Corruption, Elite Contestation, and Parliaments: Why Do Legislatures Become Stronger in Authoritarian Regimes?

The last decade has seen several authoritarian leaders personalizing power such as Putin in Russia, Erdogan in Turkey, and Xi in China. It is, therefore, no surprise that recent research has widely studied the personalization of power in autocracies and highlighted its (negative) consequences on regime survival, leader survival, and repression (e.g., Frantz et al. 2020; Grundholm 2020; Chin et al. 2022). A related pressing issue is how to credibly constrain dictators and prevent them from centralizing power in the first place. Here, especially the role of institutions such as parliaments remains debated among scholars.¹ While some characterize them as crucial for restraining dictators and providing a forum for elite bargaining (e.g., Svolik 2012), others dismiss their usefulness or even view them as “rubberstamps” (e.g., Brancati 2014; Meng 2020).

Yet, the strength of legislatures, that is, parliaments’ powers vis-à-vis the executive, and therefore, their ability to effectively constrain dictators, differs remarkably (e.g., Fish 2006; Barkan 2009) with important policy implications, for instance, for regimes’ economic growth and investment (e.g., Wright 2008; Jensen et al. 2014; Bonvecchi and Simison 2021). While the cross-regime differences in legislative strength are increasingly well understood (e.g., Wilson and Woldense 2019; Wiebrecht 2021), there is no reason to assume that legislatures’ strength remains constant once established. Yet, research has not advanced to explain the temporal dimension of authoritarian parliaments’ power cross-nationally. In other words, why do legislatures become stronger (or weaker) over time under authoritarianism? Cases from regimes as diverse as the German Democratic Republic (Schirmer 2002), Saudi Arabia (Seznec 2002), Myanmar (Egreteau 2019), Zambia (Opalo 2019), as well as Tanzania (Collord 2022), illustrate that legislatures’ institutionalization and strengthening is an empirical reality that warrants further analysis.

This article aims to fill this gap and provides an original explanation for the changes in legislative strength over time in authoritarian regimes. Drawing on recent scholarship on institutional development in African regimes (e.g., Opalo 2019; Demarest 2021; Collord 2022), this paper puts forward the argument that stronger legislatures are associated with higher levels of corruption in a given regime. Although seemingly counterintuitive, engaging in corrupt activities allows delegates to accrue economic and political influence which in turn, enables them to establish power bases independent of that of the dictator. Consequently, they can shift (or manifest from the outset) the balance of power between them and the executive and ultimately, advance legislative powers that constrain the powers of dictators.

Utilizing panel data on parliaments and legislative corruption in authoritarian regimes from 1946 to 2010, this paper finds that authoritarian parliaments are stronger under higher levels of legislative corruption in the respective regimes. This relationship holds in different regions across the globe and is particularly pronounced in party-based and military regimes. An alternative explanation that puts the increased competition in legislative and electoral processes at the forefront, on the other hand, generates mixed results. While some indicators that point towards more competitiveness such as the existence and autonomy of opposition parties strengthen authoritarian legislatures, others including the openness of elections to multiple parties and narrower margins of victory of the authoritarian incumbents do not. This may be due to a great deal of uncertainty added by some competitive processes that may ultimately undermine the role of legislatures as a venue for elite bargaining.

This work contributes to several strands of literature. First, while a growing body of literature examines how dictators concentrate power in authoritarian regimes (e.g., Tolstrup and Souleimanov 2022; Meng 2020), research on how institutions can become a more credible constraint for dictatorial rule is more limited. Thus, this research complements the growing interest in legislative development primarily in Africa (e.g., Barkan 2009; Opalo 2019; Collord 2019) with the first comparative cross-regime analysis of how authoritarian legislatures become stronger over time. It shows that legislative strength is primarily driven by intra-elite power relations and bargaining but benefits from more competitive electoral and legislative procedures only to a limited extent. As such, this research advances our understanding of when and how elites can enhance the role of institutions in authoritarian regimes and transform them into constraints for the dictator (e.g., Meng 2020). More generally, this study’s findings also help us understand how elites can become counterweights to the personalization of power in autocracies.

