Beyond economic voting: Do voters hold incumbents to account for the country’s happiness?

George Ward

06 May 2015



Politicians have long focused on GDP as the main measure of national performance. And with good reason — one of the most-replicated and well-known findings in economics and political science is that voters reward incumbents at the ballot box during periods of economic prosperity, and punish them during downturns (e.g. Fair, 1978; Kramer, 1971).

However, the last few years have seen a growing interest in measures of societal progress that go “beyond GDP” (Fleurbaey 2009). One option that has gained in popularity is to ask people directly how happy or satisfied they are with their lives (Krueger and Stone, 2014). In line with recommendations made by the Stiglitz Commission (2010) as well as reports by the EU (2009), OECD (2013) and US National Research Council (2014), many countries and international organisations are beginning to collect subjective wellbeing (SWB)—or ‘happiness’—data in order to gauge success and guide policymaking.1

But do voters themselves judge performance in such broad terms at the polls? And is it in politicians’ electoral interest to focus policy on anything beyond GDP?

Well-being and voting

In a recent paper (Ward, 2015), I examine whether the electorate holds incumbents to account for the happiness of society. Analysing wellbeing data alongside general election results from a long-run panel of European countries, I find evidence that well-being shapes election outcomes: good times keep incumbents in office, misery throws them out.

Figure 1 reports the extent to which a country’s aggregate level of life satisfaction is able to account for the variance in vote shares won by governing parties in general elections over the past four decades. Plotted is the actual vote share against the value predicted from a regression of cabinet vote share on aggregate SWB (together with a set of country and year dummies as well as various covariates standard in the cross-country economic voting literature).2 The regression reveals a significant positive relationship between aggregate happiness indices and cabinet vote share. A one standard deviation change in a country’s self-reported well-being over time is associated with a swing in incumbent vote share of around 8.5 percentage points.

Figure 1. Using SWB data to account for variance in the electoral fate of governing parties

Political incentives

The key implications of this lie in the incentives that voting behaviour gives to politicians. A growing literature on political economy shows that, provided the benefits of re-election are sufficiently large, the threat of being voted out of office induces politicians to act in voters’ interests (Besley, 2006).

Evidence of ‘economic voting’ presents politicians with electoral incentives to ensure a healthy economy, but the empirical focus on economic accountability leaves open the possibility that officeholders may be left unrestrained in a range of other policymaking areas that matter to people. Evidence of a link between happiness and voting over and above the state of the economy suggests there is also an electoral dividend for incumbents that focus policy on a broader range of policy domains influencing how satisfied people are with their lives. A burgeoning economic literature on the determinants of SWB suggests people’s well-being is sensitive to a number of policy-relevant factors such as mental and physical health, noise and air pollution, education, community cohesion and social relations, corruption, the effectiveness of government services, crime rates, as well as income and employment (Clark, Layard and Senik, 2012; Helliwell et al., 2014). 

Going beyond economic voting

The economy is itself important for people’s well-being (Di Tella et al. 2003, Stevenson and Wolfers 2008).  Nevertheless, even once macroeconomic controls like the economic growth, unemployment and inflation rates are included in the equation, SWB remains a significant predictor of electoral outcomes. When included together in the equation, both economic growth and life satisfaction are significant predictors of cabinet vote share, with the effect size of SWB twice that of growth.3 Moreover, SWB is able to account for more of the variance in government vote share than any one of the standard macroeconomic variables.

That retrospective voting goes beyond GDP is perhaps unsurprising.  Given that governments act across a variety of policy domains, there seems no immediate reason for voters to evaluate incumbent performance solely on the basis of economic outcomes. For example, recent evidence on the impact of the Moving to Opportunities experiment in the US – which gave the opportunity (via a random lottery) to families living in high- poverty areas to move to more prosperous neighbourhoods – shows that the programme led to improvements in SWB but had no significant effect on employment or earnings (Ludwig et al. 2013).  We might reasonably expect the well-being increases associated with such a policy to have at least some bearing on participants’ opinions about government performance.  If we take a narrow focus on economic outcomes, however, we would not expect this kind of policy to have any effect on voting behaviour at all.

None of this is to say that the economy is unimportant.  Electoral outcomes are best predicted in the sample by a combination of both SWB and other economic variables, each of which provides additional explanatory power. The finding that both are separately important accords with recent work suggesting that well-being is multifaceted and dependent upon a number of different factors including both life satisfaction as well as material prosperity (Benjamin et al. 2014; Glaeser et al. 2015).

Future directions

The use of SWB data in political economy is in its infancy, but as a broad proxy for social welfare it offers potentially interesting new avenues for future empirical research. One advantage of SWB is that it sits well with the informational asymmetry at the heart of many political economy models. Typically, in such models voters decide whether to re-elect politicians but are unable to directly observe their actions and/or type, and are instead left - as in Ferejohn’s (1986, p.11) seminal model - ‘to assess the effects of governmental performance on their own well-being’. This informational asymmetry relates to the popular notion of rational ignorance (Downs 1957), which highlights the high cost of acquiring and filtering information relative to the low probability of a vote being decisive to an electoral outcome. Observing one’s own satisfaction with life is essentially costless and does not require voters to have a thoroughgoing understanding of complex economic and political issues. As Fiorina (1981, p.6) notes, even though people often lack relevant knowledge, they ‘typically have one comparatively hard bit of data: they know what life has been like during the incumbent’s administration’.

