data. Time dummies included in the main analysis should control for this difference, but we nonetheless check for robustness to the exclusion of this time period. In this case, the findings do reveal some sensitivity. In particular, in the full specifications, there is not a statistically significant relation between interest groups and policy volatility. The 1999-2003 cross-section overlaps with the surrounding periods which has the potential to produce autocorrelation in the error term of our regressions. Tests revealed no evidence of autocorrelation but we also ran additional regressions dropping this period for completeness. When we exclude the 1999-2003 period, the relation between interest groups and fiscal volatility remains statistically significant, but the relation between interest groups and import volatility does not. The results are thus somewhat sensitive to the exclusion of these two time periods. Finally, we examine whether the relation between interest groups and volatility differs in OECD and non-OECD countries. The findings in panel E indicate that it does not, with the exception of the case of export volatility. A t-test of equality between the coefficients on interest groups in OECD countries and non-OECD countries indicates no statistically significant difference in the cases of fiscal and import volatility. In the case of export volatility, while the findings in Table 3 indicate no relation, these findings indicate that interest groups are associated with export stability in OECD countries, though not in non-OECD countries.
IV. ETHNIC FRAGMENTATION AND GROUPS
Finally, we examine whether the relation between groups and policy volatility is conditional on the extent of ethnic fragmentation in society. Our measure of interest group activity is the number of interest groups. While this measure arguably captures the potential for lobbying, it may not capture the type of lobbying that takes place. While all interest groups may compete over a limited economic pie, some group interests will coincide with other groups (e.g., all firms, regardless of industry, may prefer low corporate taxes) while other groups will have opposing interests (e.g., the logging industry benefits from tariffs on timber whereas housing construction industries are harmed.) We presume that in more ethnically fragmented societies, additional groups which form are more likely to represent divergent or opposing interests rather than common or coincident interests, leading to differences in policy preferences. Thus, in more ethnically fragmented societies, policy may be more likely to fluctuate over time as more lobbying will entail attempts to overturn the status quo successfully implemented by past lobbying of other groups. Put differently, fragmentation may proxy for the extent to which groups compete with one another. The more fractionalized is a society, the more groups are expected to compete with one another, and the less influence groups are expected to have on policy. In this case, we expect the stabilizing impact of groups on policy volatility to be offset by the extent of fragmentation. (7)
This hypothesis suggests that the marginal impact of groups on policy volatility is conditional on the extent of social fragmentation. We test this notion by interacting our interest group variable with a measure of social fragmentation. We consider two measures of fragmentation--ethnolinguistic fractionalization and ethnolinguistic polarization. Both measures are taken from Desmet, Ortuno-Ortin, and Wacziarg (2012), who use language trees and linguistic differentiation to construct indexes of social fragmentation. The fractionalization index represents the probability that two randomly chosen people belong to different social groups, and is maximized when each individual belongs to a different group. In contrast, polarization is maximized when there are two groups of equal size.
Findings are reported in Table 5, and reveal mixed evidence of the sort of hypothesized competition among groups in fragmented (via fractionalization or polarization) societies. We focus the discussion on the polarization findings, as significance levels are always higher in the case of polarization than in the case of fractionalization. In column (1), the interest group coefficient implies that in a perfectly homogenous society (Polarization = 0), interest groups reduce fiscal policy volatility, although not to a statistically significant extent. The coefficient on polarization implies that in the absence of any interest groups (Interest Groups = 0), greater polarization leads to greater fiscal volatility, perhaps as policy makers look to placate a greater variety of perceived interests on their own. Competing interest groups (Polarization*Groups) reduce fiscal volatility, as indicated by the negative coefficient associated with the interaction term.