Second, this article speaks to the literature on the role and effects of corruption in authoritarian regimes. There is a long tradition of research illustrating how corruption undermines nominally democratic structures and institutions (e.g., Bratton and Van de Walle 1997; Seligson 2002; Anderson and Tverdova 2003). My research, however, highlights that by altering the underlying distribution of power, corruption can contribute to strengthening institutions instead. It thus adds to a recent strand of literature applying a more nuanced view to these practices, especially in the context of authoritarianism (e.g., Jiang 2018; Jiang et al. 2022; Demarest 2021; Collord 2022). In combination with this line of research, my work points to the need to study the consequences of corruption in authoritarian regimes more systematically.

In the following, this article will introduce its theoretical framework explicating why corruption is associated with legislative strength. Subsequently, the empirical strategy to test these claims will be introduced and followed by the presentation of the results. Lastly, the findings of this article will be discussed with reference to the broader literature on authoritarian regimes.

Parliaments and other institutions have been identified as cornerstones of authoritarian rule (e.g., Gandhi 2008; Wright 2008). Dictators create them in order to coopt opposition actors (e.g., Gandhi 2008) or provide a credible forum for power-sharing with other elites (e.g., Svolik 2012). Given these alleged benefits, it is unsurprising that, by now, virtually all authoritarian regimes have established parliaments as part of their institutional framework (e.g., Meng 2020).

On a general level, legislative strength refers primarily to parliaments’ relative power in the political system and especially in relation to the executive. In his pioneering work on classifying legislatures, Mezey (1979, p. 153), for instance, understands legislative power as “the importance of the legislature in the policy-making process” and holds that “legislatures will be salient in the policy-making process to the extent that their presence and prerogatives act as a constraint on the executive.” In other words, it is the relationship between legislature and executive that defines the strength of the former (and latter). Relatedly, scholars have for decades warned of the “legislative decline,” claiming that executives have and will continue to become more powerful at the expense of parliaments (e.g., Wheare 1963; Khmelko et al. 2020). Thus, changes in policy-making powers are often conceptualized as zero-sum games in which power flows from the legislature to the executive or vice versa (e.g., Chaisty and Power 2023). Essentially, then, legislative strength denotes the balance of power between parliament and executive, that is, the dictator in authoritarian regimes.

Yet, Geddes (1999, p. 121) famously held that “different kinds of authoritarianism [significantly] differ from each other” and the same applies to authoritarian regimes’ parliaments. Country-specific evidence suggests that in the shadow of party-based authoritarian rule, many legislatures are relegated to a mere consultative character (e.g., Thomas and Sissokho 2005) or “public forum for the airing of party propaganda” (Forrest 1987, p. 113). Others, however, have maintained significant veto powers throughout authoritarian periods such as the Legislative Advisory Commission in Argentina (Bonvecchi and Simison 2017). The Turkish Grand National Assembly also retained many of its constitutionally enshrined powers during the country’s authoritarian periods and the Parliament of Singapore continues to be more powerful than other authoritarian legislatures with its limited, but nevertheless, existent checks on the executive (Rodan 2009).

However, there is no reason to assume that legislatures’ strength remains uniform over time. In addition to differences in parliaments across regimes, many of them also undergo varying development trajectories over time within regimes. Recent literature from the African context, for instance, illustrates divergent institutional evolutions qualitatively in Kenya, Zambia, Tanzania, and Uganda (e.g., Collord 2019; Opalo 2019). Even the Consultative Assembly of Saudi Arabia (e.g., Seznec 2002), the People’s Chamber in the German Democratic Republic (e.g., Schirmer 2002), and the Chinese People’s Political Consultative Conference (Wiebrecht 2022) have been largely powerless institutions that showed signs of increased activity in different periods of time. These cases also illustrate that legislative strengthening is not necessarily linked to sequences of democratization but can also occur in their absence. Thus, more research is needed regarding the question of why parliaments become stronger under authoritarianism.

