Political Institutions, Political Careers and Environmental Policy - SSRN

KYKLOS, Vol. 67 – February 2014 – No. 1, 54–73

Political Institutions, Political Careers and Environmental Policy Per G. Fredriksson and Jim R. Wollscheid*

I. INTRODUCTION We argue that party discipline, party strength, and political instability are joint determinants of environmental policy outcomes. We follow Grossman and Helpman (2005) and define the level of party discipline as the degree to which elected legislators keep their party’s campaign promises after being elected. With high party discipline, legislators are induced to toe the party line, as their future political careers may depend on it (Longley, 1998; Grossman and Helpman, 2005; Carey, 2007). High party discipline arises when legislative party leaders have control over nominations for reelection to the same or another office, and thus national party leaders have the ability to hamper or improve a legislator’s career prospects.1 The level of party strength relates to the extent that a party has a wellmaintained and well-financed party machine at both the grassroots and national level (Enikolopov and Zhuravskaya, 2007; Keefer and Khemani, 2009). A strong party can provide political support in the form of local volunteers going doorto-door in support of its candidates, increase turnout by transporting voters to

*

Fredriksson: Professor of Economics, Department of Economics, University of Louisville, Louisville, KY 40292, USA; Email: [email protected]. Wollscheid: Associate Professor of Economics, Department of Economics, University of Arkansas – Fort Smith, Fort Smith, AR 72913, USA; Email: [email protected]. We are grateful to two insightful and helpful referees, the managing editor Alois Stutzer, Theo Eicher, Mohammad Farzanegan, Rajeev Goel, Steve Gohmann, Dann Millimet, Salvatore Nunnari, John Pepper, Hem Paudel, Dan Phaneuf, Hilary Sigman, the participants at the World Congress of Environmental and Resource Economists, Montreal, the MPSA Conference, Chicago, the NCDE Conference, Copenhagen, and at seminars at Clemson and Illinois State University, for constructive comments and suggestions. Dann Millimet kindly provided aspects of the data. Fredriksson gratefully acknowledges research and travel support from the College of Business, University of Louisville. Fredriksson worked on this paper while visiting Resources for the Future, and he is grateful for the hospitality. The usual disclaimers apply. 1. Fredriksson and Wollscheid (2010) find that greater party discipline raises environmental policy strictness when the level of corruption is low, but it reduces the policy strictness when corruption is high (they abstract from party strength and political stability).

54

© 2014 John Wiley & Sons Ltd.

POLITICAL INSTITUTIONS, POLITICAL CAREERS AND ENVIRONMENTAL POLICY

polling stations, and provide campaign finance support.2 National party strength induces local politicians and Congressional level legislators to pay more attention to the wishes of their national party bosses, since stronger parties are expected to furnish more successful careers to their representatives (Riker, 1964; Enikolopov and Zhuravskaya, 2007; Primo and Snyder, 2010).3 The literature has not yet investigated the effect of party strength on environmental policy outcomes. A parallel literature suggests that the level of political stability is an important determinant of policy outcomes, including deforestation rates, resource extraction rates and environmental policies (see, e.g., Deacon, 1994; Bohn and Deacon, 2000; Fredriksson and Svensson, 2003). The above strands of literature have so far developed independently of each other. We begin to remedy this gap in the literature by investigating whether: (i) the empirical effects of party discipline and party strength on the stringency of environmental policy are conditional on the degree of political stability, and (ii) the empirical effect of political stability is conditional on the levels of party discipline and party strength. The analysis aims to help us better understand instances of political failure in the area of environmental policy across countries (see, e.g., Babcock et al., 1997; Wu and Boggess, 1999; Wu et al., 2001). The analysis may help us understand the puzzle why some countries with higher GDP/capita than their neighbors may still have weaker environmental policies.4 Studying environmental policy is also of interest as this area offers society an important trade-off between short term costs and long term benefits.5 What are the policy effects of party discipline and party strength? The literature suggests that both greater party discipline and stronger national political parties bring environmental policies closer to their optimal levels, assuming the level of political stability is high. This is because both these institutions shift policy decisions to party officials at the national level and reduce the influence of 2.

Enikolopov and Zhuravskaya (2007) find that the outcome of fiscal decentralization (measured as economic growth, quality of government, and public good provision) improves when national parties are stronger. 3. The literature discussing party discipline and party strength includes, e.g., Riker (1964), Mayhew (1986), Grossman and Helpman (2005), Carey (2007), Enikolopov and Zhuravskaya (2007), and Primo and Snyder (2010). 4. For example, Chile (GDP = $10,100 in year 2000) scored lower than Uruguay (GDP = $9,300 in year 2000) on the Institutional Capacity index (CIESIN, 2002) (higher values represent stricter policies): 57.5 versus 68.1, respectively. While both countries have relatively high political stability (Kaufmann et al., 2003), Chile has low party discipline and Uruguay high party discipline. On the other hand, Russia (GDP = $7,700 in year 2000) scored 26.8 on the Institutional Capacity index, while Lithuania (GDP = $7,300 in year 2000) scored 50.9. Both countries have high party discipline, but Russia scores lower on Kaufmann et al.’s (2003) political stability index (−0.4) compared to Lithuania (0.9). 5. Moreover, environmental policies addressing national pollution problems may be suitable for this study as they appear relatively unaffected by multilateral agreements. This is in contrast to, e.g., trade agreements which tie governments’ hands to a greater degree.


