Political Budget Cycles in Autocracies: The Role of Religious Seasons and Political Collective Action

Abstract

For autocrats ruling over religious populations as in many Muslim majority countries, threats of mass collective action and religious mobilization are risky to ignore, being potentially detrimental to authoritarian survival. Religious seasons, such as Ramadan, could raise the seriousness of such threats. Accordingly, incumbents might adopt expansionary fiscal policies to avoid the escalation of political discontent at these times. Focusing on Egypt's fiscal policy between 2006 and 2019, I find that although the religious season of Ramadan is associated with modest increases in government expenditure and welfare spending, this relationship is dependent on the level of political threats facing the regime. Government spending is higher in Ramadan's season when it is preceded by more episodes of anti-regime collective action. This evidence suggests that the interaction between the religious and political contexts could generate political budget cycles outside electoral seasons.

Although several studies document the presence of electorally driven macroeconomic policy cycles in autocracies (e.g., Gonzalez 2002; Pepinsky 2007; Blaydes 2010), less attention has been paid to the role of non-electoral pressures in generating political budget cycles. In authoritarian regimes, threats of mass collective action could be more detrimental to autocratic survival than elections, incentivizing autocrats to increase government spending to contain public discontent when such threats are credible (Brownlee et al. 2013). And, for autocrats ruling over religious populations as in Muslim majority countries, religious mobilization could heighten the seriousness of such threats (Butt 2016). In these contexts, how do autocrats utilize fiscal policy to maintain their political survival?

To address this question, this article studies Egypt's fiscal policy during the period (2006–2019) which covers: Mubarak's regime (2006–2011), two transitional periods interrupted by the Muslim Brotherhood's rule (2011–2013), and al-Sisi's regime (2014–2019).¹ Given Egypt's political instability between 2011 and 2014, the analysis focuses primarily on fiscal policy-making under Mubarak's and al-Sisi's regimes. There is some evidence that both regimes increase their spending in electoral seasons to attract supporters (Blaydes 2010; Ketchley 2021), proposing that government expenditure follows the electoral calendar. While acknowledging that, I argue that electoral pressures are not the only source of political budget cycles under both regimes. Anecdotal evidence suggests that both regimes increase their spending and delivery of services and economic benefits in anticipation and during the religious season of Ramadan, the ninth month of the Islamic calendar. This is manifested in establishing temporary markets for government-subsidized goods and providing financial allowances for government employees. For example, in 1995, Mubarak's regime started an initiative to sponsor government-subsidized markets in Ramadan which lasted till 2010. A similar initiative was undertaken by al-Sisi's regime under the name of “Welcome Ramadan (Ahlan Ramadan)” markets. Besides suggesting that the religious calendar regulates government spending, this observation raises a question regarding the political rationale behind these religious fiscal policy cycles.

My argument is that autocrats expand their fiscal policy in the religious season of Ramadan to avoid the escalation of political threats during the season. The religious norms and rituals of Ramadan rationalize political discontent and facilitate collective action, raising the credibility of anti-regime collective action threats. In response, the government expands its spending, particularly on welfare channels, in Ramadan to contain such threats. My empirical findings support this argument. Although Ramadan's season has a modest effect on government expenditure, this effect is proportional to the level of collective action threats facing the regime. The religious season is associated with a significant increase in government expenditure when threats of anti-regime collective action and religious mobilization are higher.

The findings expand our understanding of the distinctive political rationales behind economic policy-making in authoritarian regimes, contributing to recent arguments linking autocrats' distributive policies to their political survival (e.g., Albertus et al. 2018). By departing from the predominant focus on elections in explaining economic policy cycles, this note underlines the role of the religious environment and collective action threats in generating alternative -yet overlooked- patterns of economic policy-making in Egypt and possibly other authoritarian regimes in the Muslim World.

Argument

Mass collective actions are hard to anticipate. Even regimes with expansive domestic surveillance apparatuses might fail to foresee the timing of mass collective actions and their consequences (Kuran 1991). The events of the Arab Spring took the world by surprise, while most observers were invested in analyzing the persistence of authoritarianism in the region (Bellin 2012; Shihade and Shihade 2012). More importantly, citizens' collective actions can impose high costs on autocrats. Escribà-Folch (2013) and Tanaka (2016) find that autocrats deposed by non-electoral means are more likely to face jail, murder, or exile after losing power. Even if the autocrat does not lose power, mass collective action could undermine the incumbent's image domestically, inspire further resistance, and signal discontent to the international community.

