1 Introduction
Does foreign income in the form of aid and direct investment influence power accumulation by dictators? Official Development Assistance (ODA) and Foreign Direct Investment (FDI) are among the main sources of external finance of countries in the Global South, many of which are still governed by nondemocratic regimes. Driven by multilateral development goals and economic liberalization, these cross-border flows have surged in recent decades. According to UNCTAD (2023), developing countries received over 900 billion dollars in FDI and about 200 billion in foreign aid in 2022. Both income flows entail a transnational transfer of capital and funds. Yet, these (and other) cross-border flows differ from each other in important ways. In particular, they do so not only in size (and volatility) but also, importantly, in who the senders and the recipients are. ODA is distributed by official government agencies in high-income countries and mostly flows to governments in low-income countries. In contrast, FDI decisions are made by profit-seeking multinational corporations (MNCs), with capital accruing to private businesses and partly to governments and state-owned companies.
The impact of external finance on regime durability and the quality of governance in recipient countries has received ample scholarly attention. For example, oil and gas rents have been repeatedly found to undermine democracy and to extend authoritarian regime duration (e.g. Andersen and Ross, Reference Andersen and Ross2014; Wright et al., Reference Wright, Frantz and Geddes2015). For some of these foreign income inflows, however, the existing evidence is less clear and even contradictory. For example, while FDI has recently been found to help autocrats survive longer in power (Bak and Moon, Reference Bak and Moon2016; Powell and Chacha, Reference Powell and Chacha2016; Escribà-Folch, Reference Escribà-Folch2017); other research reports a positive relationship between foreign investment and human rights and democracy levels (Richards et al., Reference Richards, Gelleny and Sacko2002; Li and Reuveny, Reference Li and Reuveny2003; Rudra, Reference Rudra2005; Eichengreen and Leblang, Reference Eichengreen and Leblang2008). Similarly, while foreign aid has been typically found to entrench nondemocratic rule as well as decrease institutional quality and democraticness (e.g. Knack, Reference Knack2004; Djankov et al., Reference Djankov, Montalvo and Reynal-Querol2008; Kono and Montinola, Reference Kono and Montinola2009; Licht, Reference Licht2010; Ahmed, Reference Ahmed2012); for the post-Cold War period, scholars have related aid to the adoption of multiparty elections and democratization (Dunning, Reference Dunning2004; Dietrich and Wright, Reference Dietrich and Wright2015; Escribà-Folch and Wright, Reference Escribà-Folch and Wright2015; Bermeo, Reference Bermeo2016). Finally, while some research suggests that migrants’ remittances foster democracy (Bearce and Park, Reference Bearce and Park2019; Escribà-Folch et al., Reference Escribà-Folch, Meseguer and Wright2022); other scholars claim that they stabilize autocratic governments and worsen governance (Abdih et al., Reference Abdih, Chami, Dagher and Montiel2012; Ahmed, Reference Ahmed2012, Reference Ahmed2013).
We leverage these differences to examine (and compare) the relationship between these two forms of foreign income and the degree of personalism in dictatorships, that is, the extent to which power is concentrated in the hands of the leader. Personalism is conceptually different from leader and regime duration, which have been the main focus of past research. The impact of external finance on intra-regime dynamics, particularly power distribution within dictatorships, remains under-explored; existing research focuses primarily on oil (Fails Reference Fails2020). This article addresses a key gap in the comparative authoritarianism literature by examining the relationship between foreign aid, FDI, and personalism in dictatorships.
Personalization refers to the gradual accumulation of power by the leader (Geddes et al., Reference Geddes, Wright and Frantz2018). When dictators manage to usurp control of decision-making and political appointments at the expense of their top allies, they come to dominate the entire state apparatus and can exercise power with little or no restraint. Personalism is one of the outcomes that can emerge from the (often conflicting) interactions between dictators and their support coalition (Svolik, Reference Svolik2012; Geddes et al., Reference Geddes, Wright and Frantz2018). When elites cannot credibly threaten the rulers, they will concentrate enough power to reign unconstrained. In other contexts, dictators establish power-sharing agreements that give elites access to rents and influence—often enforced via formal political institutions (Svolik, Reference Svolik2012; Meng, Reference Meng2020; Meng et al., Reference Meng, Paine and Powell2023).