55

PER G. FREDRIKSSON/JIM R. WOLLSCHEID

the local constituency. National party leaders put a higher value on the national party’s overall electoral future (and thus on overall national welfare) than do local districts’ representatives (Grossman and Helpman, 2005; Enikolopov and Zhuravskaya, 2007). They will therefore tend to promote policies that lead to the internalizing of spillover damage across districts, and avoidance of sub-optimal capital competition using weak environmental regulations. In contrast, with low party discipline and low national party strength local party officials and district legislative representatives ignore the national party’s electoral future and overall national welfare to a greater extent. Instead, their local constituents’ narrow interests carry more weight. For example, local constituencies will not internalize spillover damage on other districts, and may engage in sub-optimal capital competition using weak environmental regulations. Thus, a national perspective among legislators may be expected to raise environmental policy stringency because it is more likely to address national transboundary pollution spillovers (see, e.g., Oates and Portney, 2003). What are the policy effects of increased political stability? As political stability rises, party leaders have a greater probability of being in office long enough to benefit from delivering welfare. Thus, political stability lengthens the decision-making horizon and will tend to bring policies closer to the optimal level.6 We focus on the behavior of national party leaders (and not lobby groups or other bribe givers). We assume that higher political stability brings policy closer to the optimal level of environmental policy stringency (which we assume implies more stringent national environmental policies). Drawing on our discussion of the different strands of literature, the effects of party discipline and party strength on policy outcomes is expected to be conditional on the level of political stability. In particular, greater party discipline and party strength, and thus a more national perspective among political representatives should on the one hand, raise environmental policy strictness where the level of political stability is high (party leaders have a long time horizon). This is because of a decline in free-riding behavior among local district representatives; office holders are induced by national party leaders to take a longer term and more national perspective. This benefits legislators’ careers when party discipline and/or party strength are high.

6.

56

For example, Deacon (1994) explains that forest preservation is a form of investment. With less volatile or predatory political institutions, the level of uncertainty regarding property rights is lower and the incentive to invest increases. In Olson’s (1991) “roving bandit” theory, a more stable incumbent will find it optimal to steal less today, and to rather formulate policies that build future wealth. Bohn and Deacon (2000) find that while deforestation declines with lower ownership risk, extraction of petroleum increases. Fredriksson and Svensson (2003) argue that with higher political stability, the government will care more about delivering social welfare and thus tend to set more optimal policies addressing national (or local) environmental problems.



On the other hand, when the degree of national level political stability is low, high party discipline and party strength may be expected to cause weaker national environmental policies. This occurs if policy choices of national-level political leaders are severely affected by low political stability. Instability induces them to adopt a short time horizon, as the probability of their party losing power is high. The long term benefits of environmental policies become less important to national leaders, and therefore the short term costs become more important. As a result, national leaders may also be more inclined to seize the opportunity to derive personal benefits. In turn, this leads them to dictate weak environmental policy choices to legislators.7 We also argue that the effect of political stability is conditional on the levels of party discipline and party strength. When the levels of party discipline and party strength are high (and free riding is lower), a reduction in political stability should have a relatively large marginal impact on environmental policy stringency. This is so because the instability affects party leaders who can dictate policy. In this case, party leaders switch their policy preferences towards weaker environmental policies as their time horizon is shortened. This in turn translates into weaker policy as lower-level politicians listen attentively to their leaders due to career concerns. Conversely, when the levels of party discipline and/or party strength are low, national party leaders are unable to dictate policy. In this case, free riding behavior is already high and the party leaders are consequently not in a position to induce weaker policy stringency. Although shorter time horizons may affect policy stringency negatively, the marginal impact is likely to be small or negligible in the presence of low party discipline and party strength. We present ordinary least squares (OLS) results utilizing data from up to 94 democracies. Measures of party discipline and party strength come from Beck et al. (2001). As our main variable, we create a combined variable indicating the presence of both high party discipline and high party strength. With both high discipline and high party strength, political careers are particularly tightly guided by the party leadership. Both the “stick” and the “carrot” are available to the national party bosses in this case, creating the strongest incentive for legislators 7.

For this to occur, political instability should have a relatively large impact at the national level. The literature provides examples of this situation. Bardhan and Mookherjee (2000) discuss a model which includes an absence of effective electoral competition due to loyalty biases favoring one party at the local level. Suppose two parties participate in elections at both the national and local levels, and local districts have sharply different party loyalties (due to socioeconomic or ethnic composition, e.g.). In each district, the dominant party is the overwhelming favorite. In this case, the electoral outcome is more uncertain at the national level. Moreover, Bardhan and Mookherjee (2000) also argue that the number of competing parties may be greater at the national than at the local level, perhaps due to differences in party strength across districts, or because the stakes are greater at the national level. If more political parties produce a greater number of possible government coalitions, political uncertainty may be greater at the national level.


57


and party officials to follow their national party leaders. Our political stability index comes from Kaufmann et al. (2003).8 We evaluate the effects of the above measures on three different measures of environmental policy stringency (CIESIN, 2002; Metschies, 2003). Our empirical results suggest that the presence of high party discipline and strong political parties raise the stringency of environmental policies when the level of political stability is high, but the reverse effect tends to occur when the degree of political stability is low. The effect of political stability on environmental policy stringency is positive when party discipline and party strength are high, but for lower such values the effect declines and eventually disappears. These findings constitute novel contributions to the literature, to our knowledge. The identified interactions may have more general implications and may apply to other economic policies. The paper is structured as follows. Section II describes the empirical model and data. Section III reports the empirical results, Section IV offers a robustness analysis, and Section V concludes. Appendix I contains summary statistics. Appendix II provides variable definitions and sources, and Appendix III provides further robustness analysis.

II. EMPRICAL MODEL AND DATA In this section we discuss the approach used to test whether countries with high party discipline and high party strength set tougher environmental policies when the level of political stability is high and weaker environmental policies when the level of political stability is low. Moreover, we test whether the effect of political stability is stronger when party discipline and party strength are high. We present OLS results. The equation estimated is

ti = α + τ 1DiscStrengthi + τ 2 ( DiscStrengthi *Stabilityi ) + λ Stabilityi + β xi + μi, (1) where ti is a measure of environmental policy stringency in country i, DiscStrengthi is an interaction variable combining our measures of party discipline and party strength, Stability is our measure of political stability, xi is a vector of control variables, and μi is a well-behaved error term. While we focus on 94 democracies from the late 1990’s and the early 2000’s, data is available for a total of 128 countries. We classify as democracies those 8.

