“The people want the fall of the regime”: Schooling, political protest, and the economy

Highlights

• Individual economic circumstances are a key link between schooling and political protest.

• Highly educated individuals with low income outcomes are more likely to protest.

• Rising education with macroeconomic weakness is associated with political turnover.

• This combination is associated with subsequent democratization.

Abstract

We provide evidence that economic circumstances are a key intermediating variable for understanding the relationship between schooling and political protest. Using the World Values Survey, we find that individuals with higher levels of schooling, but whose income outcomes fall short of that predicted by their biographical characteristics, in turn display a greater propensity to engage in protest activities. We discuss a number of interpretations that are consistent with this finding, including the idea that economic conditions can affect how individuals trade off the use of their human capital between production and political activities. Our results could also reflect a link between education, “grievance”, and political protest, although we argue that this is unlikely to be the sole explanation. Separately, we show that the interaction between schooling and economic conditions matters too at the country level: Rising education levels coupled with macroeconomic weakness are associated with increased incumbent turnover, as well as subsequent pressures toward democratization.

Introduction

Every so often, longstanding and seemingly stable incumbents have been challenged by a groundswell of discontent and public protest, and even been forced out as a result, as vividly illustrated by the recent events of the Arab Spring. While it is understood that simmering public dissatisfaction can suddenly erupt in revolution (Kuran, 1989), we are naturally led to ask what systematic forces or structural conditions might lead such dissatisfaction to arise and get translated into widespread political action.

A natural starting point for this investigation is one of the best-known empirical relationships in the social sciences, namely the strong positive correlation between schooling and political participation. It is well-known that more educated citizens display a greater propensity to engage in virtually all forms of political activity, ranging from more mundane acts such as voting and discussing politics, to the more public forms of mobilization such as attending political events and demonstrating.¹ Several recent studies have shown, however, that favorable aggregate economic circumstances in the labor market can mute the propensity of individuals to engage in political activities (Campante and Chor, 2012a, Charles and Stephens, 2013). Conversely, the high unemployment rates that persisted in many Arab countries for a broad swath of educated workers have proved to be relevant for understanding the political upheaval of the Arab Spring (Campante and Chor, 2012b).

This paper attempts to extract broader lessons from this interaction between schooling and economic conditions. We start off by looking at individual-level data, asking whether personal economic circumstances affect the propensity of educated individuals to be involved in political activities. To address this, we use data from the World Values Survey (WVS), which includes information on different forms of political participation across a broad sample of countries. We first construct a measure of the extent to which an individual’s actual income deviates from that which is predicted by a comprehensive set of biographical characteristics, including education. This residual thus constitutes a measure of relative income performance, in comparison with other individuals with similar attributes.

We find robust evidence that the interaction between this income residual and one’s education level is negatively correlated with political participation. In particular, the more positive an individual’s income residual, the less responsive to education is one’s propensity to engage in political activities. Conversely, more highly educated individuals whose actual income nevertheless under-performs that predicted by their biographical characteristics are more likely to devote their human capital toward active political involvement. Of note, the above pattern holds true precisely for protest modes of participation such as demonstrations, strikes or occupying buildings, but not for more civic and less time-intensive activities such as signing petitions or discussing politics. In short, we see that economic under-performance at the individual level conditions the link between schooling and political protest.

We discuss at some length several interpretations of this strong empirical regularity. First, the above patterns could be driven by the decisions that individuals make over the competing uses of their human capital in production versus (effort-intensive) political participation: Positive productivity shocks or shifters that raise one’s ability to earn income from one’s human capital would naturally discourage its use in political activities, an idea which we develop formally in the Model Appendix. Alternatively, it could be that the link between economic under-performance and political protest operates through “grievance”: An individual’s disappointment with his personal economic situation can lead to political disaffection, and perhaps especially so for the more educated. We will present some evidence indicating that grievance motives are relevant, but that they are also unlikely to fully account for our findings, as our key results related to the income residual survive when we control directly for survey measures of individual dissatisfaction.² A third possibility is that our results could reflect a form of self-selection. For example, there could be an omitted individual characteristic (such as ideology) which tends to raise one’s intrinsic propensity to protest, but which also could be correlated with education and labor market outcomes. Though the nature of our data makes it difficult for us to tease out a causal effect, we can attempt to proxy for such an omitted characteristic with variables that would presumably correlate strongly with it. As it turns out, our results are robust to including such proxies as an individual’s political positioning and religiosity. This suggests that the relationships we have found are unlikely to be due solely to the self-selecting behavior of ideologically-extreme individuals, although we acknowledge that other non-causal stories cannot be ruled out.