In Muslim majority countries, the threat of political collective action could rise during the religious season of Ramadan through two main channels. First, Ramadan's religious norms call for empathy with the poor and inter-class solidarity, requiring Muslims to support the less fortunate. As a result, religious and charity organizations intensify their campaigns during Ramadan to collect donations to fund their projects. Ramadan's charity campaigns promote altruistic behaviors and increase the supply of negative economic information by emphasizing social disparities to attract donors (Keenan and Yeni 2003; Atia 2013; Sherra 2017).² This could raise the salience of distributive issues, reveal government failures in addressing economic insecurities, and fuel citizens' discontent with the government's economic performance. For example, one day before Ramadan of 2010, Egyptian opposition groups organized an anti-regime protest criticizing Mubarak's government for failing to control food prices and leaving the poor more economically vulnerable in the month of charity (Saoud and Al-Khouli 2010).

Second, Ramadan's religious rituals might facilitate religious mobilization, adding to the potential threat of collective action during the season. Mass religious congregations are held daily during Ramadan with attendance rates exceeding any other time of the year. The popularity of these religious congregations weakens the organizational challenges to political mobilization, as is the case for Friday's congregations (Lynch 2013; Hoffman and Jamal 2014; Butt 2016). For example, the Jordanian anti-austerity protests that erupted in Ramadan of 2018 were facilitated by the evening mass congregations of the month, which offered a focal point for protesters to coordinate their movements.³ These religious congregations could also create an opportunity for religious opposition groups to recruit and mobilize new members. In fact, we see that incumbents take this threat seriously during Ramadan. In 2018, the head of Egypt's Parliamentary Committee on Religion stated, “it is a must that mosques are monitored by the Ministry of Religious Affairs during Ramadan to avoid their infiltration by extremist groups during Ramadan” (Al-Emari 2018).

This argument leads us to two main predictions. While raising the potential threat of political collective action by rationalizing and facilitating mobilization, Ramadan makes this threat more predictable. The incumbent can anticipate rising economic concerns by expanding its fiscal policy and increasing its expenditure, particularly on items strongly related to citizens' economic welfare, in Ramadan's season. Yet, given the economic costs of expansionary fiscal policies, we would expect the incumbent to resort to this strategy more when the threat of collective action against the regime is more credible. For example, if Ramadan is preceded by an episode of mass protests, the regime might worry that the threat might escalate during Ramadan and respond by expanding its economic provision. Therefore, Ramadan's season would be positively associated with increasing government expenditure (overall and toward welfare spending) when the threat of anti-regime collective action is higher, i.e., in more contentious times.

Data and Methodology

To test these theoretical predictions, I construct a monthly time series dataset of the Egyptian government's budgetary operations during the period (2006–2019). This dataset utilizes monthly statistical bulletins published by the Central Bank of Egypt, which detail the government's monthly expenditures and the breakdown of its spending on various items on monthly basis. The frequency and detail of these reports enable us to study within-year variations in government expenditure and characterize its use, which improves on the predominant focus on annual fluctuations in macroeconomic policies in the existing literature on political budget cycles (see: Brender and Drazen 2007; Dubois 2016).

The first outcome is the monthly change in total government expenditure (i.e., expenditure growth). I also consider the monthly change in government expenditure on two welfare items: salaries and the purchase of goods and services. Although welfare expenditure could be directed toward other channels, these two items are closely related to citizens' welfare (and welfare policies) in Ramadan. Egyptian governments often provide financial allowances to public employees as Ramadan approaches. They also expand their provision of free and subsidized goods before and during Ramadan, which would translate into a rise in spending toward the purchases of goods and services. Furthermore, unlike other outcomes, these two items experience no changes in their reporting and calculation over the period of the study.