58

One benefit of this measure appears to be that it captures the perceived level of political stability at the national level.



countries that have an average Polity2 score greater than zero during years 1990–2000 (Marshall et al., 2006).9 Three different dependent variables measure environmental policy stringency. Two composite indices come from CIESIN (2002), produced in collaboration with the Yale Center for Environmental Law and Policy, the Global Leaders of Tomorrow World Economic Forum, and Columbia University’s Center for International Earth Science Information Network (CIESIN): (i) Institutional Capacity and (ii) Environmental Governance. Institutional Capacity measures the extent to which a country has in place institutions and underlying social patterns of skills, attitudes, and networks for effective responses to environmental problems. It includes measures of a country’s scientific and technological capacity to address environmental challenges, access to environmental data in publicly available datasets, measures of the responsiveness of the private sector to environmental challenges, and finally the (less noisy) sub-index Environmental Governance. Environmental Governance measures the institutions, rules and practices that shape environmental policies, using the following variables: the quality of environmental regulations, the existence of sectoral guidelines for environmental impact assessments, the degree of transparency and absence of corruption in environmental decision-making, the extent of protected areas and the degree of certification of forest areas for sustainable management, and existence of subsidies that may lead to over-exploitation of resources. The third dependent variable is the year 2000 price of super gasoline (Super2000) from Metschies (2003).10 Discipline classifies our 94 democratic countries into high or low party discipline systems. Discipline equals 1 if a “closed list” system is used, i.e., voters cannot express preferences for candidates within a party list and thus the position on the list (decided by party leaders) determines a candidate’s election probability; 0 otherwise (Beck et al., 2001).11 As a measure of party strength we use average party age, following Enikolopov and Zhuravskaya (2007). Strength is 9.

Persson and Tabellini (2009) use a similar rule. Using an average Polity2 score avoids that countries that are in recent transition to democracy are included. The inclusion of such observations would potentially be problematic since environmental policy making and implementation involve some lags. 10. Data on gas taxes would be preferred, but such data is available only for OECD countries. We therefore use gas prices. Differences in gasoline prices across countries are affected by domestic demand and openness to international trade, environmental taxes, congestion taxes aimed at externalities, and possible other taxes. Taxes represent the major share of the variation in gasoline prices among OECD countries (see OECD/IEA (2000), Figs. 16 and 18). 11. While in this paper we discuss open- versus closed party list systems, party discipline may be created by other institutions. Nacif (2002) argues that Mexican parliamentary leaders can also use their procedural powers to monitor individual legislators throughout the law-making process, including their committee work. Cheibub Figueiredo and Limongi (2000) argue that despite an open list system, Brazil has high party discipline (in the sense that the party coalition supporting the president votes in favor of most of the president’s proposals) due to the legislative powers of the president and the centralized organization of legislative work.


59


defined as the average age of the two main government parties and the main opposition party, and it comes from Beck et al. (2001).12 By interacting Discipline and Strength, we create DiscStrength, which measures the simultaneous presence of both high party discipline and high party strength. To be clear, the assumption implies that party strength matters only in the presence of high party discipline. Thus, high party discipline and high party strength are each necessary conditions for party leaders having a strong grip over legislators, but each condition is not by itself sufficient for this to occur. Stability combines several indicators measuring perceptions of the probability that a government in power will be destabilized or overthrown by possibly unconstitutional and/or violent means, including terrorism and domestic violence (Kaufmann et al., 2003). It ranges from −3 to 3; higher values indicate higher political stability. We now turn to our control variables. GDP/CAP captures the increase in demand for environmental quality as income per capita rises; GDP/CAP2 captures potential Environmental Kuznets curve effects; and Trade Openness (the ratio of exports plus imports to GDP) proxies the potential effect of international trade on environmental policies (Damania et al., 2003). Demographic variables include measures of the age distribution (Age 15–64) (proxy for the number of drivers), population (Population), and population density (PopDensity). The latter two variables are proxies for the exposure to pollution damage. We also include the percentage of population adhering to Islam in 2000 (Muslim). The included institutional variables are the number of years of independence (Independence), a UK colony dummy interacted with Independence (UK Colony), a French colony dummy interacted with Independence (French Colony), Africa (Africa), East Asia (East Asia), Latin America (Latin America) dummies, dummies for parliamentary governance system (Parliament), federal system (Federal), and proportional electoral system (Proportional). See Table A1 in Appendix I for descriptive statistics. Appendix II provides data sources.

III. RESULTS Table 1 reports the main estimation results of Equation (1) using DiscStrength and Political Stability. Table 1 provides two models each for our three dependent variables. DiscStrength is never significant by itself. However, the interaction 12.

According to Huntington (1968), a higher age of the main political parties reflects a more stable party system and stronger parties. Beck et al.’s measure closely follows the party strength ranking by Garman et al. (2001) for four Latin American countries (Argentina, Brazil, Mexico, and Venezuela), see Enikolopov and Zhuravskaya (2007). The exception is Colombia which appears due to its unusual party system. Note also that Strength does not equal OECD membership; the simple correlation coefficient between these variables is 0.3415.