The marginal impact of groups is not directly discernable from the coefficient estimates in Table 5, as it is conditional on the level of fragmentation. We therefore plot the conditional marginal effect of groups in Figure 1. As depicted in the top graph of Figure 1, the conditional marginal impact of groups on fiscal volatility is statistically significant for all but the least fragmented societies. (8) This may be because the additional competing groups are able to block the lobbying efforts of other groups. Thus, group formation may be more complete in the Becker (1983) sense, where groups compete upfront and only the most efficient win (consistently), rather than groups alternating in the success of their lobbying efforts as proposed above. In the case of export volatility, the positive coefficient on the interaction term in column (3) of Table 5 reveals that greater polarization reduces the stabilizing impact of groups as originally expected. Still, the overall marginal impact of groups remains negative (reducing volatility) for all levels of polarization (see the middle graph of Figure 1). The conditional marginal impact of groups is found to be statistically significant for all but the highest levels of polarization. More specifically, the impact is statistically significant in all but three countries in the available sample. (9) Likewise, in the case of import volatility, greater polarization reduces the stabilizing impact of groups, but the marginal impact of groups remains negative (stabilizing) for all levels of polarization. The conditional marginal impact of groups is statistically significant for all but the largest levels of polarization. More specifically, the impact is statistically significant in all but four countries in the available sample. (10)
V. CONCLUDING REMARKS
In this paper we have examined the relationship between democracy, interest groups, and policy volatility. Previous studies have identified an inverse relationship between democracy and policy volatility, and postulated this occurs due to the checks and balances and decentralization endemic to democratic institutions. Our dataset supports this finding for all three measures of policy we test--fiscal, exports, and imports. But we find that the effect of democracy is attenuated by the presence of interest groups. Democracies tend to foster the proliferation of interest groups, and in line with our expectations, we find that interest groups directly reduce the variation in fiscal and trade policy, and controlling for interest groups negates the effect of democracy. Thus, the results suggest that democracy only indirectly influences policy volatility, through the promotion of interest groups rather than directly from the characteristics of democratic institutions. The finding is most robust in terms of fiscal policy. Trade policy results differ somewhat between exports and imports, with the latter more robust than the former.
We also find that interest groups have significant effects in both OECD and non-OECD economies, except for export volatility which only manifests among the OECD economies. Finally, ethnic polarization, which may proxy for competition among groups, partially offsets the policy stabilizing impact of interest groups. This effect is the strongest for import volatility, but of the opposite direction for fiscal policy. On net, though, interest groups significantly reduce volatility in all three categories.
The differing effects of polarization between fiscal and trade policy may be due to differences in group goals. An alternative not tested here would incorporate diversity and competition in industry, rather than population. The listing of groups in the World Guide to Trade Associations (used to construct the aggregate group counts for the original dataset) is stratified by sector as well as by country, which would allow for the development of measures for industry fragmentation. A dataset comprised of sector and country-specific counts of groups could then be used to analyze the relation between sector-level interest group activity and volatility. For example, manufacturing groups are probably more directly interested in trade policy than are banking interests.
As noted in the introduction, the normative implications of policy stability induced by interest groups are ambiguous. If policy stability reduces uncertainty and thereby stimulates productive investment, the stabilizing influence of groups may be good. If group-induced policy stability is the result of successful efforts by groups to block rent extraction by politicians, the effect might again be understood to be good. In contrast, if group-induced policy stability is the result of groups' successful efforts to maintain the status-quo in the context of socially enhancing policy change, their impact may well be adverse. While our study identifies groups as a source of policy stability, our aggregated analysis does not allow us to discern the particular mechanism via which groups generate policy stability. It therefore does not allow us to conclude whether the stabilizing impact of groups is good or bad on net. Future research might be directed toward efforts to empirically identify the mechanism via which groups generate policy stability in order to facilitate normative policy recommendations.
CIA: Central Intelligence Agency
GDP: Gross Domestic Product
GMM: Generalized Method of Moments
OECD: Organization for Economic Co-operation and
VARIABLE DEFINITIONS AND DATA SOURCES
The dataset is an unbalanced panel of a maximum 791 observations that covers 159 countries over six 5-year periods, 1973-1977, 1980-1984, 1985-1989, 1995-1999, 1999-2003, 2002-2006.
Fiscal Volatility. Standard deviation of the annual share of general government final consumption expenditures in GDP. Source: World Bank's World Development Indicators.
Export Volatility. Standard deviation of the annual share of exports in GDP. Source: World Bank's World Development Indicators.
Import Volatility. Standard deviation of the annual share of imports in GDP. Source: World Bank's World Development Indicators.
INDEPENDENT VARIABLES OF PRIMARY INTEREST
Interest Groups: natural logarithm of the number of interest groups in a country, in 1973, 1980, 1985, 1995, 1999, and 2002. Source: First through sixth editions of World Guide to Trade Associations.
Democracy: annual average of a political rights index of the degree of freedom in the electoral process, political pluralism and participation, and functioning of government. The original index ranges from one to seven, and is decreasing in extent of democracy. We invert the index so that it is increasing in the extent of democracy. Source: Freedom House. "Freedom in the World Country Ratings 1972-2011
ADDITIONAL CONTROLS IN ALL SPECIFICATIONS
Time dummies: dummy variables indicating the years 1973, 1980, 1985, 1995, and 1999.