We present several arguments linking foreign aid and FDI to personalism levels. Particularly, we posit that aid should be related to higher personalism levels since it is a windfall that accrues to the government and does not require cooperation from elites to obtain it. As a result, it allows autocrats to distribute patronage rents and create dependency ties with elite members. Furthermore, aid signals foreign support for the regime, strengthening the dictator’s position vis-à-vis his ruling coalition. In contrast, while FDI can also generate rents for the incumbent government, its more decentralized nature and the role of private senders push the regime to reshape its internal organization in ways that constrain the ruler’s discretion. This is so because constraints reduce risks for foreign investors and FDI inflows boost mutual dependence and reshape elites’ incentives and influence as they may acquire new preferences, connections, and exit options. To test our hypotheses on a global sample of dictatorships, we analyze foreign aid data from 1960 to 2010, FDI data from 1970 to 2010, and a latent index of personalism in dictatorships (Geddes et al., Reference Geddes, Wright and Frantz2018; Wright, Reference Wright2021). We find empirical evidence that ODA is not correlated with personalism, while FDI is negatively and significantly correlated with it, but with a small substantive effect. Additionally, after disaggregating the outcome variable, we find that aid increases the likelihood that the ruler will create a paramilitary unit for regime protection, while reduces the probability of new party creation. FDI’s relation with personalism runs mostly through the security items; in particular, it significantly reduces the likelihood of the dictator controlling military promotions and using purges.
Since the end of World War II, and especially, after the end of the Cold War, personalism levels have risen almost steadily, and so have aid and FDI flows. This article contributes to our understanding of the conditions behind the emergence of personalistic dictatorships. Most existing work has concentrated on domestic determinants, such as the initial balance of power between the ruler and the ruling coalition and failed coups (Geddes et al., Reference Geddes, Wright and Frantz2018; Meng, Reference Meng2020; Timoneda et al., Reference Timoneda, Escribà-Folch and Chin2023). Our findings shed new light on the debate about whether foreign income is a curse or a blessing for recipient countries. We do so by analyzing a different (but highly consequential) dimension of autocratic rule, personalism, which foreign actors—i.e. governments and MNCs—might have a role in shaping via aid and investment.
2 Extant work: foreign income and autocratic rule
While some dictators manage to accumulate power within their hands, others share it with elite members in predictable ways. However, despite the growing global importance of personalist rule and its (many) negative consequences, the literature on the emergence of personalism is still rather limited. Extant studies suggest that power accumulation is mainly made possible by factors that give the dictator a bargaining advantage over elites so that the latter are less able to monitor or punish him via a coup. Conversely, factors that enhance elites’ capacity to coalesce and credibly threaten the dictator push regimes towards the establishment of power-sharing agreements (Svolik, Reference Svolik2012).
Most of such factors shaping dictator or elite advantages identified in the literature are domestic. Geddes et al. (Reference Geddes, Wright and Frantz2018) and Meng (Reference Meng2020), for example, both focus on the initial balance of power as shaped by internal cohesion of seizure groups and the strength of the leader—as determined by characteristics such as rulers’ mode of entry—upon taking over, respectively. Relatedly, contextual events, like failed coups, can also give dictators temporary informational and power opportunities to accumulate power (Timoneda et al., Reference Timoneda, Escribà-Folch and Chin2023).
International factors can also give dictators a military or a resource advantage vis-à-vis the ruling coalition. Concerning military ones, some works find that external support in the form of sponsorship or defense alliances makes the adoption of certain coup-proofing practices more likely (Boutton, Reference Boutton2019; Casey, Reference Casey2020). Resource advantages, on the other hand, can come from transnational income flows. Comparativists have long studied how external finance influences several political outcomes. However, the only work connecting foreign income to power concentration in dictatorships is that of Fails (Reference Fails2020), which examines the impact of oil. He finds that oil rents are related to higher personalism levels since generous unearned resources allow rulers to finance patronage networks, reduce the need to share power, and resist foreign pressure to liberalize.
The political effects of aid have been mostly evaluated in terms of regime duration and institutional quality, not personalism.Footnote 1 An initial set of works reported an unconditional negative effect of ODA on democracy and governance and a positive one on autocratic regime duration (e.g. Knack, Reference Knack2001, Reference Knack2004; Bräutigam and Knack, Reference Bräutigam and Knack2004; Rajan and Subramanian, Reference Rajan and Subramanian2007; Djankov et al., Reference Djankov, Montalvo and Reynal-Querol2008; Kono and Montinola, Reference Kono and Montinola2009; Licht, Reference Licht2010; Ahmed, Reference Ahmed2012). Theoretical frameworks often equate aid to oil, viewing both as unearned windfalls that largely benefit governments. Like oil rents, foreign aid generates substantial nontax revenue, enabling regimes to forgo taxation, reward political support, and strengthen repressive forces. However, recent evidence suggests aid’s impact has evolved. Unlike oil, aid is distributed by government agencies whose political priorities shift over time. As Bermeo (Reference Bermeo2016, 6) clearly puts it, “The end of the Cold War brought not only the demise of strategic importance for many developing countries but an increased emphasis on democratization for many democratic aid donors.” Consistent with this, ample evidence shows that aid had a stabilizing effect during the Cold War, but it has fostered democratic change in the post-Cold War period (Dunning, Reference Dunning2004; Bermeo, Reference Bermeo2011, Reference Bermeo2016; Dietrich and Wright, Reference Dietrich and Wright2015; Escribà-Folch and Wright, Reference Escribà-Folch and Wright2015).