Explaining this phenomenon is not only important in its own right but also has important implications for our understanding of authoritarian politics. By now virtually all authoritarian regimes have legislatures, and therefore, I echo concerns that their mere presence may not be suggestive of constraints to dictatorial rule (e.g., Meng 2020). The powers that legislatures possess, however, give an indication of their ability to constrain leaders and that of the respective members of parliaments. By and large, the stronger a legislature, the more of a counterweight it constitutes for authoritarian rule. Thus, in light of increasingly personalist authoritarian politics under “strongmen” such as Putin and Erdogan (e.g., Frantz et al. 2020; Baturo and Elkink 2021), it is crucial to assess how legislatures and elites can form a credible constraint for dictators.

Beyond constraining leaders, stronger legislatures have also been associated with important policy consequences. Jensen et al. (2014) find evidence that stronger, that is, binding authoritarian assemblies are associated with more private investment and economic growth (see also Wright 2008). Bonvecchi and Simison (2017) hold that stronger legislatures can also shape public policy more decisively by rejecting and amending government bills. Finally, how powerful an authoritarian assembly is may also have consequences on wealth and resource distribution in society (Collord 2019).

The question of why legislatures are and become stronger over time is linked to the question of how elites can shift the balance of power vis-à-vis the dictator in their favor. This perspective is also anchored in scholarship from historical institutionalism and the political settlements approach (e.g., Khan 2018) that see institutions as “distributional instruments laden with power implications” (Streeck and Thelen 2005, p. 9) and characterizes them as arenas for conflict between “rule-makers” and “rule-takers.” Often serving the interests of the former, institutions “reflect, and also reproduce and magnify, particular patterns of power distribution in politics” (Thelen 1999). Institutional change, then, often goes hand-in-hand with changes in the underlying balance of power between actors and coalitions (e.g., Mahoney and Thelen 2009). This also applies to legislatures, as Collord (2022, p. 3) holds that “variation in legislative institutional strength matters in that it reflects and magnifies the distribution of power within a ruling elite.”

Building upon recent advances in the study of legislatures’ political economy and institutional development, primarily from the African context (e.g., Collord 2022; Demarest 2021), I argue that a key factor in enabling elites to shift this balance of power is their engagement in corruption. Corruption is widespread in many developing and authoritarian regimes and is often seen as undermining development, democracy, and political institutions (e.g., Svensson 2005; Rose-Ackerman and Palifka 2016; Rothstein 2011). Yet, under authoritarianism, engaging in corruption is also a crucial pathway for elites of the ruling coalition, including legislators, to accumulate wealth and consequently, build bases of power independent of that of the dictator.

Political corruption refers to the unlawful use of public office for private gains (e.g., Shleifer and Vishny 1993; Rose-Ackerman and Palifka 2016). In addition to merely enriching themselves, authoritarian elites also frequently utilize state resources to retain and expand their political power (e.g., Manzetti and Wilson 2007; Fjelde and Hegre 2014). Fjelde and Hegre (2014, p. 270) describe a multitude of actions through which this takes place including “offering selective tax exemptions, public appointment, land allocations, lucrative government contracts, discriminatory enforcement of the law, purchase votes, or condone graft from lower level public officials.” In many cases, political corruption is also decentralized in the sense that elites compete with each other for personal benefits and influence (e.g., Collord 2016) but that the precise strategies of doing so are under individual discretion.

Legislative offices particularly offer opportunities for elites to engage in corruption. More so than questions of policy-making, issues of distribution are often at the forefront of parliamentary business in authoritarian regimes (Lust-Okar 2006; Manion 2015). Opalo (2022a), for instance, describes how corruption increased under the Harambee Movement which in practice required legislators to tap into public resources to satisfy constituent demands. Now, constituency development funds have institutionalized the process but are still frequently misappropriated (Opalo 2019). Beyond misdirecting public funds for constituents, legislators often also siphon off money into their own pockets. Committee roles are especially lucrative posts for these activities (e.g., Demarest 2021; Opalo 2022b).