60


POLITICAL INSTITUTIONS, POLITICAL CAREERS AND ENVIRONMENTAL POLICY Table 1 Ordinary Least Squares Estimation Results Model

DiscStrength DiscStrength × Stability Stability GDP/CAP GDP/CAP2 PopDensity

Environmental Institutional Super2000 Environmental Institutional Governance Capacity Governance Capacity (1) (2) (3) (4) (5) −9.9e-06 (.002) .003* (.002) .10 (.08) .00004* (.00002) 6.7e-11 (5.6e-10) −.00006*** (.00001)

.06 (.06) .057 (.051) 4.85*** (1.62) .0008* (.0004) 6.0e-09 (1.5e-08) −.002*** .0003

.08 (.13) .22* (.13) −3.67 (3.29) .002** (.001) −4.2e-08 (4.0e-08) −.002* (.0009)

41.09*** (2.45) 83 .6918

48.59*** (4.89) 85 .3707

Population Age15–64 Age65+ Trade Openness Independence Proportional Parliament Federal Africa East Asia Latin America Constant Observations R2

−.31*** (.12) 83 .5778

−.0002 (.003) .005** (.002) .03 (.09) .00009** (.00004) −1.7e-09* (8.9e-10) −.00009*** (.00003) .0003 (.0003) −.02 (.02) −.0007 (.02) −.0004 (.003) −.24 (.49) −.32** (.12) −.17 (.14) .03 (.16) .19 (.24) −.13 (.17) −.25 (.18) 1.19 (.99) 82 .6801

Super2000 (6)

.05 .05 (.07) (.13) .09 .24** (.07) (.11) 3.44** −7.15** (1.87) (5.07) .001* .003*** (.0008) (.001) −9.9e-09 −6.2e-08 (2.0e-08) (4.6e-08) −.002*** .0005 (.0008) (.002) .004 .03*** (.006) (.01) −.05 −1.13 (.39) (.78) −.70 2.45** (.50) (1.13) .05 .08 (.07) (.10) −10.05 35.74 (15.43) (27.44) −2.57 1.23 (2.67) (5.22) 1.43 −6.60 (2.95) (5.38) −3.78 −15.13** (3.91) (6.72) 3.77 9.12 (4.80) (11.35) −3.71 8.43 (3.53) (7.63) −5.42 9.67 (5.21) (10.48) 61.69*** 66.25 (25.39) (53.30) 82 83 .7496 .5514

Notes: Models 4–6 include Muslim, UK Colony, and French Colony (not reported, available upon request). Standard errors in parenthesis. ***(**)[*] indicates significant at the 1(5)[10]% level, respectively. The control variables in Models 1–3 [4–6] are denoted Control Set # 1 [2].

term of interest is positive and significant in four of the six models. The marginal effect of DiscStrength is found by adding the DiscStrength coefficient and the DiscStrength × Stability interaction coefficient, taking the level of Stability into account. The significant coefficients suggest that DiscStrength has a positive effect on environmental policy when Political Stability is high, but the reverse effect occurs when Political Stability is low (recall that Political Stability ranges © 2014 John Wiley & Sons Ltd.

61

PER G. FREDRIKSSON/JIM R. WOLLSCHEID Figure 1

The Marginal Effect of DiscStrength on Environmental Governance, conditional on Stability

Marginal Effect of DiscStrength

Enviromental Governance

0.04 0.02 0 -3

-2

-1

0

1

2

3

-0.02 -0.04

Stability

Notes: Calculations based on Model 4, Table 1. Dashed lines represent the 90 percent confidence interval.

from 3 to −3). Figure 1 uses Model 4 to illustrate the marginal effect of DiscStrength on Environmental Capacity, conditional on the level of Stability.13 The figure suggests that the marginal effect is positive for high Stability levels; while at low levels of Stability, the marginal effect is negative. The effect of Political Stability is not consistent across the models. The direct effect of Political Stability is significant and positive in two models, but negative in a third (Model 6).14 The marginal effect of Stability is found by adding the Stability coefficient and the DiscStrength×Stability interaction coefficient, taking the level of DiscStrength into account. Using Model 4, Figure 2 illustrates the marginal effect of Stability, conditional on DiscStrength. Stability has a larger positive marginal effect on Environmental Capacity, the greater is DiscStrength. Among the control variables, GDP/CAP and PopDensity are significant in most models, while the remaining controls presented in Table 1 do not show consistently significant coefficients. IV. ROBUSTNESS ANALYSIS While these initial results are informative, further robustness analysis is needed. We present robustness checks using OLS in Table 2, where Control Set #1 (see Table 1) is used throughout (see also Appendix III). First, in Panel A, Models Model 4 in Table 1 yields the expression Environmental Governance = −0.0002 + 0.005 × Stability. In Figure 1, Stability is then allowed to take values along its range. The analogous approach is used to create Figure 2. 14. Political Stability still has a positive effect on Super2000 in Model 6, however, as long as DiscStrength is above 29.79. While this is 0.55 std. dev. above the mean of DiscStrength, note that Discipline (part of DiscStrength) is a dummy variable. When Discipline = 1, Stability has a positive effect even if DiscStrength (and Strength) is slightly below the mean (equal to 32.64).

13.

62


POLITICAL INSTITUTIONS, POLITICAL CAREERS AND ENVIRONMENTAL POLICY Figure 2

The Marginal Effect of Stability on Environmental Governance, conditional on DiscStrength

Marginal Effect of Stability 1.2

Environmental Governance

1 0.8 0.6 0.4 0.2 0 -0.2

0

20

40

60

80

100

120

DiscStrength Notes: Calculations based on Model 4, Table 1. Dashed lines represent the 90 percent confidence interval.