In contrast to aid and oil, other forms of external finance are more decentralized, as they target a broader range of actors and are less susceptible to government capture or diversion. Surprisingly, however, despite their volume being significantly larger than that of ODA, the political impact of private funds such as FDI (and remittances) has remained mostly unexplored until recently. Furthermore, studies examining FDI’s political effects yield seemingly contradictory results. On the one hand, a first set of works suggests that international investment fosters political liberalization and respect for human rights (Apodaca, Reference Apodaca2002; Richards et al., Reference Richards, Gelleny and Sacko2002; Li and Reuveny, Reference Li and Reuveny2003; Rudra, Reference Rudra2005; Eichengreen and Leblang, Reference Eichengreen and Leblang2008). It is argued that FDI can have a modernization effect and thus indirectly stimulate democratization by promoting long-term development. More specifically, these authors posit that private capital flows undermine dictators’ control over the economy, reduce the dependence of firms on state resources, reshape the preferences of some local groups, and can strengthen proliberalization actors (Maxfield, Reference Maxfield1998; Malesky, Reference Malesky2008; Li and Reuveny, Reference Li and Reuveny2009; Levitsky and Way, Reference Levitsky and Way2010).
On the other hand, researchers focusing on autocratic stability have found that FDI reduces the likelihood of regime/leader failure and coups. Several arguments inform this finding. First, FDI increases the costs associated with irregular political change (Powell and Chacha, Reference Powell and Chacha2016). Second, FDI can generate patronage rents that dictators can use to buy off political support (Bak and Moon, Reference Bak and Moon2016; Escribà-Folch, Reference Escribà-Folch2017; Tomashevskiy, Reference Tomashevskiy2017). Finally, FDI helps mitigate commitment problems by making defection more costly for both the ruler and elite members (Bak and Moon, Reference Bak and Moon2016).
The extant research presents several challenges. First, the focus on regime durability obscures the distinct political dynamics that may underlie regime (in)stability. Both power-sharing and personalism can lead to greater regime or leader stability and lower coup risk but via distinct mechanisms. While the former reduces incentives by providing access to power, the latter mostly operates by undermining the elites’ ability for coordinated action (Chin et al., Reference Chin, Escribà-Folch, Song and Wright2022). Second, the current state of the literature provides a limited understanding of how foreign income can affect these intraregime dynamics. The mechanisms existing studies underscore seem to point to contradictory effects that can lead regimes to different equilibria. Rent-sharing, for example, might facilitate power concentration or might be accompanied by a reallocation of power (Meng et al., Reference Meng, Paine and Powell2023).
3 The arguments: external finance and personalism in dictatorships
We argue that aid can help dictators concentrate power in their hands by providing extra (discretionary) resources that can be used to buy off elite support and create dependency ties. On the contrary, we contend that FDI, by increasing mutual dependency, pushes rulers to share power and, consequently, reduces the degree of personalism. These expectations stem from the characteristics of each of these external flows and from the way they may reshape intraregime interactions between the dictator and his support coalition.
We assume the dictator and his inner circle are rational and survival-driven actors, who, as Geddes et al. (Reference Geddes, Wright and Frantz2018, 67) underscore, “engage simultaneously in two kinds of strategic interaction: (1) a cooperative effort aimed at keeping all of them (the regime) in power and (2) noncooperative interactions in which different members/factions seek to enhance their own power and resources at the expense of others in the inner circle.” As said, this struggle over the distribution of power may result in two distinct trajectories: one where the dictator concentrates power, and another where the ruler shares rents and power with elites. The next subsections explore how aid and FDI might affect such outcomes.
3.1 Foreign aid and personalism
Foreign aid may transform intra-regime dynamics in a way that spurs power accumulation. A first mechanism underpinning such expectation has to do with the fact that aid is a form of non-tax revenue and, hence, unearned foreign income (Smith, Reference Smith2008). Aid thus generates additional economic resources for governments, which have ample discretion over their allocation. This is in large part due to the fungibility of aid, that is, the fact that aid has been mostly delivered in the form of budget support, or to fund specific projects, which in turn allowed recipient governments to free-up funds that could be diverted to fund patronage or strengthen the security apparatus (Feyzioglu et al., Reference Feyzioglu, Swaroop and Zhu1998; Remmer, Reference Remmer2004).Footnote 2 Consequently, foreign aid enhances the dictator’s resource advantage over their inner circle, strengthening their position and enabling them to diminish the influence of allies, thereby accumulating personal power.