Transition economy dummy: dummy variable indicating transition economies--Albania, Armenia, Belarus, Bulgaria, Cambodia, China, Croatia, Czech Republic, Georgia, Hungary, Kazakhstan, Laos, Latvia, Lithuania, Macedonia, Moldova, Poland. Romania, Russia, Slovakia, Slovenia, Tajikistan, Ukraine, Uzbekistan, Vietnam.
Oil: dummy variable for oil exporters. Source: Global Development Network Growth Database.
GDP: natural logarithm of initial value of real GDP. Source: World Bank's World Development Indicators.
Population: natural logarithm of initial value of population. Source: World Bank's World Development Indicators.
Growth Volatility: standard deviation of annual real per capita GDP growth. Source: World Bank's World Development Indicators.
Growth: average annual growth rate of real per capita GDP. Source: World Bank's World Development Indicators.
Fractionalization: a measure of ethnolinguistic fractionalization (Tables 3 and 4) and a measure of ethnolinguistic polarization (Table 4). Both measures are available at various levels of aggregation. We use the measures with the least aggregation (ELF(15) and POL(15)). Source: Desmet, Ortuno-Ortin, and Wacziarg (2012).
Policy Level: average of the annual share of general government final consumption expenditures in GDP (Fiscal Volatility regressions); average of the annual share of exports in GDP (Export Volatility regressions); average of the annual share of imports in GDP (Import Volatility regressions). Source: World Bank's World Development Indicators.
Independence: Date of independence. Source: Encyclopedia Britannica.
OECD: Dummy variable indicating OECD membership prior to 1973. Source: OECD.
Muslim: Dummy variable indicating majority Muslim population. Source: World Christian Encyclopedia.
British Law: Dummy variable indicating British legal origin. Source: La Porta et al. (1998, 1999); updated with Central Intelligence Agency (CIA) World Factbook.
French Law: Dummy variable indicating French legal origin. Source: La Porta et al. (1998,1999); updated with CIA World Factbook.
German Law: Dummy variable indicating German legal origin. Source: La Porta et al. (1998,1999); updated with CIA World Factbook.
Scandinavian Law: Dummy variable indicating Scandinavian legal origin. Source: La Porta et al. (1998,1999); updated with CIA World Factbook.
Alesina, A., R. Baqir, and W. Easterly. "Public Goods and Ethnic Divisions." The Quarterly Journal of Economics, 114(4), 1999, 1243-84.
Alesina, A., A. Devleeschauwer, W. Easterly, S. Kurlat, and R. Wacziarg. "Fractionalization." Journal of Economic Growth, 8, 2003, 155-94.
Barro, R. "Economic Growth in a Cross Section of Countries." Quarterly Journal of Economics, 106(2), 1991, 407-43.
Baum, C., M. Schaffer, and S. Stillman. "Enhanced Routines for Instrumental Variables/Generalized Method of Moments Estimation and Testing." The Stata Journal, 7(4), 2007,465-506.
Becker, G. "A Theory of Competition among Pressure Groups for Political Influence." Quarterly Journal of Economics, 98(3), 1983, 371-400.
Bischoff, 1. "Determinants of the Increase in the Number of Interest Groups in Western Democracies: Theoretical Considerations and Evidence from 21 OECD Countries." Public Choice, 114(1/2), 2003, 197-218.
Coates, D., J. C. Heckelman, and B. Wilson. "Determinants of Interest Group Formation." Public Choice, 133(3-4), 2007, 377-91.
--. "Special-Interest Groups and Growth." Public Choice, 147(3-4), 2011, 439-57.
Desmet, K., I. Ortuno-Ortin, and R. Wacziarg. "The Political Economy of Linguistic Cleavages." Journal of Development Economics, 97, 2012, 322-38.
Dutt, P., and A. Mobarak. "Democracy and Policy Stability." INSEAD Business School Research Paper No. 2007/50/EPS, INSEAD, Fontainebleau, France, 2007.
Dutt, P., and D. Mitra. "Inequality and the Instability of Polity and Policy." The Economic Journal, 188(531), 2008, 1285-314.
Grossman, G., and E. Helpman. "Protection for Sale." American Economic Review, 84(4), 1994, 833-50.