Through engaging in (far-reaching) corruption, authoritarian elites can build up networks of support and influence. They can share public resources with their supporters and thus, manifest their power base (e.g., Manzetti and Wilson 2007; Engvall 2022). Closely related to clientelism, these networks and exchanges help legislators to become important actors in their own right (e.g., Bratton and Van de Walle 1994; Sigman and Lindberg 2019). However, while clientelism involves a hierarchical element and often primarily refers to exchanges between legislators and voters (e.g., Rothstein and Varraich 2017), corruption can span more interactions. For instance, corruption can (and frequently does) also take place between political elites across government branches and between political and economic elites (e.g., Gray 2015; Demarest 2021; Collord 2022).

Thus, when corruption is more prevalent, more public resources are being used for private gains and, in consequence, legislators can be expected to be in a better position to challenge dictators (e.g., Fjelde and Hegre 2014). Collord (2022, p. 8), for instance, holds that with

“greater private accumulation [of wealth] spread across an expanding class of domestic capitalists, leaders are less able to control factional rivalries […] or to prevent intra-party contestation from surfacing in parliament, where it motivates more legislative institutional strengthening and assertiveness”.²

Indeed, many authoritarian elites have built up support networks challenging dictators in and outside of parliaments through corruption and patronage. Opalo (2019), for instance, highlights how legislators in Kenya have become powerful local patrons and consequently, can withstand pressures from the executive and remain autonomous. Felix Houphouet-Boigny also saw himself confronted with barons that had established independent support networks across the Ivory Coast (e.g., Medard 2014). Additionally, in Zimbabwe, Morgan Tsvangirai strengthened his position similarly prior to challenging dictator Robert Mugabe (Brancati 2014). At the same time, dictators’ anti-corruption efforts and campaigns are often not only strategies to eradicate corruption as such but also to eliminate rivals’ power networks (e.g., Zhu and Zhang 2017; Carothers 2022).

With established independent sources of power, legislators also have an interest in strengthening the role of the legislature vis-à-vis the executive. Opalo (2020, p. 1349), for instance, holds that “all else equal, legislators want to play as big a role as possible in bargaining over policy and supporting legislation, including the ability to determine the distributional benefits of legislation.”³ Material independence will provide elites with the necessary resources to balance dictators and engage in intra-regime bargaining (e.g., Boix and Svolik 2013; Reuter and Robertson 2015; Reuter and Szakonyi 2019). For example, legislators will have an interest in improving oversight and acquiring a say in approving legislation that would shield them from executive encroachment. With greater financial and political power they likely also demand more involvement in how resources are being used, for instance, for the question of whether their country should enter a war. This is a similar mechanism to how medieval European rulers were constrained by emerging parliaments (e.g., Stasavage 2011; Cox et al. 2023).

Therefore, in essence, the argument presented here is that corruption offers a mechanism for authoritarian elites in the ruling coalition, including legislators, to develop economic and political influence and consequently, alter the distribution of power between them (legislature) and the dictator (executive). Importantly, following the conceptualization of legislative strength as powers vis-a-vis the executive, with increased economic and political power, elites may constrain the dictator in several different areas. Although individual legislative powers are specific and can be quite different from each other, that is, whether parliament has to approve proposed legislation or wars, they all constrain dictatorial rule in one way or another or strengthen the position of parliament in the political system (e.g., through legislatures controlling their own resources). Therefore, increased legislative corruption is expected to be reflected in a general strengthening of legislative powers across the board.⁴

On the other hand, the question of which particular additional powers the legislature gains as a result of this intra-elite bargaining may be highly context-dependent. For instance, it may depend on the question of which powers the parliament already has and which powers strengthened elites deem more important (i.e., sometimes a specific clause on war approval may not be their first choice). Figure 1 illustrates this mechanism visually. Therefore, the formal hypothesis is as follows:

Higher levels of legislative corruption are associated with stronger parliaments in authoritarian regimes.