This advantage arises from two factors. First, being unearned income (similar to oil rents), aid revenues do not require domestic cooperation or activity for generation and collection. Second, these funds typically accrue to the government, giving the dictator significant discretion in their allocation for political gain. Aid receipts enable the ruler to distribute patronage in a highly centralized manner, fostering elite dependency and internal competition, which weakens incentives and capacity for defection or rebellion. Increased elite dependency on the ruler is thus the key mechanism linking foreign aid to personalism. The dictator uses rent distribution to reward loyalists and neutralize potential challengers, making defection less appealing compared to remaining within the ruling coalition. Furthermore, centralized and discretionary rent distribution fosters elite competition for access to state resources, enabling rulers to employ divide-and-rule tactics and pit factions against each other. This weakens elites’ ability to coordinate in efforts to remove the ruler and lowers the cost of securing their loyalty.
A second mechanism linking aid to higher personalism concerns aid’s potential to enhance the regime’s coercive capacity. This is a mechanism similarly connecting oil rents to increased autocratic resilience, as noted by several scholars (e.g. Ross, Reference Ross2001; Wright et al., Reference Wright, Frantz and Geddes2015). Increased aid resources and the ability to divert funds enable governments to boost security spending, as reported in several studies (Collier and Hoeffler, Reference Collier and Hoeffler2007; Kono and Montinola, Reference Kono and Montinola2013; Langlotz and Potrafke, Reference Langlotz and Potrafke2019). This, in turn, enhances the dictator’s control over the security apparatus by allocating budgets to reward loyalty and increase dependency, or by funding parallel paramilitary units for regime protection (Escribà-Folch et al., Reference Escribà-Folch, Böhmelt and Pilster2020). In addition, generous aid disbursements also signal to both internal and external foes the support of foreign countries to the incumbent leader.
Based on these arguments, our first hypothesis is as follows: foreign aid is associated with higher personalism levels in dictatorships.
As emphasized earlier, the end of the Cold War opened the door for (more credible) democratizing pressures linked to aid disbursements resulting from a shift in donors’ strategic intent. This resulted in a positive impact of aid on governance and democratization as numerous cross-national studies report (e.g. Dunning, Reference Dunning2004; Escribà-Folch and Wright, Reference Escribà-Folch and Wright2015; Bermeo, Reference Bermeo2016). This would initially suggest that conditionality should be related to lower personalism levels, as dictators were coerced into sharing power and institutionalizing their regimes by adopting multiparty systems and even elections. However, as Dietrich and Wright (Reference Dietrich and Wright2015) note, many such reforms fell short of full democratization. Moreover, most aid-supported investments in governance and state capacity continued to target governments and, in turn, ended up bolstering incumbent regimes. Consequently, the potential effect of aid on personalism in the post-Cold War period is unclear and we explore it in the empirical section below.
3.2 FDI and personalism
Recent comparative work has found that foreign investment positively affects autocratic stability (Bak and Moon, Reference Bak and Moon2016; Powell and Chacha, Reference Powell and Chacha2016; Escribà-Folch, Reference Escribà-Folch2017; Tomashevskiy, Reference Tomashevskiy2017). However, as emphasized earlier, this outcome may arise either from dictators eliminating all threats to their rule or from the elites’ strengthened institutional capacity to uphold power-sharing agreements.
One line of reasoning would suggest that, like other windfalls, FDI also generates rents that the ruler can partially capture, allowing him to increase patronage distribution to reward loyalty, and induce competition between elites for access to such benefits. The resulting resource advantage for dictators could in turn open opportunities for power concentration. To extract such rents, autocratic governments can, for example, charge entry licenses, get kickbacks, collect taxes on new firms and their benefits, impose the obligation to create joint ventures with local partners, sell public assets, or divert funds to state-owned companies (e.g. Malesky et al., Reference Malesky, Gueorguiev and Jensen2015; Bak and Moon, Reference Bak and Moon2016; Escribà-Folch, Reference Escribà-Folch2017). The dictator’s ability to intermediate and distribute business opportunities to handpicked local partners could boost the leader’s power relative to elites and make the latter more dependent on the former. Additionally, as Wright and Zhu (Reference Wright and Zhu2018) posit, some MNCs in the primary sector may invest in more autocratic environments to obtain monopoly rents. Since a part of these rents accrues to personalistic rulers, they can then use them to further consolidate their position.
However, despite these potential opportunities for rent distribution, we expect FDI to lead to lower levels of personalism. Such expectation hinges on several mechanisms related to the interests of MNCs and to changes in the internal balance of power within dictatorships that FDI may bring about.