Heckelman, J. C., and B. Wilson. "Interest Groups and the 'Rise and Decline' of Growth." Southern Economic Journal, 81(2), 2014, 435-56.
Henisz, W. "Political Institutions and Policy Volatility." Economics and Politics, 16(1), 2004, 1-27.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. "Law and Finance." Journal of Political Economy, 106(6), 1998, 1113-55.
--. "The Quality of Government." Journal of Law, Economics, & Organization, 15(1), 1999, 222-79.
Lindblom, C. "The Science of Muddling Through." Public Administration Review, 19, 1959, 79-88.
--. The Intelligence of Democracy: Decision Making Through Mutual Adjustment. New York: Free Press, 1965.
Mobarak, A. "Democracy, Volatility and Economic Development." Review of Economics and Statistics, 87(2), 2005, 348-61.
Mokyr, J. "Innovation and Its Enemies: The Economic and Political Roots of Technological Inertia," in A Not-So-Dismal Science, edited by M. Olson and S. Kahkohen. New York: Oxford University Press, 2000, 61-91.
Mueller, D. Public Choice III. Cambridge: Cambridge University Press, 2003.
Murrell, P. "An Examination of the Factors Affecting the Formation of Interest Groups in OECD Countries." Public Choice, 43(2), 1984, 151-71.
Olson, M. The Rise and Decline of Nations: The Political Economy of Economic Growth, Stagflation, and Social Rigidities. New Haven, CT: Yale University Press, 1982.
Parente, S., and R. Zhao. "Slow Development and Special Interests." International Economic Review, 47(3), 2006, 991-1011.
Roodman, D. "How to Do Xtabond2: An Introduction to Difference and System GMM in Stata." The Stata Journal, 9(1), 2009a, 86-136.
--. "A Note on the Theme of Too Many Instruments." Oxford Bulletin of Economics and Statistics, 71(1), 2009b, 135-58.
Staiger, D., and J. H. Stock. "Instrumental Variables Regression with Weak Instruments." Econometrica, 65(3), 1997, 557-86.
Stigler, G. "The Theory of Economic Regulation." The Bell Journal of Economics and Management Science, 2(1), 1971,3-21.
Stock, J. H., and M. Yogo. "Testing for Weak Instruments in Linear IV Regression," in Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg, edited by D. W. K. Andrews and J. H. Stock. Cambridge: Cambridge University Press, 2005, 80-108.
Tsebelis, G. "Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism, Multicameralism and Multipartyism." British Journal of Political Science, 25(3), 1995, 289-325.
Tullock, G. "The Welfare Costs of Tariffs, Monopolies, and Theft." Western Economic Journal, 5(3), 1967, 224-32.
(1.) We supplement the data with measures of ethnic fragmentation as explained in Section IV.
(2.) The irregular start date of each period is dictated by the interest group data, which are available as of 1973, 1980, 1985, 1995, 1999, and 2002.
(3.) Freedom House also reports an index of civil liberties. We use political rights to follow Dutt and Mobarak (2007) but note that our results are not sensitive to this choice. The two indexes are correlated at better than 0.94.
(4.) There are 29 observations that have zero recorded interest groups. To avoid losing these observations, we assign them the arbitrarily small value of 0.01 before taking the natural log. Our primary results in Table 2 reported below are robust to dropping these observations.
(5.) In our sample, the bivariate correlation between the number of interest groups in a country and the level of democracy is 0.52.
(6.) The number of observations available for the specifications in columns (2), (4), and (6) is smaller than the number available for the specifications in columns (1), (3), and (5). This difference does not explain the different findings regarding democracy. When we estimate all specifications using the smaller number of observations, the findings in columns (1), (3), and (5) remain unchanged.
(7.) At the suggestion of a reviewer, we also examined whether the relation between groups and policy volatility is conditional on the extent of democracy. In the case of fiscal volatility, the findings indicate that the marginal impact of groups is smaller in magnitude in more democratic nations, though the sign remains unchanged. Thus, groups have a stabilizing impact on policy in all regimes, but the effect is smaller in more democratic nations. The estimates imply that in the most democratic nations the marginal impact of groups on policy volatility is -0.196, and in the least democratic nations the marginal impact of groups is--0.315. The values are not substantially different from our baseline estimate of --0.208, in column (2) of Table 2. In the case of export and import volatility, the findings indicate that the stabilizing impact of groups on policy is statistically significant only in the most democratic nations.
(8.) More specifically, the impact is statistically significant in all but 32 of the 158 countries in the available sample.