1–3, we restrict the sample to countries that have been independent for at least 75 years (CIA, 2003). This allows the main political parties the time to reach a sufficient age to be viewed as strong by political participants. The sample size falls, but two interaction coefficients and two Stability coefficients are significant. Second, the effect of Strength may not be monotone but may exhibit a declining marginal effect beyond some threshold level of average party age. Political parties that survive some given threshold number of years may be viewed as strong and thus be important for politicians’ careers. We therefore create a dummy variable, Strength25, which takes a value of unity when average party age equals 25 years or more, and zero otherwise. We then create a corresponding DiscStrength25 interaction variable. The results reported in Panel A, Models 4–6, suggest that Strength25 may indeed capture aspects of political behavior as two DiscStrength25 × Political Stability interactions are significant. DiscStrength25 and Political Stability are individually significant in Model 5. In Panel B, we study the effect of DiscStrength in proportional electoral systems and in parliamentary systems. The effects of party discipline and party strength may play out differently in different systems because of differences in incentive structures. We find only one significant interaction coefficient in Panel B, for parliamentary systems (Model 6). Next, in Panel C we study the individual effects of Discipline and Strength, rather than as parts of an interaction. The Discipline×Stability interaction is positive and significant in Model 1, while the Strength×Stability interaction is positive and significant in Models 4–6. Panel C suggests that Strength may be driving the results in Panel A, as well as the results in Table 1. © 2014 John Wiley & Sons Ltd.

63

64 Institutional Capacity (2)

−0.002 (.002) .004* (.002) .20** (.10) 50

Old Democracies .029 (.065) .071 (.060) 6.46*** (2.46) 50 Proportional Systems .0025 .13 (.0027) (.09) .0011 0.005 (.0024) (.08) .19** 6.21** (.09) (2.46) 61 61 Democracies −0.09 −.57 (0.09) (2.15) 0.18* 2.62 (0.10) (2.22) 0.07 4.68** (0.08) (1.60) 83 83

Environmental Governance (1)

5.54 (4.46) 7.41 (5.00) −3.37 (3.16) 85

0.05 (.20) .18 (.17) −2.75 (4.39) 64

.011 (.15) .25* (.15) −1.96 (4.94) 49

(3)

Super2000

Stability

Strength × Stability

Panel C Strength

Panel B

Stability

DiscStrength25 × Stability

Panel A DiscStrength25

Outcome Variable Institutional Capacity (5)

Democracies 5.29* (3.17) 3.37 (2.74) 5.14*** (1.54) 83 Parliamentary Systems 0.008 .12 (.007) (.11) −.004 −0.03 (.005) (.09) .28 11.75*** (.22) (3.82) 41 41 Democracies .003* .11*** (.001) (.04) .002** .06** (.001) (.02) .08 3.97** (.09) (2.00) 83 83 .01 (.13) .21** (.11) 0.09 (0.07) 83


−.009 (.11) .15* (.08) −5.34 (4.56) 85

−0.08 (.14) .24** (.13) −3.60 (6.64) 44

1.66 (6.74) 13.99* (7.47) −3.80 (3.36) 85

(6)

Super2000

Notes: All models use Control Set #1, i.e. the control variables in Models 1–3 in Table 1. Standard errors in parenthesis. ***,**,* represent significance at the 1%, 5%, 10% level respectively.

Observations

Stability

Discipline × Stability

Discipline

Observations

Stability

DiscStrength × Stability

DiscStrength

Observations

Stability

DiscStrength × Stability

DiscStrength

Treatment Variable

Table 2 Robustness Analysis I: Ordinary Least Squares




In Table 2, Stability is significant in eight models. Only Model 5 in Panel C (using Strength) displays both significant Stability and interaction coefficients. This model suggests that an increase in political stability has a positive effect on Institutional Capacity in all countries, but the effect is stronger in countries with high party strength. V. CONCLUSION In this paper, we provide cross-country evidence that the presence of high party discipline and strong political parties have a positive [negative] effect on environmental policy in stable [unstable] democracies. Moreover, political stability has stronger positive effect in countries with high party discipline and strong political parties. Party strength appears to play a relatively more influential role than party discipline. These findings may apply more generally to other policy areas. REFERENCES Babcock, Bruce A., P.G. Lakshminarayan, JunJie Wu and David Zilberman (1997). Targeting Tools for the Purchase of Environmental Amenities, Land Economics. 73: 325–339. Bardhan, Pranab and Dilip Mookherjee (2000). Capture and Governance at the Local and National Levels, American Economic Review. 90: 135–139. Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer and Patrick Walsh (2001). New Tools in Comparative Political Economy: The Database of Political Institutions, World Bank Economic Review. 15: 165–176. Bohn, Henning and Robert T. Deacon (2000). Ownership Risk, Investment, and the Use of Natural Resources, American Economic Review. 90: 526–549. Carey, John M. (2007). Competing Principals, Political Institutions, and Party Unity in Legislative Voting, American Journal of Political Science. 51: 92–107. Center for International Earth Science Information Network (CIESIN) (2002). 2002 Environmental Sustainability Index, Yale Center for Environmental Law and Policy, Yale University, www.ciesin.columbia.edu. Cheibub Figueiredo, Angelina and Fernando Limongi (2000). Presidential Power, Legislative Organization, and Party Behavior in Brazil, Comparative Politics. 32: 151–170. CIA (2003). The World Factbook, CIA, Washington, DC. Deacon, Robert T. (1994). Deforestation and the Rule of Law in a Cross-Section of Countries, Land Economics. 70: 414–430. Damania, Richard, Per G. Fredriksson and John A. List (2003). Trade Liberalization, Corruption, and Environmental Policy Formation: Theory and Evidence, Journal of Environmental Economics and Management. 46: 490–512. Egger, Peter H. and Maximilian von Ehrlich (2013). Generalized Propensity Score for Multiple Continuous Treatment Variables, Economics Letters. 119: 32–34. Enikolopov, Ruben and Ekaterina Zhuravskaya (2007). Decentralization and Political Institutions, Journal of Public Economics. 91: 2261–2290. Forum of Federations (2005). www.forumfed.org. Fredriksson, Per G. and Daniel L. Millimet (2007). Legislative Organization and Pollution Taxation, Public Choice. 131: 217–242. © 2014 John Wiley & Sons Ltd.