First, FDI requires the cooperation of domestic and foreign actors to generate economic (and political) gains. Concerning foreign actors, entailing the creation or acquisition of a production facility in another country, MNCs making FDI decisions are mainly concerned about expected profits and risks—such as that of expropriation. Indeed, a large literature examining the determinants of FDI underscores that the institutional environment of the host economy is crucial in shaping long-term investment decisions in foreign countries. Accordingly, democracies have been typically found to receive more FDI inflows than autocracies due to their more credible ability to commit to protecting property rights and maintain a stable policy environment (e.g. Ahlquist, Reference Ahlquist2006; Jensen, Reference Jensen2008; Li, Reference Li2009).Footnote 3 A similar logic underpins the variation in the volume of FDI inflows across autocracies. Moon (Reference Moon2015, Reference Moon2019) shows that autocracies with long-time horizons and better property rights protection, as well as those with formally democratic institutions, attract more FDI than other nondemocracies.Footnote 4
Against this backdrop, increased FDI enhances the bargaining power of MNCs relative to the host government, which MNCs can leverage to encourage reforms. When MNCs invest in a dictatorship, some rents may become available, but the flow of private funds depends on the MNCs’ evaluation of the institutional environment and the perceived risk of expropriation. Consequently, placing constraints on the discretionary power of the leader (i.e., reducing personalism) should make the country more attractive to foreign investors, giving dictators incentives to institutionalize power-sharing arrangements.
Secondly, and regarding domestic actors, FDI creates a window of opportunity for elite members to coordinate and enforce mutual dependency vis-à-vis the dictator. As Levitsky and Way (Reference Levitsky and Way2010) claim, linkages to the West such as foreign investment can reshape the interests of some domestic actors and increase their resources. On the one hand, involvement in foreign investment ventures can prompt a change in preferences of elites with business interests in certain sectors toward more liberalized or constrained forms of governance as well as the emergence of new (regional or sectoral) elites with ties to international companies, networks, and markets. Consequently, “Linkage may alter the balance of power within autocratic parties, helping to strengthen reformist tendencies” (Levitsky and Way, Reference Levitsky and Way2010, 49). Ascending and old elites’ interest in getting access to the benefits associated with FDI projects creates strong incentives for them to coalesce and keep (or demand) the institutional conditions that ensure the continuity and long-term profitability of such investments.
On the other hand, the more decentralized nature of FDI inflows prompts the emergence of elites with connections with transnational economic actors and increased economic resources, which reshapes the internal distribution of power in favor of the ruling coalition. Transnational connections provide military, regional, and sectoral elites controlling businesses with informational advantages as well as exit options that can facilitate defection from the ruling coalition. This boosts their bargaining power, their political influence, and their ability to enforce power-sharing agreements. While dictators share opportunities for rent-seeking with elites and may selectively grant access to joint projects, once in place, the long-term cooperation and autonomy of those insiders participating in FDI projects become necessary for mutual gains to be generated and for rents to flow back to the regime (Bak and Moon, Reference Bak and Moon2016). Such cooperation requires information flows and credibly delegating authority to elites, which are both facilitated by the existence of institutionalized power-sharing mechanisms (Gehlbach and Keefer, Reference Gehlbach and Keefer2012).
Based on these arguments, our second hypothesis contends that: FDI decreases personalism in dictatorships.
4 Data and research design
Our main dependent variable is the latent personalism measure developed by Geddes et al. (Reference Geddes, Wright and Frantz2018) and Wright (Reference Wright2021). The index is a time-varying, continuous variable that ranges between 0 and 1, where 0 represents minimum personalism and 1 maximum levels of power concentration. The measure results from computing an item response theory (IRT) two-parameter logistic model (2PL) model using eight constitutive items. These dichotomous items are coded as 1 when dictators use strategies aimed at controlling the security apparatus and the political party (if one exists). Particularly, the military indicators are (1) security apparatus, which describes if dictators personally control the security forces; (2) creation of paramilitary, which is coded as 1 when dictators have created a loyal paramilitary force outside of the normal military chain of command; (3) promotions, which entail that leaders have the power to personally promote loyalists to high military ranks, and finally, (4) purges, which indicates whether the leader imprisons/kills officers from groups other than his own without a reasonably fair trial. The civilian party indicators are (5) new party, which entails the creation of a new political party; (6) appointments to high office, which evaluates the discretion of dictators to control appointments to high office; (7) party rubber stamp, which identify how dictators use the party executive committee to pursue personal policies and choices, and finally, (8) party executive committee, which describe if dictators have the power to choose party leaders (Geddes et al., Reference Geddes, Wright and Frantz2018, 79–82). This measure captures a variation that traditional typologies cannot due to their time-invariant nature, which codes the entire autocratic spell as one ideal regime type. In contrast, personalism varies across units and over time.
Our two key independent variables are foreign aid and foreign direct investment. We measure foreign aid as net ODA per capita in constant 2020 USD. Specifically, ODA “consists of disbursements of loans made on concessional terms [..] and grants by official agencies of the members of the Development Assistance Committee (DAC), by multilateral institutions, and by non-DAC countries to promote economic development and welfare in countries and territories in the DAC list of ODA recipients” (World Bank, 2023). FDI includes “the net inflows of investment to acquire a lasting management interest (10 percent or more of voting stock) in an enterprise operating in an economy other than that of the investor” (World Bank, 2023). We use net FDI inflows per capita in constant USD.Footnote 5 Negative values are transformed into 0 for both independent variables and we apply the natural logarithm to normalize the distribution so that outliers do not drive the results. Both variables are lagged one year to avoid reverse causality.
