Elections and Economic Policy in Developing Countries

This paper investigates whether elections in developing countries have improved economic policies and economic governance. Both casual empiricism and casual theorizing suggest that they have done so. As contested elections have become more common since the 1990s, the policy ratings from the World Bank and the International Country Risk Guide have both improved markedly. These improvements accord with the fundamental notion that elections discipline governments into good performance. Yet this view from on high often collides with the actual experience of individual elections. The Kenyan election of December 2007 triggered a catastrophic implosion of the society, polarizing it on ethnic lines. To date, the legacy of that election is a policy paralysis: for example, the number of government ministers has been doubled with a resulting loss of policy coherence. In Zimbabwe the prospect of contested elections in 2002 and 2008 clearly failed to discipline President Mugabe into adopting good economic policies: he chose hyperinflation, using the revenues to finance patronage. The most celebrated economic reform episode in Africa is Nigeria 2003-6, when a group of technocrats led by Ngozi Nkonjo-Iweala as Minister of Finance turned the economy around. This episode was ushered in by the replacement of a military dictator, General Abacha, with an elected president, Obasanjo, suggesting that elections indeed improved government performance. However, reform only began in Obasanjo’s second and final term, when he no longer faced the discipline of an election. He told Nkonjo-Iweala that the window for reform was only three years, not the full four years of his term: as he said, ‘the last year will be politics’.2 Indeed, that Nigeria failed to harness the first oil boom was primarily the responsibility of a democratic government, elected in 1978. That government adopted very poor economic policies, including borrowing heavily in order to finance public consumption; it was also famously corrupt. Despite its disastrous performance it was re-elected in 1983. As these examples suggest, in the conditions typical of many developing countries, elections may be two-edged swords. The effect of elections on policy in low-income countries is of considerable importance. Since aid was first used conditionally to promote ‘Structural Adjustment’ in the 1980s the international community has recognized that policy improvement is fundamental to development. During the 1990s the approach to how good policies should be promoted shifted from conditionality, which was increasingly seen as both ineffective and unacceptable, to the promotion of democracy. Electorates rather than donors would coerce governments into good performance. At the core of the promotion of democracy was the promotion of elections: for example, in 2006 donors provided $500m to finance elections in the Democratic Republic of the Congo. Yet the premise that elections are effective in such conditions has yet to be evaluated. At a more pragmatic level, since elections periodize political decision taking, they might also periodize policy reform: some times might be ripe for good policy. For example, it would be useful both to political leaders and to donors to know whether the year just prior to an election is indeed unsuited to policy reform as President Obasanjo evidently thought. There is a large general literature on the relationship between democracy and economic performance but it does not provide much guidance as to these questions. The conclusion from the literature is that any such relationship is weak (Drazen, 2000; Feng, 2003; Przeworski et al., 2000). However, these studies do not focus specifically on the characteristics prevalent in developing countries many of which democratized during the 1990s. Several recent studies find that democracy has distinctive effects in the context of such characteristics. Alesina and La Ferrara (2005) and Collier (2001) focus on the consequences of ethnic diversity for growth. They find that diverse societies benefit significantly more from democracy than homogenous societies. Collier and Rohner (2008) focus on the relationship between democracy and the risk of large-scale political violence. They find that whereas in developed economies democracy increases security, below an income threshold of around $2,700 per capita it significantly increases the risk of political violence. Finally, Collier and Hoeffler (2008) focus on the relationship between democracy and the economic performance of resource-rich countries. They find that whereas below a threshold of resource wealth democracy is significantly beneficial, above the threshold it significantly worsens performance. These results suggest that no simple model of how democracy affects economic policy may be globally applicable. Models designed to describe how elections affect political incentives in OECD societies may prove seriously misleading if applied to contexts such as Afghanistan and the Democratic Republic of the Congo. Our purpose in this paper is to investigate empirically the various effects of elections. As we discuss in Section 2, since there are potentially several distinct and offsetting effects, an appropriate empirical strategy needs to distinguish between them: simple composite empirical measures are likely to be misleading. In many developing countries governments are failing to provide their citizens with the rudiments of social provision and economic opportunities now considered both normal and feasible. A reasonable inference is that in such states the ruling politicians are either ill-motivated or incompetent. We focus directly on policies rather than on economic outcomes. The typical developing country is subject to large shocks that introduce much noise into the mapping from policy choices to outcomes and we are concerned with the variables that are more directly under the control of politicians. The direct observation of policies is difficult, but in this respect the researcher is at an advantage over the electorate. We rely upon two international data sets which rate economic policies and governance, neither of which has been available to citizens of rated countries. Hence, while citizens must largely rely upon observable economic performance, we are able to observe policies directly, albeit with the limitations implied by these international rating systems. In Section 2 we discuss the theory which informs our empirical analysis. We show that while democracy may have both structural and cyclical effects on policy, a priori there are offsetting effects so that the net effect is ambiguous: the issue must therefore be resolved empirically. In Section 3 we discuss our empirical strategy: no one approach is ideal and so we test the robustness of each against the reasonable alternatives. In Section 4 we present our core results, subjecting them to a range of robustness tests in Section 5. Section 6 extends the analysis to those developing countries at the extremes of poverty and poor policies, and investigates possible interactions between elections and other political variables. Section 7 draws out the implications for policy, both in terms of international support for the process of policy reform, and for the better functioning of the new low-income democracies.