(9.) These countries are Austria, Djibouti, and Yemen.
(10.) These countries are Austria, Djibouti, Yemen, and Fiji.
JAC C. HECKELMAN and BONNIE WILSON *
* We thank participants in the meetings of the Western Economic Association and the Public Choice Society for helpful remarks, as well as the editor and two anonymous reviewers.
Heckelman: Professor, Department of Economics, Wake Forest University, Winston-Salem, NC 27109. Phone 336 758 5923, E-mail firstname.lastname@example.org
Wilson: Associate Professor, Department of Economics, Saint Louis University, St. Louis, MO 63108-3397. Phone 314 977 3844, Fax 314 977 1478, E-mail wilsonbe @ slu.edu
TABLE 1 Summary Statistics Fiscal Export Import Volatility Volatility Volatility Democracy Mean 1.314 3.688 4.174 4.429 Standard deviation 1.299 3.048 3.606 2.131 Minimum 0.006 0.008 0.008 1.000 Maximum 15.317 22.634 33.083 7.000 Observations 761 780 780 780 Interest Groups Interest Groups (raw count) (In form) Mean 202.055 3.012 Standard deviation 821.312 2.054 Minimum 0 -4.605 Maximum 14,964 9.613 Observations 780 780 TABLE 2 Groups and Policy Volatility (Baseline Model) Fiscal Volatility Export Volatility (1) (2) (3) (4) Interest groups -0.208 -0.400 (0.000) (0.003) Democracy -0.150 0.033 -0.569 -0.086 (0.014) (0.508) (0.022) (0.778) Political instability 0.069 0.139 -0.779 -0.533 (0.690) (0.282) (0.193) (0.353) A constant, time, transition economy, and oil exporter dummies also included. AR(2) p value 0.278 0.096 0.908 0.861 Hansen p value 0.164 0.479 0.234 0.496 Cragg-Donald 69.66 24.55 71.37 24.89 Kleibergen-Paap rk 22.08 14.05 22.60 Observations 772 761 791 780 Countries 158 158 159 159 Instruments 30 50 30 50 Import Volatility (5) (6) Interest groups -0.679 (0.000) Democracy -0.543 0.122 (0.007) (0.511) Political instability -1.002 -0.801 (0.057) (0.008) A constant, time, transition economy, and oil exporter dummies also included. AR(2) p value 0.935 0.725 Hansen p value 0.134 0.265 Cragg-Donald 71.37 Kleibergen-Paap rk 22.60 14.20 Observations 791 780 Countries 159 159 Instruments 30 50 Notes: p values reported in parentheses are based on heteroskedasticity-corrected and country-clustered standard errors. Errors are Windmeijer-corrected. Orthogonal deviations are used to transform the data to remove fixed effects. Weighted estimation is used, with real GDP per capita serving as the weighting series. Excluded instruments are Independence, OECD, Muslim, British Law, French Law, German Law, Scandinavian Law. Interest Groups is treated as predetermined but not strictly exogenous. Democracy is treated as endogenous. TABLE 3 Groups and Policy Volatility (Extended Model) Fiscal Export Import Volatility Volatility Volatility (1) (2) (3) Interest groups -0.189 -0.256 -0.287 (0.001) (0.149) (0.015) Democracy 0.047 0.043 -0.020 (0.385) (0.769) (0.892) Political instability 0.255 -0.431 -0.443 (0.078) (0.376) (0.202) GDP 0.058 -0.100 0.026 (0.541) (0.726) (0.886) Population -0.009 0.358 -0.008 (0.915) (0.114) (0.962) Growth volatility 0.129 0.339 0.220 (0.013) (0.103) (0.170) Growth 0.016 0.025 0.040 (0.647) (0.740) (0.601) Fractionalization 0.287 -0.446 -0.077 (0.016) (0.229) (0.848) Policy level 0.054 0.072 0.063 (0.000) (0.000) (0.000) A constant, time, transition economy, and oil exporter dummies also included. AR(2) p value 0.053 0.873 0.982 Hansen p value 0.652 0.899 0.345 Observations 753 771 771 Countries 157 159 159 Instruments 101 101 101 Notes: p values reported in parentheses are based on heteroskedasticity-corrected and country-clustered standard errors. Errors are Windmeijer-corrected. Orthogonal deviations are used to transform the data to remove fixed effects. Weighted estimation is used, with real GDP per capita serving as the weighting series. Excluded instruments