We add several controls that could confound the relationship between external finance and personalism. Logged GDP per capita, economic growth of income per capita, Logged population (World Bank, 2023) capture economic performance and country size: lower values of income, economic growth, and population are related to higher personalism (Wright, Reference Wright2008), and might influence the receipt of aid and FDI. We also include logged oil rents per capita (Ross and Mahdavi, Reference Ross and Mahdavi2015), which has been found to increase personalism (Fails, Reference Fails2020) and might be related to FDI by increasing resource rents (Escribà-Folch, Reference Escribà-Folch2017). Finally, we include civil war (Uppsala Conflict Data Program, UCDP, 2022), which might increase aid and discourage foreign investment and logged leader duration, which is correlated with higher levels of personalism (Geddes et al., Reference Geddes, Wright and Frantz2018). All control variables are lagged one year.
We employ two-way fixed effect (TWFE) regressions to evaluate empirically the effect of aid and FDI on the level of personalism in dictatorships. The inclusion of leader FE helps controlling for unobserved unit- and time-invariant characteristics of leaders, whereas the year FE account for time-specific shocks or trends. In addition, for the case of aid, we also run separate models for the Cold War and post-Cold War periods to explore if its potential association with personalism changes due to increased donor conditionality in the latter period. The analyses below thus focus on the within-leader relationship between a change in the level of foreign income and the subsequent change in the level of personalism. Standard errors are clustered at the leader level.Footnote 6
5 Results
Table 1 presents the results testing the impact of aid on personalism. Models 1–2 show the results for the whole sample period (1960–2010), while Models 3–4 and Models 5–6 show the results of the split samples analysis for the Cold War and post-Cold War periods, respectively. Additionally, for each period, we report naive models without controls (columns 1, 3, and 5), and models adding all confounders (2, 4, and 6). Interestingly, and contrary to our expectations, aid is not significantly correlated with personalism. The estimated coefficients are positive in the naive models, turn negative once the controls are added, and are not significant in any case.
Table 1. Foreign aid per capita and personalism, 1960–2010
Note: Clustered (Leader) standard errors in parentheses.
*** p< 0.01, ** p < 0.05, * p < 0.1.
To further check the robustness of the findings, we conduct other models that we report in the Online Appendix. Table A2 shows models that include the negative values of ODA. The coefficients remain negative and insignificant. Table A3 reports the results of models where time is modeled by adding time trend polynomials rather than year fixed effects, and models where we exclude the year FE. These alternative specifications yield positive coefficients, but still insignificant. Tables A4 and A5 show the results of ODA per capita in current USD, excluding and including negative values, respectively. The estimates are again nonsignificant. Table A6 shows the results of ODA per capita using data from the AidData Project instead of the World Bank’s (Tierney et al., Reference Tierney, Nielson, Hawkins, Roberts, Findley, Powers, Parks, Wilson and Hicks2011).Footnote 7 The coefficients are stronger and positive across all specifications but, again, are not different from zero. Table A7 reports estimates using two alternative independent variables: economic and democracy aid, both from the AidData Project. Overall, aid data conflate both modalities of funding which could confound our results, so we follow Dietrich and Wright (Reference Dietrich and Wright2015) and distinguish between economic and democracy aid and re-run our main models for both the Cold War and post-Cold War periods. Interestingly, and consistent with our first hypothesis, we find economic aid to have a positive and significant correlation with personalism for the Cold War period, suggesting that these fungible funds facilitated dictators’ concentration of power. The coefficient is 0.012. In substantive terms, however, this entails that an increase in one (overall) standard deviation (s.d. = 1.28) results in a modest 0.015 unit change in the personalism index.Footnote 8 Democracy aid is not significant in any specification.
Finally, we check the influence of individual units (leaders) on the main findings and report Figures in the Online Appendix showing the results of leave-one-out tests to examine the extent to which the estimates change after excluding one leader at a time. The distribution of the coefficients reveals that the main reported result is overly dependent on data from one case (see Figure A3). Curiously, removing Zaire under Mobutu makes the aid coefficient turn positive (0.004). Some other cases have some influence, but it is not notable, as Figure A4 shows. None of the coefficients is significant though.
Table 2 reports the models testing the relation between FDI and personalism. Model 7 shows the results of the naive model, while Model 8 includes the full set of controls. The estimates in Table 2 show that FDI is significantly and negatively correlated to personalism. Albeit significant, the estimated effect is not robust (see below) and is substantively small: an increase in one (overall) standard deviation in FDI (s.d. = 3.98) leads to a −0.012 unit change in personalism levels. Note, however, that our TWFE models focus on within-unit variation, which is also relatively small, as personalism does not vary much on average during dictators’ tenure.Footnote 9 The effect size relative to the within variation still represents a relatively small fraction of the outcome variable variability.
