The dynamics of within-regime stability: Party tenure and economic performance

Duration and economic growth

Economic growth requires the investment of savings in capital and technology to improve future production possibilities. Yet, political risk – the potential of capricious actions by the state – undermines investors’ incentives and saps performance (North and Thomas, 1973). New ruling parties may have substantially different policy agendas from their predecessors; indeed, changes in ruling parties have been found to be associated with large shifts in macroeconomic policy (Alesina and Rosenthal, 1995; Bunce, 1980; Cameron, 1978; Castles, 1982; Cowart, 1978; Hibbs, 1977; Lewis-Beck and Rice, 1985; Mosley, 1984; Schmidt, 1982). Absent information concerning particular governments’ policy preferences, investors face uncertainty and are unable to assess these risks effectively. Heightened uncertainty has a chilling effect on investment as it precludes the accurate comparison of risk-adjusted returns; policy uncertainty rewards waiting and hence dampens investment incentives (Dixit, 1989).

Candidate party platforms give policy cues but cannot completely eliminate the uncertainty associated with a new ruling party. Parties create reputations for certain economic policies over time such that some may be more welcomed by investors than others (Petrocik, 1996). However, skepticism cannot be eliminated altogether because a party’s ability as currently constituted to enact an economic program is uncertain, particularly if the party has not been a part of the government in the past. Even when parties have experience in prior governments, current party members may not have served in the government and, moreover, macroeconomic circumstances may have changed. This especially applies to newer democracies since they often have little or no track record in economic policy (Keefer and Vlaicu, 2008).

Even fully rational and forward-looking actors may lack information about a government’s policy preference or competence (Cukierman and Meltzer, 1986; Persson and Tabellini, 1990; Rogoff, 1990; Rogoff and Sibert, 1988). Empirically, economic actors are wary of predicting the policy behavior of ruling parties with short track records (Backus and Driffill, 1985). In the US, such simple partisan mappings of behavior would not have predicted Clinton’s retrenching of welfare, Reagan’s base-broadening tax measures, or Nixon’s wage and price controls and devaluation.

Observation of government policy behavior does alleviate this uncertainty. Governments’ difficulty in making credible commitments makes assessing a new ruling party’s policy preference on the basis of platforms difficult. Observed behavior, however, gives a clear indication of policy preferences. Even ruling parties that are initially perceived as potentially more hostile to business will see improved performance with duration, as investors become better able to calculate political risks. As the largest gains in the reduction of uncertainty come with the initial observation of behavior, economic performance should improve most in the initial phases of party tenure.

However, extended duration in power of ruling parties leads to institutional sclerosis and deteriorating economic performance. As the marginal returns from the stabilization of policy expectations diminish, ruling parties come under increasing special interest pressure to redistribute resources, hindering growth in four important ways. First, market intervention to create rents for narrow interests distorts incentives, drawing resources from efficient uses (Becker, 1983, 1985). Second, the widespread prevalence of redistributive policies incentivizes rent-seeking, diverting real resources from productive activity to lobbying (North, 1990). Third, vested interests lobby for government policies that hamper innovations (Krusell and Rios-Rull, 1996). Fourth, incumbent parties use the public budget for partisan purposes instead of acting as effective gatekeepers (Geddes, 1994; Shefter, 1994).

Ruling party duration should be associated with decreasing economic performance as rent-seeking draws resources away from productive activity. Responsiveness to rent-seeking suggests that narrow interests have achieved political victory over broad interests who benefit from the growth of the economy as a whole. Olson (1982, 1983) suggests that rent-seeking interest groups accumulate over time during periods of stability. Similarly, in Hellman’s (1998) partial reform equilibrium model, ruling party stability allows the winning constituency to lock in reforms that provide them with private benefits at a cost to greater economic efficiency. Extended government duration also increases the likelihood that vested interests maintain political connections with the ruling party and use them to stymie new, innovative competition.

The Liberal Democratic Party (LDP) in Japan provides a quintessential example of a sustained tenure contributing to institutional sclerosis. Ruling continuously from 1955 to 1993, the LDP subsidized public construction works, supporting 560,000 construction companies with 6.7 million employees. Although this lavish infrastructure spending initially stabilized the economy and contributed to the Japanese miracle, it was increasingly criticized over time as expensive and inefficient. However, this spending became entrenched due to an iron triangle among LDP politicians belonging to the construction “tribe,” bureaucratic agencies, and construction companies. The construction companies colluded and submitted artificially high bids, while the LDP politicians received kickbacks and election contributions (Broadbent, 2002). Such activities were implicated in the collapse of the bubble economy in the 1990s.