65

PER G. FREDRIKSSON/JIM R. WOLLSCHEID Fredriksson, Per G. and Jakob Svensson (2003). Political Instability, Corruption and Policy Formation: the Case of Environmental Policy, Journal of Public Economics. 87: 1383–1405. Fredriksson, Per G. and Jim R. Wollscheid (2010). Party Discipline and Environmental Policy: The Role of “Smoke-filled Back Rooms,” Scandinavian Journal of Economics. 112: 489–513. Garman, Christopher, Stephan Haggard and Eliza Willis (2001). Fiscal Decentralization: A Political Theory with Latin American Cases, World Politics. 53: 205–236. Gehlbach, Scott and Philip Keefer (2012). Private Investment and the Institutionalization of Collective Action in Autocracies: Ruling Parties and Legislatures, Journal of Politics. 74: 621– 635. Grossman, Gene M. and Elhanan Helpman (2005). A Protectionist Bias in Majoritarian Politics, Quarterly Journal of Economics. 120: 1239–1282. Huntington, Samuel P. (1968). Political Order in Changing Societies. New Haven: Yale University Press. Kaufmann, Daniel, Aart Kraay and Massimo Mastruzzi (2003). Governance Matters III: Governance Indicators for 1996–2002, World Bank Policy Research Working Paper 3106. Keefer, Philip (2011). Collective Action, Political Parties and Pro-Development Public Policy, Asian Development Review. 28: 94–118. Keefer, Philip and Stuti Khemani (2009). When Do Legislators Pass on Pork? The Role of Political Parties in Determining Legislator Effort, American Political Science Review. 103: 99–112. Longley, Neil (1998). Legislative Systems with Absolute Party Discipline: Implications for the Agency Theory Approach to the Constituent-legislator Link, Public Choice. 97: 121–140. Marshall, Monty G., Keith Jaggers and Ted R. Gurr (2006). Polity IV: Political Regime Characteristics and Transitions, 1800–2004. http://www.cidcm.umd.edu/polity/ Mayhew, David R. (1986). Placing Parties in American Politics. Princeton: Princeton University Press. Metschies, Gerhard P. (2003). International Fuel Prices, 3rd Ed. Eschborn (Germany): Deutsche Gesellschaft für Technische Zusammenarbeit GmbH (GTZ). www.zietlow.com/docs/Fuel-Prices -2003.pdf. Millimet, Daniel L. (2013). Empirical Methods for Political Economy Analyses of Environmental Policy, in: Jason Shogren (ed.), Encyclopedia of Energy, Natural Resource, and Environmental Economics. Oxford: Elsevier. Millimet, Daniel L. and Rusty Tchernis (2009). On the Specification of Propensity Scores, with Applications to the Analysis of Trade Policies, Journal of Business & Economic Statistics. 27: 397–415. Nacif, Benito (2002). Understanding Party Discipline in the Mexican Chamber of Deputies: The Centralized Party Model, in: Scott Morgenstern and Benito Nacif (eds.), Legislative Politics in Latin America. Cambridge: Cambridge University Press. Oates, Wallace E. and Paul R. Portney (2003). The Political Economy of Environmental Policy, in Karl-Göran Mäler and Jeffrey R. Vincent (eds.), Handbook of Environmental Economics 1. Amsterdam: Elsevier. OECD/IEA (2000). Energy Policies of IEA Countries, 2000 Review. Paris: Organization for Economic Cooperation and Development and International Energy Agency. Olson, Mancur (1991). Autocracy, Democracy, and Prosperity, in: R. Zeckhauser (ed.), Strategy and Choice. Cambridge: MIT Press. Pearl, Judea (2009). Causality: Models, Reasoning and Inference, 2nd ed. Cambridge: Cambridge University Press. Persson, Torsten and Guido Tabellini (2002). Do Constitutions Cause Large Governments? Quasiexperimental Evidence, European Economic Review. 46: 908–918. Persson, Torsten and Guido Tabellini (2003). The Economic Effects of Constitutions. Cambridge: MIT Press.

66


POLITICAL INSTITUTIONS, POLITICAL CAREERS AND ENVIRONMENTAL POLICY Persson, Torsten and Guido Tabellini (2009). Democratic Capital: The Nexus of Political and Economic Change, American Economic Journal: Macroeconomics. 1: 88–126. Primo, David M. and James M. Snyder, Jr. (2010). Party Strength, the Personal Vote, and Government Spending, American Journal of Political Science. 54: 354–370. Riker, William (1964). Federalism: Origins, Operations, Significance. Boston: Little, Brown and Co. Rosenbaum, Paul R. and Donald B. Rubin (1983). The Central Role of the Propensity Score in Observational Studies for Causal Effects, Biometrika. 70: 41–55. Rubin, Donald B. (2007). The Design versus Analysis of Observational Studies for Casual Effects: Parallels with the Design of Randomized Trials, Statistics in Medicine. 26: 26–36. World Bank (2003). World Development Indicators 2003. Washington, DC: The World Bank. Wu, JunJie and William G. Boggess (1999). The Optimal Allocation of Conservation Funds, Journal of Environmental Economics and Management. 38: 302–321. Wu, JunJie, David Zilberman and Bruce A. Babcock (2001). Environmental and Distributional Impacts of Conservation Targeting Strategies, Journal of Environmental Economics and Management. 41: 333–350.

APPENDIX I Table A1 Summary Statistics

Discipline Strength Strength25 DiscStrength DiscStrength25 Stability Institutional Capacity Environmental Governance Super 2000 Proportional Parliament Federal GDP/capita Trade Openness Population (millions) Age 15–64 Age 65+ PopDensity Independence Africa East Asia Latin America Muslim UK Colony French Colony Democracy


Obs

Mean

S.D.