The main independent variable is a dummy capturing whether a given month coincides with Ramadan or the 2 months before. In practice, the government's expansion of welfare policies takes place during Ramadan, as well as the month preceding it. For example, in 2018, the government claimed that one million households benefited from government distribution in Ramadan (Kandil 2018), indicating an increase in welfare provision in the season. Part of that distribution took place through the provision of subsidized goods sold in outlets managed by the government, which were launched a full week before the beginning of Ramadan (Wahba 2018). Hence, the government anticipates Ramadan with economic provision before the month itself. If so, then we expect the impact of these policies on the budget to extend beyond Ramadan and include the 2 months before it as well, when preparations for the season take place. This measurement strategy accounts for such a possibility while simplifying the analysis by utilizing a single measure for the independent variable of interest.

Ramadan's effects are estimated using first-order autoregressive OLS models. Besides controlling for the lagged dependent variable, the models control for the lagged monthly growth in total government revenues, the lagged monthly inflation rate, and an indicator for the first month of the fiscal year. Since earlier work documents electoral budget cycles in Egypt (Blaydes 2010), I add a dummy variable for whether a given month coincides with the electoral quarter (the month of elections and the two preceding months),⁴ to employ elections as a benchmark for Ramadan's effects.⁵ All models contain fixed effects for years, with robust standard errors clustered for years.

The second part of my argument suggests that government spending in Ramadan's season is driven by the regime's concerns over the escalation of political threats to its rule during that period. Thus, we expect Ramadan's effects to be stronger when public dissatisfaction with the regime is higher and political threats are more credible, as Ramadan can lead to further intensification of public discontent. To test this claim, I interact the indicator of Ramadan's season with three alternative measures of political discontent and threat (over the past quarter to any given month) obtained from the Armed Conflict Location and Event Data Project (ACLED) (Raleigh et al. 2010). The first is the number of violent collective actions (protests and riots) where violence occurred such as violent demonstrations, protests with interventions or excessive use of force against protesters, and mob violence. These actions might represent the highest threat to the regime as they involve clashes with security forces, signal citizens' willingness to take costly actions against the regime, and undermine political stability. The second measure includes only peaceful protests. The third variable is the number of protests and riots by Islamists. Under both Mubarak's (2006–2011) and al-Sisi's (2014–2019) regimes, political mobilization by Islamist groups constituted a key political threat. During the transitional period (2011–2013), various Islamist groups emerged. Competition among these groups presented challenges to incumbent governments, even when some Islamists factions (i.e., the Muslim Brotherhood and Salafis) had a grip on power for that brief period. Service provision was a primary strategy by which Islamist political groups recruited members and garnered political support, with Ramadan being an important season for these groups to demonstrate to the public their activism by providing food handouts and reaching out to new constituents for support (Masoud 2014). Hence, at times when the threat of Islamist opposition is high, we should expect incumbents to expand their role in service provision in Ramadan to crowd out their opponents.⁶

Throughout the analysis, I provide three sets of results: the full sample, Mubarak's regime, and al-Sisi's regime. The first provides an overall view of Ramadan's effects. During Mubarak's era, distribution in Ramadan was mostly targeted toward the government's employees who received allowances during the month. Mubarak's regime also provided subsidized goods during Ramadan, yet this strategy was far less pronounced than it has become under al-Sisi's regime. During the latter, the government increased its seasonal markets of government-subsidized goods to appeal to the public and contain discontent, due to the implementation of a set of austerity measures. Having said that, economic provision in Ramadan is financed differently under both regimes. While Mubarak's regime relied on budgetary resources, al-Sisi's governments utilized extra-budgetary resources to fund its distributive efforts in Ramadan. The most notable of these resources is Tahia Masr Fund, which is a government-managed fund that collects donations from the public to provide social support for the poor and finance infrastructure projects, yet it remains separate from the government's budget. This means that Ramadan's effects on the government's budget under al-Sisi's regime would only constitute a lower bound for the fiscal changes in the religious season.