However, a classic competing explanation puts the development of legislative and electoral competition at the forefront. Based on the development of the American Congress, for instance, rational choice theories highlight that legislators are investing resources into institutional development and strengthening when they are motivated by re-election (e.g., Mayhew 2004; Ferejohn 1986). Elections have therefore been highlighted as an important source of legislative power for several reasons. Opalo (2019) describes how legislators can utilize public support to improve their bargaining power vis-à-vis the executive. Barkan (2009) highlights that elections are a necessary precondition for the creation of “coalitions of change” that will advance legislative strengthening. In the broader literature, transitions to multiparty elections have long been characterized as defining moments in the institutional development of legislatures (e.g., Linz and Stepan 1996; Bratton and Van de Walle 1997; Diamond 1999; Barkan 2009; 2013). Co-optation theory in authoritarian regimes also suggests that legislatures should be stronger once they allow for electoral competition and house an opposition because the latter is only incentivized to join parliament when it actually has the power to influence policy-making processes (e.g., Gandhi 2008).

Thus, a broad strand of the literature suggests that legislative strength goes hand in hand with more competitive processes of candidate (s)election and policy-making. The following empirical analyses on authoritarian parliaments’ strength will also test this potential explanation.

In the following, I examine the abovementioned argument with a quantitative empirical test using linear regressions with Time-Series Cross-Sectional (TSCS) data on authoritarian regimes between 1946 and 2010. To identify the universe of authoritarian regimes during that time, I rely on the TSCS dataset constructed by Geddes et al. (2018) (GWF). The analyses will be restricted to the authoritarian regimes that have established a legislature which can also be identified from the same dataset. The unit of observation is the (authoritarian) country-year.

The main empirical analysis will focus on the changes in legislative strength. The dependent variable for the main analysis is legislative strength, that is, “the accumulation (or absence) of different powers of legislatures vis-à-vis the executive” (Wiebrecht 2021, p. 1078). This concept has been operationalized in different ways, most prominently perhaps with the help of the Parliamentary Powers Index (PPI) (Fish 2006). Yet, the PPI only allows for cross-sectional analyses and does not account for variation within regimes across time. Nevertheless, relying on V-Dem data, recent scholarship has been able to observe changes in legislative strength over time both, in authoritarian regimes as well as in democracies (e.g., Wilson and Woldense 2019; Coppedge et al. 2021; Wiebrecht 2021).

This paper follows previous studies’ approach in constructing an index of a total of 16 equally weighted legislative powers that, in line with the conceptualization of legislative strength offered previously, can capture the relationship between legislature and executive. This is an additive index that includes both de facto (10 powers) as well as de jure powers (6). De jure powers are often easier to identify cross-nationally and are often important cues about the balance of power between legislature and executive. Yet they may not accurately reflect the power relations ’on the ground’, especially in authoritarian regimes. Therefore, it is beneficial to include powers of both categories in the index. Some of the de facto powers have been assessed by expert coders estimating the abilities of legislatures, for instance, if it were to attempt to remove the head of state from office (Wilson and Woldense 2019). The relevant legislative powers comprised in this index include, for instance, legislatures’ abilities to introduce bills in all policy areas by law, to amend the constitution, and to regularly question the executive branch. In line with the expectation laid out above, most of the powers in the index are directly constraining the dictator or otherwise strengthening the parliament’s position in the political system (e.g., controlling its own resources). Despite giving parliaments very different kinds of powers, the indicators converge on the outcome, namely, that with every additional power, legislatures are stronger players in the political system and pose more of a constraint to dictatorial rule. A full list of all 16 powers is provided in Table A1 in the Appendix.⁵

The independent variable used in the analyses is legislative corruption. Naturally, corruption is difficult to observe and operationalize empirically. Due to its illicit nature, scholars have predominantly relied upon perception-based measurements for empirical analyses, Transparency International’s Corruption Perceptions Index in particular (e.g., Anderson and Tverdova 2003; Heywood and Rose 2014). This paper, however, uses V-Dem’s “Legislature corrupt activities” indicator that is based on an expert survey and asks, “Do members of the legislature abuse their position for financial gain?” (Coppedge et al. 2021). This operationalization focuses more precisely on members of parliament and is therefore the one closest in line with the theoretical mechanism proposed in this paper. For ease of interpretation, this variable has also been recoded so that higher levels of legislative corruption are reflected in higher values.

