This paper explores the impact of elections on economic policies and governance in developing countries. We distinguish between a structural effect, which increases accountability, and a cyclical effect which may be disruptive. Since the effects are offsetting, neither can be analysed in isolation. We implement an econometric analysis on more than 80 developing countries using positive changes in the Country Policy and Institutional Assessment of the World Bank and the International Country Risk Guide as signalling improvements in economic policy and governance. We find that both structural and cyclical effects matter. The cyclical effect suggests that mid-term is the best moment for policy change. We investigate the structural effect by comparing different frequencies of elections. Except at the extremes, a higher frequency of elections improves both policy and governance net of any cyclical effect. The important exception to this benign net effect is if the electoral process is badly conducted. Badly conducted elections have no structural efficacy for policy improvement. A reasonable interpretation of our results is that honest elections increase accountability and thereby discipline governments to improve economic policy and governance, but that if candidates can win by fraud this chain is broken.
Economic Policy (2009) 24 (59) 509-550. Also published as Dial Working Paper DT/2008-11. Dec 2008.