Minimum

Maximum

94 94 94 94 94 93 83 83 85 94 94 94 94 94 94 92 92 94 94 94 94 94 94 94 94 94

.52 32.64 .46 16.27 .26 .22 53.14 .14 66.88 .72 .5 .19 9424 35.26 37.09 62.71 8.51 70.55 .45 .16 .15 .21 12.68 .15 .34 7.05

.50 30.60 .50 24.79 .44 .87 16.32 .62 22.60 .45 .50 .40 8548 25.78 112.59 5.85 5.17 545.42 .36 .37 .36 .41 24.15 .31 .41 2.83

0 3.07 0 0 0 −1.9 20.9 −1 12 0 0 0 720 4.35 .41 48.56 2.18 .22 .008 0 0 0 0 0 0 .18

1 144 1 130.27 1 1.63 91.5 1.47 119 1 1 1 36200 133.21 1029.99 72.05 18.34 5297.93 1 1 1 1 100 .92 .98 10

67


APPENDIX II

Data Description Discipline. A dummy variable equal to 1 if he voters cannot express preference for candidates within a party list. Source: Beck et al. (2001). Strength. The average age of the first government party, second government part, and first opposition party or the subset of these. Source: Beck et al. (2001). Strength25. A dummy variable equal to 1 if the average age of the first government party, second government part, and first opposition party or the subset of these for which the age of the party is known is greater than 25 years. Source: Beck et al. (2001). DiscStrength. Discipline and Strength interacted. DiscStrength25. Discipline and Strength25 interacted. Stability. A point estimate that measures the likelihood that the government in power will be destabilized or overthrown. Source: Kaufmann et al. (2003). Institutional Capacity. The extent to which a country has in place institutions and underlying social patterns of skills, attitudes and networks that foster effective responses to environmental situations. Source: CIESIN (2002). Environmental Governance. A measure that examines the institutions, rules and practices that shape environmental policy. Source: CIESIN (2002). Super2000. The price of super gasoline in 2000 in US cents per liter. Source: Metschies (2003). Proportional. A dummy variable equal to 1 if the winning party needs to gain a majority of the districts to gain power and Democratic equals 1. Source: Persson and Tabellini (2002). Parliament. A dummy variable equal to 1 if the country has a parliamentary form of government. Source: Persson and Tabellini (2002). Federal. A dummy variable equal to 1 if the country has a federal political structure. Forum of Federations (2005). PopDensity. Population divided by land area, 2000. Source: World Bank (2003). Population. Measures the total population for the country, 1999. Source: World Bank (2003). Age15–64. Percentage of the total population between 15 and 64 years old, 1999. Source: World Bank (2003). Age65+. Percentage of the total population over the age of 65, 1999. Source: World Bank (2003). GDP/CAP. Per capita gross domestic product in US dollars. Source: CIA (2003). Trade Openness. Trade in good as a percent of GDP. Total Export and Total Imports divided by GDP, 2000. Source: CIA (2003). Muslim. Percent of population following the religion of Islam, 2000. Source: www.factbook.net/muslim_pop.php.

68



Independence. (250 - number of years independent from 1748)/250. Source: CIA (2003). UK Colony. Interaction between a dummy for a country being a UK colony (excluding the US) and Independence. Source: Persson and Tabellini (2002) and CIA (2003). French Colony. Interaction between a dummy for a country being a French colony and Independence. Sources: Persson and Tabellini (2002) and CIA (2003). Africa. A dummy equal to 1 if the country is located on the continent of Africa. East Asia. A dummy equal to 1 if the country is located in East Asia. Latin America. A dummy equal to 1 if the country is located in Latin America or South America. Democracy. A dummy equal to 1 if a country’s average Polity2 score from 1990–2000 is greater than 0. Source: Marshall et al. (2006). APPENDIX III

IV. Robustness Analysis (continued) While the general results lend some support to our hypotheses, a well-known issue arises when using OLS. Countries are not randomly assigned their party discipline or party strength, but they have instead self-selected into their particular category. These institutions result from, e.g., multiple historical choices made by political leaders, geography, and from existing culture. For example, cultural characteristics may cause countries to have stable and durable political parties and/or to select institutions which induce high party discipline. If these countries also have strong [weak] preferences for environmental policies, a failure to control for this correlation will yield an estimated effect on environmental policy that is biased upwards [downwards]. The propensity score estimation method (PSM) is an alternative method to measure the treatment effect in this situation (Rosenbaum and Rubin, 1983; Rubin, 2007; Pearl, 2009; Egger and von Ehrlich, 2013). By using PSM, we attempt to overcome the issue of self-selection.15 To test for the average treatment effect (ATE), we estimate the propensity score in the first stage (the predicted probability that each observation belongs to the treatment group) utilizing a probit model, and a second stage OLS regression. Note that the creation of the Strength25 dummy discussed in the previous robustness analysis has the advantage that it enables us to analyze the effect of party strength (and its interaction 15.

PSM has been used by, e.g., Persson and Tabellini (2002, 2003); we use the same approach as Fredriksson and Millimet (2007) and Fredriksson and Wollscheid (2010). See Millimet (2013) for a detailed discussion.


69


with party discipline) using PSM (treatment variable required). The first stage propensity score estimation includes all variables in Control Set #2 used in Table 1, Models 4–6. Following Millimet and Tchernis (2009), we expand the propensity score estimation to include higher order terms in the propensity score specification. The results from our PSM analysis are reported in Table A2, with samples similar to those presented in Tables 1–2. Strength25 is significant in two models (Panels B and C), while thirteen Stability coefficients are significant, all with a positive sign. Half of the interaction coefficients are significant with the expected positive sign. Among those are three Strength25 interactions in Panel C, but no Discipline interactions. While these findings (and those in Table 2) suggest that primarily Strength25 contributes to the estimated effects of DiscStrength25, we can compare the two significant (and corresponding) Strength25 and DiscStrength25 interactions using the democracies sample in Panels A and C. In both cases, the DiscStrength25 interaction coefficient is greater than the Strength25 interaction coefficient, suggesting that Discipline also plays a role. To further study the interaction effects of Strength25 and Discipline, in Table A3 we undertake additional robustness analysis in alternative samples using PSM (except Models 4–6 in Panel C, where OLS is used). In Panel A we use our sample of old democracies (≥75 years of independence). Panel B utilizes alternative samples dependent on governance systems. In Panel C we include all available observations in the sample, including autocracies.16 While the Discipline interaction is significant in one model in Panel A, the Strength25 interaction is significant in all three models. Strength25 × Stability is significant in ten out of the twelve models in Panels B and C.17 The direct effect of Strength25 is positive and significant in five models. Moreover, Stability is significant in twelve models in Table A3. Overall, our results appear reasonably robust.18

16.