Findings

Figure 1 presents the coefficients on the indicators of Ramadan and elections estimated from three different models with the outcomes: change in total expenditure, change in compensation for employees (salaries), and change in government purchases of goods and services. Although Ramadan's effect on total expenditure growth is not statistically significant (and imprecisely estimated), it remains substantively meaningful. Ramadan's quarter is associated with growth in overall expenditure by 14%. Ramadan's effect is stronger for Mubarak's regime with a growth rate of 28%, compared to only 14% during al-Sisi's time. Note that electoral seasons are associated with a decline in expenditure growth, particularly under al-Sisi's regime when electoral quarters are associated with a 40% drop in expenditure growth. And, under both regimes, the effect of Ramadan is significantly higher than that of elections (at the 90% level) as concluded from testing the equality of their coefficients.

Figure 1. Change in Government Expenditure in Ramadan

Note: All models contain year fixed effects and the controls specified in Section “DATA AND METHODOLOGY”. Standard errors are clustered for years. Confidence intervals are estimated at the 90 and 95% levels.

These same patterns persist when we look at spending on welfare items. Expenditure toward compensation of employees grows in Ramadan's quarter by 18%. This effect is statistically significant and distinguishable from the effect of elections. The magnitude of the effect is higher for Mubarak's regime (23%) compared to 15% under al-Sisi, but the latter coefficient is more precisely estimated. Similarly, Ramadan's quarter is associated with a significant rise (29%) in the growth of spending toward purchases of goods and services. Again, this change is higher for Mubarak's regime with an increase of 67% compared to 9% under al-Sisi. Meanwhile, electoral seasons have a negative association with growth in government purchases.

These results suggest that total government expenditure and welfare spending grow during the Ramadan season. Interestingly, these positive fluctuations contrast the patterns observed during electoral seasons, when the outcomes remain unchanged or decline. Breaking the analysis by political regimes reveals that Ramadan's effect is stronger under Mubarak, which might be attributed to the regime's reliance on budget resources for spending in Ramadan. The milder effects under al-Sisi could be related to the regime's utilization of extra-budgetary resources for welfare spending during the season.

The second part of my argument suggests that the fiscal expansion during Ramadan is driven by the regime's concerns over rising political threats during the season. Hence, we expect spending to increase in Ramadan's season when such threats are more credible, i.e., after periods of political contention. Table 1 presents the interaction effects of Ramadan with the three measures of political collective action on total expenditure. The results provide consistent support to my theoretical prediction. In panel (a), the interaction terms in models (1) and (3) are statistically significant. For a given month during Ramadan's season, total expenditure growth increases by only 1% in the absence of any violent political collective action. Yet, at the average level of violent contention, Ramadan's season would be associated with a 9.2% increase in expenditure growth. The government's fiscal expansion is also sensitive to Islamists' political activities: Ramadan's effects are almost twice as large at the mean level of the moderating variable (Islamist Action) in column (3). Note that none of the collective action measures is statistically significant and they all have a negative sign. Thus, government spending does not generally increase with collective action threats, but it does react to such pressures closer to Ramadan.

Table 1. OLS Analysis of the Relationship between Ramadan Quarter and the Change in Total Expenditure by Collective Action

Note: All models include year fixed effects. Robust standard errors in parentheses besides the coefficients are clustered for years. All models include all the controls specified in Section “DATA AND METHODOLOGY”, but only controls with theoretical relevance are displayed. ⁺p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

When we split the analysis by political regimes, we find consistent evidence that violent, peaceful, and Islamist collective actions are all significant moderators of Ramadan's effects on total expenditure growth. One key observation from panels (b) and (c) is that the sign on the coefficient on Ramadan is negative for most models. Under both regimes, fiscal expansion in Ramadan, thus, is not a mechanical reaction to predictable seasonal macro-economic changes in local markets during the season. And if such automatic policy response to Ramadan exists, it would be the opposite (contractionary fiscal policy) of what we observe. For example, in model (1) of panel (b), Ramadan's season would be associated with a 37% drop in expenditure growth absent any violent contention, which would turn into a 6% increase at the mean level of violent contention during Mubarak's time. A similar story emerges under al-Sisi's regime. For example, in model (3), Ramadan would bring an 11% increase in the outcome when preceded by the average level of Islamist collective action, despite being negatively correlated with expenditure growth in the absence of Islamist mobilization. Furthermore, while the two regimes react to collective action threats outside Ramadan's season differently, the coefficients on the three measures of collective action differ from their interaction terms either in magnitude (for Mubarak's regime), or direction (for al-Sisi's regime). This underlines the role of Ramadan in translating these political pressures into fiscal responsiveness. Mubarak's regime reacts to collective action threats by increasing expenditure, despite modestly. Its reaction, however, is amplified in Ramadan's season. Al-Sisi's regime responds to collective action by constraining expenditure, but this strategy is reversed in Ramadan's season.