In the second half of the paper, we explore whether the interaction between schooling and economic circumstances also has aggregate consequences, in particular for the stability of incumbent regimes. We assemble a panel comprising more than 120 countries and spanning 1976–2010, combining updated data on incumbent turnover collected in Campante et al. (2009) and information on country schooling from the Barro and Lee (2010) dataset; in particular, we focus on average years of secondary schooling as a measure of country-level schooling outcomes. Motivated by our earlier findings, we first proxy for country economic circumstances by using the median income residual (computed from the WVS) for the subset of individuals in each country who had attained at least some secondary education. We later turn to real GDP per capita as an alternative measure of economic conditions, as this latter variable offers more balanced coverage that allows us to apply more powerful panel data estimation techniques (including controlling for country fixed effects). Our results here indicate that the combination of high levels of country secondary schooling and relatively weak economic prospects is indeed associated with a higher frequency of incumbent turnover.

We also find that the same interaction between country secondary schooling and income per capita exhibits a robust negative correlation with country democracy scores, taken from the Polity IV dataset. This is consistent with the increased incumbent turnover being accompanied by some degree of institutional change, in the direction of greater democratization. These patterns suggest that any effect of increased education on the prospects for democracy could be contingent in an important way on macroeconomic circumstances.

In sum, we find that individual economic circumstances play a key intermediating role in governing the strength of the positive association between the individual’s education and her propensity toward political protest. These relationships appear to extend to the country level, where rising levels of education without commensurate economic gains are associated with greater uncertainty of tenure for the political incumbent.

This paper builds naturally on Campante and Chor (2012a). While the focus in this earlier paper is on how the interaction between individual schooling and aggregate economic conditions affects political participation, we show here that a similar interaction is present when looking at income shocks at the individual level. We further explore here the connection with country-level political outcomes, a link which had been hinted at in the specific context of the Arab Spring in Campante and Chor (2012b).

In terms of the broader literature, our findings relate to a large body of work stressing the role of the cost of political engagement in terms of other productive uses of resources. This has been emphasized in the study of regime transitions (Acemoglu and Robinson, 2005, Burke and Leigh, 2010, Brückner and Ciccone, 2011, Aidt and Leon, 2012), civil wars (Grossman, 1991, Collier and Hoeffler, 2004), and political violence in general (Besley and Persson, 2011), and is a standard feature of rational models of political participation (e.g., Riker and Ordeshook, 1968). We specifically focus on the interplay of this logic with individual education and one’s propensity toward political protest. The various interpretations of our evidence that we offer combine elements of the “resource” view that links socioeconomic status (such as education) with political engagement (e.g., Verba et al., 1987), as well as of the “disaffection” view of political protest (e.g., Gurr, 1970). The interaction with relative income performance also ties in with a longstanding sociological literature that links political protest with concepts such as “relative deprivation” and “status inconsistency” (e.g., Lenski, 1954; also, see Orum, 1974 or Gurney and Tierney, 1982 for critical appraisals).

Our argument is further related to a venerable line of thinking that has linked modernization, and specifically increases in education, to political turmoil. For instance, Huntington (1968) has argued that higher education in many modernizing countries failed to provide skills that were relevant to the countries’ needs, leading instead to alienation and instability. We provide evidence that is suggestive of such a connection, while placing an added emphasis on the tradeoff between the use of human capital in production and time-intensive political activities. We also speak to the vast literature on the “modernization hypothesis”, and in particular, the link between education and democracy. This hypothesis can be traced back at least to the seminal contributions from Dewey, 1916, Lipset, 1959, and has been the subject of considerable empirical scrutiny, recent examples of which include Acemoglu et al., 2005, Acemoglu et al., 2008, Acemoglu et al., 2009, Bobba and Coviello, 2007, Glaeser et al., 2007, inter alia. While there is still an active debate here on whether education causes democracy, our results suggest that an alternative line of inquiry would be to examine under which conditions the effects from schooling to democracy are seen to be stronger.

We proceed as follows. Section 2 describes the data and the empirical strategy that we employ to study political protest at the individual level. Section 3 presents these results. In Section 4, we turn to the country-level evidence on incumbent turnover and democratization. Section 5 concludes. Details of our model of political participation and our data sources are contained in the Appendix to this paper. (A separate online appendix contains all the Appendix Tables referred to in the text.)