Keefer (2011) and Gehlbach and Keefer (2012) also use a measure of party strength (party age) in the analysis of autocracies. In additional analysis, we studied the effect of Strength in a sample of autocracies only. This produced no significant coefficients. 17. We also run three OLS models using Strength in the full sample. Strength × Stability is significant in two models. 18. As an alternative approach using our full sample of democracies, we ran OLS regressions including the triple interaction Discipline × Strength × Stability (while also including separate interactions for Discipline × Stability and Strength × Stability). While the coefficient signs are consistent with our earlier findings, only the Strength × Stability coefficient reaches significance in the Environmental Governance model, and Strength in the Institutional Capacity model.

70



Institutional Capacity (2)

−.10 (0.17) .28** (0.13) .27*** (0.08) 82

Democracies 3.31 (4.04) 4.97 (3.14) 9.57*** (1.77) 82 Proportional Systems .09 4.99 (.22) (5.51) .19 3.70 (.14) (3.48) .35*** 10.53*** (0.09) (2.28) 60 60 Democracies −.18 −2.06 (.14) (3.33) .15 1.97 (.13) (3.14) .33*** 11.44*** (.09) (2.18) 82 82


2.66 (5.96) 6.00 (5.60) 5.47* (4.07) 83

−9.37 (9.25) 14.48** (6.18) 3.43 (4.13) 63

.63 (6.66) 14.69*** (5.67) 2.65 (3.34) 83

(3)

Super2000

Stability

Strength25 × Stability

Panel C Strength25

Panel B

Panel A

Outcome Variable Institutional Capacity (5)

Old Democracies −.19 −.55 (.23) (5.58) .34** 6.12 (.15) (3.74) .38*** 11.44*** (.09) (2.30) 49 49 Parliamentary Systems .60* 2.38 (.27) (7.97) −.09 8.57 (0.26) (6.86) 0.52*** 14.16*** (0.11) (2.86) 30 30 Democracies .11 8.03* (.16) (3.64) .26** 5.83** (.13) (2.75) .19** 6.91*** (.09) (1.96) 82 82


1.99 (7.64) 12.16** (5.90) 1.09 (4.37) 98

−12.94 (8.39) 25.66*** (7.99) 5.76 (4.72) 33

−10.53 (10.70) 18.86*** (6.91) 3.64 (4.21) 48

(6)

Super2000

Notes: Standard errors in parenthesis. ***,**,* represent significance at the 1%, 5%, 10% level respectively. The first stage variables for the propensity score include all variables from Models 4–6, Table 1, plus squared terms of GDP/cap, Population, Pop15-64, and Pop65+.

Observations

Stability


Discipline

Observations

Stability


DiscStrength25

Observations

Stability


DiscStrength25

Treatment Variable

Table A2 Robustness Analysis II: Propensity Score Estimation Method


71

72

Notes: See Table A2.

Observations

Stability


Strength25

Observations

Stability


Strength25

Observations

Stability


Discipline

Treatment Variable Institutional Capacity (2)

−.43** (.18) .23 (.14) .38*** (.10) 50

Old Democracies −8.69* (4.61) 5.54 (3.58) 11.49** (2.74) 50 Proportional Systems .33* 12.29*** (.18) (4.37) .19 4.21 (.13) (3.30) .27** 8.53*** (.11) (2.71) 60 60 All Countries .16 6.25** (.12) (2.78) .32*** 6.38*** (.10) (2.22) .12* 5.53** (.06) (1.46) 124 124


−3.05 (5.58) 12.57*** (4.54) .16 (3.21) 126

5.14 (9.13) 14.93** (6.86) −.32 (5.71) 63

−4.70 (8.77) 11.31* (6.79) 7.08 (5.19) 49

(3)

Super2000

Panel C

Panel B

Stability


Panel A Strength25

Outcome Variable

Robustness Analysis III

Table A3

Institutional Capacity (5)

(6)

Super2000

Old Democracies 8.09* −1.95 (4.62) (10.93) 8.92*** 18.32** (3.40) (7.66) 6.15** .23 (2.71) (6.05) 50 49 Parliamentary Systems .30 −.02 .30 (.26) (5.51) (.26) .36* 16.76*** .36 (.20) (4.11) (.20) .22** 7.46*** .22** (.10) (2.07) (.10) 41 41 44 All Countries (OLS) .04 3.99* −4.72 (.09) (2.33) (4.20) .25*** 4.99*** 11.44** (.09) (1.81) (5.39) .003 2.03 −5.18 (.06) (1.27) (3.78) 125 125 128 .10 (.21) .33** (.16) .23* (.12) 50




POLITICAL INSTITUTIONS, POLITICAL CAREERS AND ENVIRONMENTAL POLICY SUMMARY High party discipline and high party strength induce legislators to stay committed to the party’s promised policies after their election. This is because party leaders are able to punish deviations and reward behavior that is in the party’s interest. Higher political stability induces party leaders to take a longer-term perspective. We investigate the following interrelated hypotheses: (i) the effects of party discipline and party strength on environmental policy are conditional on the degree of political stability; and (ii) the effect of political stability is conditional on the levels of party discipline and party strength. Our empirical findings support these hypotheses.


73