Table 2. FDI per capita and personalism, 1970–2010
Note: Clustered (Leader) standard errors in parentheses.
*** p < 0.01, ** p < 0.05, * p < 0.1.
Again, to check the robustness of these results, we conducted additional tests reported in the Online Appendix. Table A10 shows the results of FDI per capita including the negative values that we transformed into 0 in our main analysis. Table A11 shows the results when time is modeled adding time polynomials instead of year FE, and when year FE are excluded. The results are consistent in both cases. Table A12 reports the results of FDI per capita in current USD, both excluding and including negative values. The coefficients remain negative but they are insignificant in this case. Table A13 shows the results of FDI per capita normalized using the cube root transformation, which allows to account for negative values. This transformation yields negative and significant results.Footnote 10 Table A14 shows the results of FDI as a percentage of GDP normalized using the cube root transformation, following Wright and Zhu (Reference Wright and Zhu2018). The coefficients are negative but not significant. Finally, in Figures A5 and A6, we explore the impact of potential outliers by conducting leave one out tests. Interestingly, as the distribution of coefficients after leaving one leader out at a time shows (see Figure A5), the results are heavily dependent on and sensitive to the exclusion of some specific cases. Specifically, as Figure A6 shows, if the case of Mobutu (Zaire)—with abnormal levels and changes in FDI after correcting for inflation—is excluded the coefficient turns much smaller (−0.0007) and becomes non-significant; while if Eyadema (Togo) is excluded, the coefficient becomes stronger (−0.0036) and significant at 0.01. This suggests that our main results should be interpreted with caution due to the potential fragility of the analysis. Overall, these additional tests show that the results for FDI are not totally robust to alternative measurements of the FDI variable and to the exclusion of influential cases.
6 Additional results: disaggregating the dependent variable
The findings above suggest that these forms of foreign funding have no clear effect on personalism. This evidence, however, cannot inform us about which item or dimension of personalism is most affected (if any) by aid and FDI. The null or small estimated effects on the overall personalism index might be the outcome of aid and FDI having opposite impacts on some of the eight constituent items of that measure, so that they cancel each other out. Indeed, in the theoretical section, we outlined two competing arguments linking FDI to power accumulation, which may be offsetting each other. To shed some light into this, we first conduct several Linear Probability Models (LPM) to explore which of the eight discrete actions dictators might adopt is more or less likely to be observed as a result of receiving more aid and FDI. Secondly, we perform a series of tests using as DVs two separate latent measures of party and security personalism. To construct the security personalization index, Chin et al. (Reference Chin, Song and Wright2023) compute a Bayesian IRT model where they include the four security items and a fifth item. This last item is created from the original Appointments to high office, but Chin et al. (Reference Chin, Song and Wright2023, 33) consider that it includes appointments to both the military and civilian high offices. The party personalization index is “analogously constructed as security personalization”.
Table 3 shows that aid is negatively and significantly related to the existence of a new party, and positively and significantly related to the existence of paramilitary units. A one standard deviation increase in the aid variable results in an increase of 0.037 in the linear probability of observing paramilitary units in a given leader-year. Aid also increases the likelihood that the dictator personally and directly controls the security apparatus.Footnote 11 Previous studies, as noted above, had found that aid increases military spending (Collier and Hoeffler, Reference Collier and Hoeffler2007; Kono and Montinola, Reference Kono and Montinola2013; Langlotz and Potrafke, Reference Langlotz and Potrafke2019). Our results suggest that the effect of aid concentrates on the structure of the security forces. Rather than strengthening the whole military, dictators may divert some additional external funds to create and maintain specialized, loyal units for personal protection—which are typically better paid and equipped than the regular forces—and, arguably, by increasing dependency, establishing more direct control over the security apparatus.Footnote 12 Even though “the diversion of scarce resources to pay and arm [paramilitaries] reduces the benefits that can be allocated to the regular military” (Geddes et al., Reference Geddes, Wright and Frantz2018, 167), the additional revenues from aid would allow dictators to fund these parallel forces to counterbalance the army and tighten their control over the security apparatus. Furthermore, the fact that aid reduces the probability of dictators creating their own new party might indicate a reduced need to create a civilian, popular counterbalance to the army due to the increased capacity to have parallel units and control the security forces.
Table 3. Foreign aid per capita and disaggregated personalism indicators, LPMS
Note: Clustered (Leader) standard errors in parentheses.
*** p < 0.01, ** p < 0.05, * p < 0.1.
Consistent with this, the results in Table 5 show that aid has a positive but insignificant effect only on the security dimension of personalism. This is unsurprising given that the effect concentrates mostly on just one policy action and it is substantively small.
