Ideological polarization and government debt

International Tax and Public Finance https://doi.org/10.1007/s10797-021-09690-1 Ideological polarization and government debt Mickael Melki1 · Andrew Pickering2 Accepted: 4 August 2021 © The Author(s) 2021 Abstract Models of strategic debt predict that public debt increases with polarization, measured by the ideological distance between the government and its likely successor. Conversely if voters are both short-termist and also more likely to switch their vote for parties offering higher spending and public good provision when the electorate is ideologically concentrated, then debt can fall with polarization, measured by dispersion of ideological preferences in the electorate. Using time-varying polarization measures generated from ideology data from party manifestos, we find a sizable and statistically significant negative association between debt levels in OECD countries and ideological polarization in the electorate. Keywords Public debt · Strategic debt · Ideological polarization JEL Classification H63 1 Introduction In 2010 average central government debt in the OECD stood at 69% of GDP. In 1974 this figure stood at 23%, down from 88% in 1945. Moreover, at any point in time there is substantial cross-country variation. 2010 debt levels varied from 22% in Switzerland to 189% in Japan. Whilst it is widely recognized that these outcomes are the product of imperfect political processes, a full explanation represents a formidable challenge to political economics.1 1 As noted by Alesina and Perotti (1995) efficiency-based explanations (e.g. Barro, 1979) alone cannot explain either the levels or variation in public debt observed across countries and time. Recent contributions include Battaglini and Coate (2008) and Yared (2010). Eslava (2011) provides an excellent recent survey. * Andrew Pickering 1 Paris School of Business, Paris, France 2 Department of Economics and Related Studies, University of York, York YO10 5DD, UK 13 Vol.:(0123456789) M. Melki, A. Pickering Persson and Svensson (1989) and Alesina and Tabellini (1990) formalized the idea of ‘strategic debt’. Given the likelihood of being replaced in the future, an ideologically motivated incumbent will encumber future (ideologically distant) governments with debt. The greater the ideological distance between the parties, the greater the level of debt.2 On the other hand, as we discuss in the next section polarized preferences serve to weaken electoral mechanisms which themselves plausibly underpin proclivity to debt. The novel hypothesis is that polarization as defined by greater ideological dispersion in the electorate will instead reduce government debt. This paper investigates the relationship between central government debt and ideological polarization empirically. Figure 1 presents a simple cross-country correlation between central government debt levels in 2010 and the polarization measure generated from the World Values Survey used by Lindqvist and Östling (2010) in their analysis of the size of government. The data sample here is a cross section of quite diverse countries.3 The scatter plot reveals a negative raw correlation between polarization and debt. Prima facie this is supportive of the argument advanced in the present paper, and somewhat contrary to the strategic debt literature. Nonetheless it is not possible to infer causality for the standard reason that analysis of cross-country performance omits considerable unobserved heterogeneity. Below the competing hypotheses are tested using time-varying polarization measures generated from ideology data from political manifestos and observed voting behaviour. One important feature of these data is that they vary over time. Previous analyses of the relationship between fiscal policy and ideological polarization have only used cross-sectional analysis, or in the context of panel data has relied on fixed measures of ideology for party positioning—wherein time variation is generated through variation in seats.4 Within countries the distance between parties is not fixed over time; hence, the data used here represent an improvement over the previous work. Within-country variation also allows us to control for unobserved fixed country-specific characteristics that might drive debt. The econometric analysis consistently finds a statistically significant negative relationship between central government debt and ideological polarization in voter 2 Persson and Svensson (1989) model preference heterogeneity over total government expenditure whilst Alesina and Tabellini (1990) focus on the composition of expenditure. In support of Persson and Svensson (1989) Pettersson-Lidbom (2001) finds that right-wing governments increase debt whilst left-wing governments reduce debt when faced with the likelihood of being replaced. However this is a different, and not mutually exclusive, hypothesis from that pertaining to the ideological distance between the two parties. For example in Persson and Svensson’s (1989) model the impetus for debt (given a conservative incumbent) is stronger when the ideological distance between the two parties is bigger. 3 The econometric analysis below focuses only on established OECD democracies for three reasons. First, both our model and the strategic debt literature rely on democratic processes. Second, debt default concerns have generally been greater outside the OECD. Third, over a meaningful time-span the manifestos data, which we exploit to generate time-varying ideological dispersion (polarization) measures, are only available for the OECD sample. 4 For example Alt and Lassen (2006). Relatedly Volkerink and de Haan (2001) and Perotti and Kontopoulos (2002) use cross-sectional policy to investigate the effect of fragmented government on fiscal policy. 13 Ideological polarization and government debt preferences. The relationship is robust to the inclusion of a number of controls and, in particular, is strengthened when a measure of fragmentation (which is distinct from polarization) is included. In contrast to the strategic debt literature we find that it is ideological dispersion in the electorate, rather than ideological distance between the government and their potential replacement, that correlates with debt. A one standard deviation increase in polarization correlates with lower central government debt by about 12% of GDP. To identify exogenous variation in polarization, we use lagged media penetration data and also the impact of the fall of the Berlin Wall on European politics. Campante and Hojman (2013) and Melki and Pickering (2014) both argue for a causal relationship from increased media penetration to reduced polarization. Using the alternative instrumental variables the results hold up in support of a causal negative relationship between polarization and debt. Furthermore, the negative relationship between debt and polarization is found to be stronger when ‘government efficiency’ is weaker. This latter variable (...truncated)