The policy expectations and institutional sclerosis effects of ruling party duration run in opposite directions, causing potential uncertainty about the total impact of duration on economic performance. However, the timing of these effects within a particular ruling party’s tenure in office allows us to make predictions about which effect will dominate. We expect the largest improvements in economic performance to come early in the ruling party’s tenure. Investors’ observation of government policy choices yields the largest improvements in stabilizing policy expectations with initial behavior, producing diminished marginal returns over time. By contrast, the institutional sclerosis effect of duration operates over longer durations, as rent-seeking activities accumulate. As a result, we expect:

Hypothesis 1: There is an inverted U-shaped relationship between ruling parties’ tenures and economic growth.

The timing of hypothesized effects in models of comparative political economy is often under-theorized and treated as an empirical matter. We face the same difficulty in predicting when the turning point suggested in Hypothesis 1 will arrive. Essentially, this requires us to posit the rate at which policy expectations stabilize for investors and compare it with the rate at which the institutional sclerosis accumulates.

For investors, reduction of uncertainty comes from the observation of ruling party behavior in fiscal, monetary, and regulatory policy. While fiscal behavior can be observed quickly given budget cycles, regulatory policy may play out over a longer period of time and monetary policy requires observation after a second election (to assess time inconsistency). This suggests that the benefits of duration associated with the reduction of uncertainty may even extend for a number of years past the first term.

New ruling parties may quickly seek to reward narrow constituencies with public resources. However, concern over losing the subsequent election due to pandering to special interests might restrain this until a second term or later, when the party has consolidated its position. Olson (1982, 1983) suggests that the damage occurs over time as the drag of rent-seeking impedes dynamism and actors increasingly shift real resources away from investment and toward seeking government benefits. Taken together with the discussion of the policy expectations effect, this suggests that the turning point on economic performance might not arrive until a ruling party’s second or third term in office, a period ranging from five to 12 years.

Veto player interactions

Extensive research has focused on the role played by political institutions in shaping the prospects for economic growth. Ruling parties do not make economic policy in isolation, but operate within an institutional framework that constrains their agendas. If political institutions moderate the two duration effects – policy expectations and institutional sclerosis – we should observe different relationships between growth and duration across institutional contexts. These relationships should change in predictable ways given prior work on the effects of institutions on policy outcomes.

We consider how veto players moderate the impact of two duration factors. Tsebelis (1995, 2002) develops the concept of veto players to assess governments’ ability to produce policy change. Veto points are institutional features that allow individual or collective actors within the political system to veto changes in policy. The combination of a large number of institutional veto points and their control by actors with divergent preferences makes policy unlikely to change.

Although there are other institutional variables, we choose to utilize the presence of veto players for two reasons. First, choosing veto players allows us to make comparisons across a broad range of countries, including presidential, semi-presidential, and parliamentary systems with a multitude of institutional variations (MacIntyre, 2001). To those interested in a detailed accounting of the interplay between these institutional features (e.g. coalition government versus single party government, legislative rules and committees, etc.) and duration, collapsing them into a single dimension will seem unsatisfying. However, this is an analytically tractable way to estimate the effects of duration given the econometric minefield we describe later. Second, extant literature on veto players generates clear hypotheses about how their presence alters the duration effects we specify.

Economic growth requires seemingly conflicting characteristics in governments (MacIntyre, 2001). On the one hand, governments must provide stable policy environments with predictable behavior from political executives. On the other hand, governments must be able to respond to exogenous shocks to the economy as well as to reverse prior policy mistakes. Haggard and McCubbins (2001) contend that these two important characteristics, which they label decisiveness and resoluteness, necessarily trade off in institutional design.

We utilize this insight to predict how the numbers of veto players and the ideological distance among them moderate the effects of ruling party duration on growth. The foregoing theoretical discussion assumes that the ruling party is relatively less constrained (i.e. it faces relatively fewer effective veto players). As Tsebelis (1995) argues, the absence of effective veto players increases the likelihood of large policy departures from the status quo. This policy flexibility creates uncertainty among investors, impeding economic performance. With time, the growing track record of the ruling party builds investor confidence in the market. However, the economic returns for building reputation diminish over time, while pressures to increase rents for narrow constituencies mount.

On the other hand, the presence of many veto players and of greater ideological distance among them stabilizes policy outcomes, altering the relationship between duration and performance. As the dominating characteristic of the government is resoluteness, checks on the policy discretion of the ruling party can help alleviate the chilling effect of unstable policy expectations on investor confidence in several ways. First, constraints on the executive can enhance the credibility of policy reforms undertaken as retrenching such measures will face greater opposition (Henisz, 2000; MacIntyre, 2001; Tsebelis, 2002). Second, veto players can diminish the likelihood of large deviations from the status quo policy environment (Tsebelis, 1995). Indeed, Alesina and Rosenthal (1995) suggest that in the US, rational voters will choose divided governments precisely for the purpose of managing this kind of risk. Third, Keefer and Stasavage (2002, 2003) find that veto players enhance the credibility of monetary policy choices, tamping down inflationary pressures. Finally, veto players have a general effect of enhancing the protection of property rights and the rule of law (Andrews and Montinola, 2004).