As for FDI, the evidence reported in Table 4 shows that it significantly reduces the likelihood of observing two types of actions: military purges and controlling military promotions, with a one standard deviation increase in FDI correlating to a decrease of 0.03 and 0.026 points in such probabilities, respectively. Interestingly, this suggests that FDI may undermine dictators’ ability to shape the composition of the ruling coalition, especially the military command structure. In accordance with this, the estimates in Table 5 confirm that FDI significantly reduces only the security personalization dimension.
Table 4. FDI per capita and disaggregated personalism indicators, LPMS
Note: Clustered (Leader) standard-errors in parentheses.
*** p < 0.01, ** p < 0.05, * p< 0.1.
Table 5. Foreign aid per capita, FDI per capita, and security and party personalism
Note: Clustered (Leader) standard errors in parentheses.
*** p < 0.01, ** p < 0.05, * p < 0.1.
“Dictator-controlled promotions and purges demonstrate the dictator’s capacity to change the command structure of the military, and thus the composition of military decision-making bodies” (Geddes et al., Reference Geddes, Wright and Frantz2018, 81–82). FDI appears to push dictatorships toward an equilibrium where (primarily military) elites gain greater control over positions of power and influence through institutionalized mechanisms. This is especially likely in contexts where the military has historically gained control or influence over key economic sectors as a form of co-optation. The military may also seek partnerships with foreign investors in strategic and mobile industries such as defense, infrastructure, or telecommunications. Through joint ventures or strategic alliances, the military can access technology, expertise, and capital, thereby enhancing its influence and operational capabilities. Moreover, some foreign investors might prefer dealing with the military due to its stability, reliability, and protective role. FDI inflows in such contexts can strengthen military elites’ bargaining power and autonomy, providing them with increased economic and informational resources, greater financial independence, and a more diversified portfolio, as well as fostering cooperation in revenue generation. As foreign entities engage with military-controlled sectors, the military becomes a key point of contact and negotiation, further solidifying its status as a central actor in both domestic and international affairs.
In Egypt, for example, the military has traditionally controlled economic sectors—such as real estate, telecommunications or energy—since Sadat’s creation of the NSPO and the access given to enterprises under the Ministry for Military Production in order to halt military discontent. To ensure military compliance, economic control expanded further under Mubarak. “Military leaders [..] used their influence and roles in the state to forge new partnerships with foreign capital and the private sector” (Brooks and White, Reference Brooks and White2022, 132). These partnerships allowed the military to deepen its foothold in strategically important industries such as infrastructure projects. FDI in these sectors bolstered the military’s financial resources and its influence within the regime, displacing civilian economic elites. Under Mubarak, the military gained control over key cabinet portfolios, participation in parliament, and, importantly, kept authority over its own internal organization (Brooks and White, Reference Brooks and White2022).Footnote 13
7 Conclusion
This article investigates how ODA and FDI influence autocratic power dynamics by focusing on the level of power concentration in the hands of the dictator. We expected foreign aid to be related to higher personalism and FDI to have the opposite effect. Overall, our findings suggest that neither aid nor FDI seem to have a substantial and robust relationship with personalism.
This article contributes to understanding whether external factors influence the distribution of power in autocracies and provides new evidence to the debate of whether foreign income is a curse or a blessing to recipient countries. Given the effect on some of the individual indicators of personalism, governments, and MNCs should be aware of the implications that external forms of funding and support can have on the structure of autocracies. Our results, however, warrant further research on the foreign determinants of personalist rule.
The article is not without limitations. A critical aspect of our analysis involves potential remaining endogeneity and selection issues. For example, while some aid donors may seek to prop up personalist leaders aligned with their geostrategic goals, foreign investors often prefer environments characterized by executive constraints and stronger rule of law. Although we utilize TWFE and lagged variables to mitigate these concerns, these methods may not fully account for the selection processes influencing aid and FDI inflows. Investigating sectoral FDI and forms of aid could shed light on how personalism might respond to distinct types of funds.Footnote 14 Addressing these issues represents an important avenue for future research. In addition, while our analysis uses yearly data to examine the impact of FDI and ODA on personalism, we acknowledge that the temporal dynamics of these processes may be more complex. Short-term changes in FDI or ODA might not immediately influence personalism, and it is possible that prolonged or cumulative exposure could have a more significant effect. Future research could explore these longer-term dynamics to better capture how repeated interactions with foreign aid and investment shape personalism over time.
Supplementary materials
The supplementary materials for this article can be found at https://doi.org/10.1017/psrm.2025.9.
Replication materials
The replication materials for this article can be found at https://doi.org/10.7910/DVN/MNXYWI.
Acknowledgements
We thank Ellen Lust, Alice Iannantuoni, Joan C. Timoneda, Irene Rodríguez, the participants at the 2023 and 2024 EPSA General Conference, the participants at the Universitat Pompeu Fabra Institutions and Political Actors Research Group Seminars, two anonymous reviewers and the editor for their helpful comments and feedback.
Competing interests
None.
Open access