To further illustrate these patterns for the various samples, Figure 2 plots the marginal effects of Ramadan on the change in expenditure over the range of the moderating variables. The figures illustrate that government expenditure growth increases significantly [with the exception of figure (e)] in Ramadan's quarter at higher levels of political contention. Therefore, the growth in government expenditure during the religious season is rooted in the regime's political concerns over its popularity and the political threats of contention it faces. More importantly, it is the combination of Ramadan's season and collective action threats (rather than the sole effect of either) that explains government spending in Ramadan.

Figure 2. The Marginal Effects of Ramadan Quarter on the Change in Government Expenditure by Collective Action

Note: Confidence intervals are estimated at the 95% level.

In the Supplementary Material, I present a similar analysis of Ramadan's effects by collective action on welfare spending growth (Appendix B). The results show similar patterns: Ramadan's quarter is associated with growth in welfare expenditure (salaries and purchases of goods and services) in more contentious times. This finding confirms that our results are -at least- partly driven by increasing welfare spending in Ramadan in response to collective action threats. These findings are also robust to using a 2-month window for measuring collective action threats (Appendix C) and the exclusion of the fixed effects (Appendix D).

Note that these findings do not imply that the threat of contention is the sole motivation for government expenditure in Ramadan. The government might provide in Ramadan to assert its religious legitimacy and enhance its popularity, or even because government officials are themselves religious. Yet, for these explanations to be sufficient in explaining government provision in Ramadan, we should expect that government welfare spending increases in Ramadan even in the absence of political contention. Empirically, this is not the case as indicated in Table 1. Ramadan has an insignificant or negative association with the outcomes of interest, absent any contention, highlighting that the shifts in government expenditure around Ramadan are driven by the threats heightened during the season.

Conclusion

Mass collective action and religious mobilization could jeopardize the political survival of autocrats in religious societies. Religious seasons, such as Ramadan in Muslim majority countries, could offer an opportunity for the materialization of these potential threats by increasing the political salience of economic issues and facilitating collective action. In line with that, I find that Egyptian governments are more likely to anticipate the religious season of Ramadan by adopting expansionary fiscal policies and increasing welfare spending as the level of anti-regime collective action threats increases.

In a relevant contribution, Klomp and de Haan (2013) find that anti-government protests in electoral seasons result in stronger manipulation of fiscal policy, especially in young democracies. My results confirm the importance of protests in explaining fiscal policy-making in non-consolidated democracies and authoritarian regimes. Yet my explanation brings into attention the overlooked role of the religious environment in translating protests into government spending decisions, departing from the predominant focus on electoral pressures (which could be less detrimental in autocracies) as a driver of expansionary fiscal policy cycles.

Furthermore, while some scholars report lower welfare spending in more religious societies (e.g., Scheve and Stasavage 2006), my findings reveal the complexity of the relationship between religion and welfare expenditure. First, religious factors might not only shape the level of welfare spending, but also its distribution over time. Second, the government's fiscal reactions to the religious environment might be conditional on the political context. The role of Ramadan's season in explaining fluctuations in government spending and the dependency of this role on political collective action attest to this convoluted association between religion and fiscal policies. This conditional nature of the relationship between the religious context and economic policy-making merits further investigation.

Finally, there remains the question of whether Ramadan is associated with political budget cycles in other authoritarian contexts and non-consolidated democracies with Muslim majority populations. Outside the Muslim World, religious seasons creating favorable conditions for popular discontent and collective action could also incentivize incumbents to increase government spending, i.e., Christmas in religious Christian societies. By taking non-electoral threats and the role of the religious and cultural contexts more seriously, future work could uncover more politically motivated, but overlooked, regularities in economic policy-making.