New ruling parties constrained by veto players will enjoy an initial advantage in credibility relative to their unconstrained counterparts that should improve economic performance by stabilizing investor confidence. However, this initial advantage decays as time passes and changing economic conditions require effective policy responses. Governments with numerous veto players are less nimble amid economic shocks as “high levels of commitment” mean “inability for political response” (Tsebelis, 2002: 204). Facing more effective veto players complicates the task of ruling parties of coordinating policy responses to these events by decreasing the win-set of policy choices. Poterba (1994) and Alt and Lowry (1994) argue that US states with divided governments respond more slowly to fiscal shocks. Under such circumstances, the government loses the capacity to create investment opportunities, contributing to economic downturn.

Some episodes from the Asian financial crisis of the late 1990s can also illustrate these points. According to MacIntyre (2001), Thailand’s fragmented party system represented a typical case of resoluteness with multiple veto players. Its coalition governments involved six or more incoherent parties whose members could vote differently from their party leaders. When Chavalit’s government attempted to reduce government spending to ameliorate the current account and to help loosen monetary policy to encourage growth, it faced strong opposition by a coalition member who feared the loss of pork-barrel construction projects. Policy rigidity created by veto players hampered an effective response to the financial crisis.

However, the declining performance of ruling parties facing many veto players may improve with further duration for two reasons. First, constrained ruling parties can overcome the policy gridlock created by multiple veto players. With a cross-national analysis, Scartascini et al. (2010) demonstrate that large numbers of veto players can achieve productive policy cooperation over time, defeating the dilemma predicted by Haggard and McCubbins.

Second, veto players can forestall rent-seeking activity by ruling parties, thwarting the institutional sclerosis in later phases of duration predicted by Olson. Evidence suggests that veto players insulate governments from special interests in the areas of state infrastructure investment (Henisz and Zelner, 2006; Keefer and Knack, 2007) and protectionist policies (Henisz and Mansfield, 2006). In the absence of effective veto players, parties become more clientelistic, creating rents for narrow constituencies at the expense of the economy as a whole (e.g. the LDP in Japan). Given the potential for policy cooperation and the prevention of institutional sclerosis, we expect:

Hypothesis 2: In the presence of numerous veto players, the inverted U-shape relationship between duration and growth is flattened and bent into a regular U-shape.

It is possible that numerous veto players could create log-rolling dynamics, leading to bloated fiscal deficits. Empirically, veto players (Perotti and Kontopoulos, 2002) and coalition or divided governments (Roubini and Sachs, 1989) tend to be associated with larger budget deficits as they increase the number of constituent groups that must be satisfied. This could be detrimental through either the crowding out of private investment or rising inflation. However, accommodating special interest pressure through executive intervention in the regulatory apparatus could create even greater distortions. In cases such as Japan’s LDP and Italy’s Christian Democrats, the abuse by a single dominant party without political constraints appears to be more damaging to the economy. This seems particularly true when these unconstrained ruling parties stay in power for an extended period of time. Budget outlays limit the distortion of incentives to the recipient, rather than rearranging them for all actors economy-wide. Moreover, they face an explicit budget constraint, whereas predatory regulations can be extended without large fiscal outlays. If veto players prevent regulatory rent-seeking and instead channel it into competition over budget resources, the net effect on economic performance may be positive.

Data and measurement

Typical of duration analysis, each case in our data set is a country-month, creating two potential problems. First, many of the variables are measured yearly, as is common in government duration models, even when cases are months. For variables with this interpolation problem, we apply the annual value to each month within the year. Though not ideal, when collapsing the models from monthly to annual analyses, we find that each of our main theoretical findings remains significant and the coefficients are correctly signed. Second, unlike much of the work on the determinants of growth, we do not utilize 10-year averages of growth rates. While this might make our findings more susceptible to business cycle effects, we see no alternative in assessing the effects of ruling party duration. Further, we note that our findings from the following growth models nonetheless correspond to those of the endogenous growth literature.

Each case in the data set represents a month for a country that has been classified as a democracy (i.e. POLITY score of six or better), including presidential and semi-presidential systems and parliamentary systems. The POLITY data are used since they cover a large number of countries, and the information is updated regularly. Table A1 in the Appendix (accessible online) shows the list of countries and years covered in the analysis. ⁸