In addition, the empirical models also include information on whether the legislature contains members of an opposition. This data is taken from the Political Institutions and Political Events (PIPE) dataset and has been recoded into a dichotomous variable distinguishing regimes that include at least one opposition party in parliament from those that do not (Przeworski et al. 2013).

In order to account for the competing explanation, the empirical models will also include covariates that capture the competitiveness of elections and parliamentary procedures. First, in addition to noting whether a legislature houses an opposition, it is also important to incorporate whether these opposition parties are independent of the authoritarian regime. Put differently, opposition parties may only be considered as such if they are credible opponents of the dictator. Yet, due to several regime-sponsored “loyal” opposition parties, this may not be taken for granted (e.g., Albrecht 2005). In order to record opposition autonomy, I utilize V-Dem’s indicator with the same name. Furthermore, as mentioned above, elections have been highlighted as a primary factor in incentivizing legislators to invest further resources in advancing legislative strengthening. Thus, a dummy variable on whether a regime holds elections, also taken from V-Dem, will also be included in the empirical analyses.

However, in addition to their mere existence, it is also important to assess the nature of these elections and specifically, their competitiveness. In order to measure to what extent elections are open to multiple parties, I utilize V-Dem’s Multiparty Elections indicator, an ordinal variable distinguishing between no-party or single-party elections, constrained elections, and multiparty ones. In addition to whether (multiparty) elections take place, the outcomes of these may also hold relevance for legislative strength. Parliamentary strengthening may especially be conceivable when an authoritarian regime loses its (legislative) dominance. Thus, the authoritarian incumbents’ seat and vote shares will also be included as part of the analyses to assess whether their legislative dominance is associated with stronger or weaker legislatures. This data has been collected from V-Dem’s indicators on the seat and vote shares won by the largest parties.

Finally, (limited) competition may be induced even within single-party regimes due to candidate selection procedures. In Vietnam, for instance, legislators that were nominated on local levels, as opposed to being selected by national party leaders, showed significantly different legislative behavior and used the legislator to challenge the executive (Malesky and Schuler 2010). Thus, more decentralized candidate selection procedures may also be associated with legislative strengthening. Data on candidate selection has been taken from V-Dem’s ordinal indicator distinguishing between national and local candidate selection procedures.

In addition to these competing explanatory variables, several key time-varying covariates will be included as control variables in the empirical models. First, the most important control variable is the level of democracy since more democratic regimes tend to have stronger legislatures (e.g., Fish 2006; Wilson and Woldense 2019). To operationalize a regime’s level of democracy, this study utilizes V-Dem’s Liberal Democracy Index. Moreover, I include regimes’ GDP per capita, log-transformed, in the analyses and rely on the Maddison Project’s Database for this data. Wealth in natural resources may also change the structure of elite relations in authoritarian regimes and simultaneously affect patterns of corruption in a given regime. Thus, using data from Haber and Menaldo (2011), I control for regimes’ natural resources, also log-transformed. Furthermore, Communist regimes and their legislatures may develop differently from other forms of dictatorships. Jowitt (1992), for instance, showed how Communist regimes have undergone similar waves of inclusion and repression throughout the Cold War era. Therefore, I also include a dummy variable noting whether a particular regime is Communist. Finally, I also include cubic duration trends to account for potentially different behavior between younger and older regimes (e.g., Lucardi 2019). In particular, in younger regimes, formal institutions may not be as stable and undergo more frequent changes than in regimes that have proven durable already. The regime duration is taken from the GWF data.