Although the evidence I present in Ward (2015) is consistent with rational accounts of electoral behaviour in which voters learn about an incumbent’s quality through observable outcomes such as their own welfare, additional research may well examine the extent to which this retrospective voting is subject to emotional and cognitive biases. For example, further models suggest that voters seem to blame the government for falls in their life satisfaction more than they reward them for equivalent improvements. This asymmetry is consistent with idea that individuals are prone to a ‘self-serving bias’ (Miller and Ross 1975), whereby they relate success to their own efforts but attribute failure to situational factors and the actions of others. Another interesting area of future investigation may well focus on whether certain aspects of people’s happiness (policy-relevant or otherwise) are more important than others in influencing voting decisions, and what this means in terms of political incentives.

Concluding remarks

Many factors go into determining an election result. Cross-national analyses of past patterns of well-being and voting are not a rival to election forecasting models that use polling on voting intentions.4 Rather, this line of research is an attempt to uncover what is behind people’s voting behaviour and shed light on what it is that people want their elected officials to be concentrating their efforts on. The data suggest it is in politicians’ interest not only to make voters financially better off, as the extensive literature on economic voting suggests, but also to formulate policy focused on improving their broader well-being.

Although policymakers have only recently begun to think of success in terms of national happiness, evidence using subjective wellbeing and electoral data stretching back to the early 1970s suggests that voters themselves seem to have long evaluated government performance in terms that go beyond GDP.


Benjamin, D J, O Heffetz, M S Kimball and N Szembrot (2014), “Beyond Happiness and Satisfaction: Toward Well-Being Indices Based on Stated Preference”, The American Economic Review 104(9), pp. 2698–2735.

Besley, T (2006), Principled agents? The political economy of good government, Oxford: Oxford University Press.

Clark, A, R Layard and C Senik (2012), “The causes of happiness and misery”, World Happiness Report, The Earth Institute, Columbia University, pp. 58–89.

Di Tella, R, R J MacCulloch and A J Oswald (2003), “The macroeconomics of happiness”, Review of Economics and Statistics 85(4), pp. 809–827.

Downs, A (1957), An Economic Theory of Democracy, New York: Harper and Row.

European Commission (2009), “Communciation from the Commission to the Council and the European Parliament - GDP and Beyond: Measuring Progress in a Changing World”, 20 August.

Fair, R C (1978), “The effect of economic events on votes for president”, Review of Economic Statistics 60, pp. 159–73.

Fiorina, M P (1981), Retrospective voting in American national elections, New Haven: Yale University Press.

Fleurbaey, M (2009), “Beyond GDP: The quest for a measure of social welfare”, Journal of Economic Literature 47(4), pp. 1029–1075.

Glaeser, E L, J D Gottlieb and O Ziv (2015), “Unhappy Cities”, Journal of Labor Economics, forthcoming.

Helliwell, J F, H Huang, S Grover and S Want (2014), “Empirical linkages between good government and national well-being”, NBER Working Paper No. 20686.

Kramer, G H (1971), “Short-term fluctuations in U.S. voting behavior 1896-1964”, American Political Science Review 65, pp. 131–43.

Krueger, A B and A A Stone (2014), “Progress in Measuring Subjective Well-Being: moving toward national indicators and policy evaluations”, Science 346(6205), pp. 42–43.

Ludwig, J, G J Duncan, L A Gennetian, L F Katz, R C Kessler, J R Kling and L Sanbonmatsu (2013), “Long- Term Neighborhood Effects on Low-Income Families: Evidence from Moving to Opportunity”, The American Economic Review 103(3), pp. 226–31.

Miller, D T and M Ross (1975), “Self-serving biases in the attribution of causality: Fact or fiction?” Psychological bulletin 82(2), pp. 213.

National Research Council (NRC) (2014), Subjective well-being: Measuring happiness, suffering, and other dimensions of experience, National Academies Press.

OECD (2013), OECD Guideline on Measuring Subjective Wellbeing, Paris.

Stevenson, B and J Wolfers (2008), “Economic Growth and Subjective Well-Being: Reassessing the Easterlin Paradox”, Brookings Papers on Economic Activity 39(1), pp. 1–102.

Stiglitz, J E, A Sen and J-P Fitoussi (2010), Report by the Commission on the Measurement of Economic Performance and Social Progress, Paris: Commission on the Measurement of Economic Performance and Social Progress.

Ward, G (2015), “Is Happiness a Predictor of Election Results?”, CEP Discussion Paper No. 1343.


1 The UK Office for National Statistics has added subjective well-being questions to a number of its Labour Force survey, for example, whilst the US Bureau of Labor Statistics has also begun to include questions on respondents’ emotional state to its Time Use Survey.

2 Linear predictions are plotted from an OLS regression. The 45-degree line is for reference, and represents what a perfect prediction would be.  Further covariates: party fractionalisation, length of parliament, previous vote share, number of parties in government. Sample is 134 elections in the so-called EU-15 countries 1973- 2012. SWB is the aggregate level of life satisfaction from the Eurobarometer survey closest in time prior to each election.

3 One standard deviation changes in aggregate life satisfaction and economic growth are significantly associated with 0.72 and 0.36 standard deviation changes in cabinet vote share respectively.

4 Although this type of analysis is not well suited to making out-of-sample predictions of future electoral outcomes, the increasing availability of high-frequency, high-quality SWB data (such as that now being collected by the ONS) may make election-forecasting models using happiness data an interesting avenue of future research.



Topics:  Frontiers of economic research

Tags:  happiness, elections, well-being, politicians, voters, Happiness economics

PhD student at MIT and research associate of the Centre for Economic Performance, LSE