are Independence, OECD, Muslim, British Law, French Law, German Law, Scandinavian Law. Interest Groups and GDP are treated as predetermined but not strictly exogenous; Democracy, Growth Volatility, and Growth, are treated as endogenous. TABLE 4 Sensitivity Analysis Fiscal Volatility Baseline Full Specification: (1) (2) Panel A: Limited instrument set Interest groups -0.207 -0.204 (0.000) (0.061) Instruments 32 57 Hansen 0.874 0.397 Panel B: Groups > 1500 excluded Interest groups -0.239 -0.194 (0.000) (0.000) Panel C: 1980-1984 excluded Interest groups -0.192 -0.099 (0.000) (0.213) Panel D: 1999-2003 excluded Interest groups -0.217 -0.143 (0.001) (0.029) Panel E: OECD and non-OECD countries Interest groups--OECD -0.211 -0.195 (0.000) (0.001) Interest groups--non-OECD -0.246 -0.137 (0.002) (0.023) Test of equality p value 0.379 0.144 Export Volatility Baseline Full Specification: (3) (4) Panel A: Limited instrument set Interest groups -0.484 -0.091 (0.000) (0.687) Instruments 32 57 Hansen 0.730 0.484 Panel B: Groups > 1500 excluded Interest groups -0.461 -0.230 (0.001) (0.166) Panel C: 1980-1984 excluded Interest groups -0.350 -0.041 (0.008) (0.850) Panel D: 1999-2003 excluded Interest groups -0.362 -0.178 (0.007) (0.324) Panel E: OECD and non-OECD countries Interest groups--OECD -0.373 -0.361 (0.002) (0.020) Interest groups--non-OECD -0.095 -0.032 (0.764) (0.852) Test of equality p value 0.274 0.036 Import Volatility Baseline Full Specification: (5) (6) Panel A: Limited instrument set Interest groups -0.808 -0.481 (0.000) (0.000) Instruments 32 57 Hansen 0.346 0.421 Panel B: Groups > 1500 excluded Interest groups -0.715 -0.299 (0.000) (0.005) Panel C: 1980-1984 excluded Interest groups -0.682 -0.134 (0.000) (0.444) Panel D: 1999-2003 excluded Interest groups -0.673 -0.132 (0.000) (0.450) Panel E: OECD and non-OECD countries Interest groups--OECD -0.576 -0.260 (0.000) (0.026) Interest groups--non-OECD -0.380 -0.237 (0.092) (0.055) Test of equality p value 0.311 0.853 Notes: p values are reported in parentheses. "Baseline" specifications are analogous to those in columns (2), (4), and (6) of Table "Full" specifications are analogous to those in columns (3), (6), and (9) of Table 3. TABLE 5 Groups and Policy Volatility (Conditional Effects) Fiscal Volatility Export Volatility (1) (2) (3) (4) Interest groups -0.082 -0.116 -0.534 -0.590 (0.315) (0.147) (0.050) (0.115) Polarization 1.699 1.454 (0.010) (0.372) Polarization*Groups -0.304 0.237 (0.028) (0.572) Fractionalization 1.661 0.515 (0.089) (0.868) Fractionalization*Groups -0.317 0.409 (0.119) (0.625) Import Volatility (5) (6) Interest groups -1.352 -1.170 (0.000) (0.000) Polarization -3.122 (0.054) Polarization*Groups 1.212 (0.018) Fractionalization -3.236 (0.203) Fractionalization*Groups 1.383 (0.062) Democracy and political instability also included. A constant, time, transition economy, and oil exporter dummies also included. Notes: p values reported in parentheses are based on heteroskedasticity-corrected and country-clustered standard errors. Errors are Windmeijer-corrected. Orthogonal deviations are used to transform the data to remove fixed effects. Weighted estimation is used, with real GDP per capita serving as the weighting series. Excluded instruments are Independence, OECD, Muslim, British Law, French Law, German Law, Scandinavian Law. Interest Groups and Fractionalization*Groups are treated as predetermined but not strictly exogenous; Democracy is treated as endogenous.
|Printer friendly Cite/link Email Feedback|
|Author:||Heckelman, Jac C.; Wilson, Bonnie|
|Publication:||Contemporary Economic Policy|
|Date:||Apr 1, 2016|
|Previous Article:||Effects of the Sunday shopping restriction in Korea.|
|Next Article:||Crises and government: some empirical evidence.|