The following models are estimated using two-way fixed-effects OLS regression models with robust standard errors. I include regime-fixed effects (as opposed to country-fixed effects) to account for time-invariant regime-specific characteristics and year-fixed effects to account for world trends commonly experienced by all regimes.⁶ All right-side variables are lagged one period (year) behind the outcome except for the existence of an opposition, whether the regime is Communist, and the duration trend. The descriptive statistics for the key variables are presented in Table A2 in the Appendix.

Figure 2 shows the differences in parliamentary powers across a randomly selected sample of regimes and illustrates that changes in legislative strength are relatively frequent among authoritarian regimes. A darker shade of blue signifies a stronger legislature in the sample of authoritarian regimes in this figure. While some regimes appear to have had stronger legislatures throughout their tenures such as the Soviet Union and Angola, some have undergone more frequent changes such as Uganda.

Figure 3 zooms in on some of the country-specific developments and shows the legislatures’ evolutions in Albania (1944–1991), Botswana (1966–2010), Egypt (1952–2010), Guinea (1958–1984), Guinea-Bissau (1980–1999), Jordan (1946–2010), Soviet Union (19171991), and Yugoslavia (1944–1990). These trajectories highlight different mechanisms ranging from gradual strengthening (Guinea-Bissau and Yugoslavia) and decline (Albania and Guinea) to extended periods of stability (Soviet Union, Botswana), and enormous fluctuation (Jordan, Egypt). Cases from these regimes also illustrate that legislative development is not always necessarily linked to processes of democratization and liberalization.

The baseline results are illustrated in Table 1. The dependent variable is a legislature’s score on the abovementioned index of legislative strength in a given country-year. Model 1 only includes legislative corruption as the primary explanatory variable alongside the control variables, whereas in Model 2 the existence of an opposition in parliament is added as a further covariate while other model specifications remain the same. The results show a highly significant and strong positive relationship between legislative strength and legislative corruption. In other words, as legislative corruption in a given regime increases, parliaments in those authoritarian regimes become stronger over time. Moving one point on the 5-point ordinal scale, for example, from no corrupt activities to very occasional incidences results in about 0.3 additional legislative powers in Model 1. This relationship remains strong and significant even when adding the existence of an opposition into the analysis.

Nevertheless, when authoritarian regimes decide to include an opposition, legislatures also tend to become more powerful (e.g., Gandhi 2008) as evidenced by the significant coefficient for Opposition. This result hints at the possibility that the alternative explanation emphasizing increased competition over legislative and electoral processes also holds some merit in explaining legislative strength. In line with prior expectations, legislatures are also stronger in authoritarian regimes with higher levels of democracy and those with higher levels of GDP. On the other hand, Communist regimes tend to have weaker legislatures while natural resources do not seem to affect the strength of legislatures in dictatorships.

In order to further analyze the role of electoral and legislative competition, several other covariates have been incorporated into the analyses. These include the autonomy of the opposition, the existence of elections, the openness of elections to multiple parties, the seat and vote shares of the authoritarian incumbents, as well as the character of candidate selection procedures. The results of these models are illustrated in Table 2. They underscore the importance of legislative corruption as it remains a strong and highly significant predictor of legislative strength. In addition, they also paint a mixed picture regarding the alternative explanation of increased competition in the legislature leading to legislative strengthening. As in the previous two statistical models, the presence of an opposition is generally associated with stronger legislatures. Beyond that, only the autonomy of the opposition parties is a positive significant predictor of parliamentary powers.

Other factors associated with higher electoral competitiveness, however, are associated with undermining legislative strength instead. For instance, the results reveal that the relationship between multiparty elections and legislative strength is a significant negative one (Model 3). Thus, more competitive elections, that is, those open to multiple parties, are associated with weaker legislatures. This points to the importance of having a unified opposition that may be in a better position to bargain with a dictator and force him to make concessions than several (smaller) opposition parties (e.g., Reuter and Robertson, 2015). In addition, legislatures tend to be stronger as the seat share of the authoritarian incumbent increases while the existence of elections, authoritarians’ vote share as well as the candidate selection methods do not appear to affect legislative strength. Therefore, the results for corruption and the presence of an opposition are the most consistent across all models, though most indicators of competitiveness do not improve legislative strength.

These results are robust to a wide variety of model specifications. First, the independent variables’ 1-year lag has also been replaced with a 3-year lag. Tables A3 and A4 in the Appendix demonstrate that the results also remain largely the same under this specification. I also estimate the baseline results using random effects in order to allow the effect size and direction to vary across country-years (see Tables A5 and A6). The results also remain robust here. Since the structure of the independent variable is such that the powers of a given legislature are counted, Poisson regression models may be equally well suited for the analyses. The results obtained using these models (Tables A7 and A8) do not diverge significantly from the baseline results presented above.

The main independent variable has also been replaced with alternative measurements in further models. Tables A15 to A18 in the Appendix illustrate the analyses using V-Dem’s Political Corruption Index and the Regime Corruption Index. The former is a broad index comprised of public sector corruption, executive corruption, legislative corruption as well as judicial corruption. Regime corruption here refers to the extent to which political actors use their political office for private gains which includes members of the executive, for instance. In addition, the liberal democracy index has been replaced with V-Dem’s commonly used electoral democracy index as a control variable. However, this change also does not affect the results significantly (Tables A19 and A20 in Appendix).

In order to account for a potential omitted variable bias, I have also estimated further models including factors that may also be theoretically important. First, corruption may only play a role in “weak” states and state capacity may be an intervening variable that determines whether delegates are able to engage in corruption in the first place. Taking this into account, Tables A21 and A22 in the Appendix includes Hanson and Sigman (2021) latent variable on state capacity. Controlling for it, however, does not change the effects of the other variables of interest. Second, other research posits that dictators choose electoral systems depending on their own strength, which may also need to be accounted for here (Chang and Higashijima 2023). In Tables A23 and A24 in the Appendix, I factor in three different electoral systems (proportional, mixed, and majoritarian) but the baseline results are also confirmed here.

Moreover, in order to avoid the danger of model overspecification, additional tests have excluded regimes’ wealth of resources (Tables A25 and A26 in Appendix), the binary variable denoting Communist regimes (Tables A27 and A28 in Appendix), and per capita GDP (Tables A29 and A30 in Appendix). Finally, in order to assess whether the results are driven by one particular legislative power, further robustness tests have dropped each of the 16 legislative powers from the analysis once. Yet, the results are robust to this specification as well.

Furthermore, given recent work suggesting limitations of two-way fixed effects models (e.g., Imai and Kim 2021), I also tested whether the models would be robust to dropping regime- and/or year-fixed effects. Tables A9-A14 in the Appendix highlight that the results remain stable when not including either one of them or even both.

Finally, it may be argued that reverse causality is a potential issue. A strengthened role for the legislature also translates into more political influence for legislators and consequently, amplified opportunities for them to further their personal agenda including financial interests (e.g., Demarest 2021; Lust-Okar 2006). While this potential problem cannot be ruled out entirely, I alleviate concerns of reverse causality and also estimate the relationship between legislative corruption and legislative strengthening using an instrumental variable approach. A neighborhood instrument provides further evidence for the causal relationship. The results and more details about this estimation can be found in the Appendix and in Table A33.

The above results provide evidence for the link between legislative corruption and legislative strengthening. However, in order to reflect the theoretical mechanism shown in Figure 1 empirically, it is also crucial to establish the intermediate step, that is, that increased legislative corruption, in fact, strengthens elites so that they will ultimately be able to challenge and constrain the dictators. Identifying this connection is a necessary condition for the argument and the main results’ credibility. Here, I draw especially on (Ziblatt 2009) who focuses on landholding inequality as a measurement of the control of clientelistic elites.

To corroborate this evidence on the cross-national level, I utilize measures of landholding inequality provided by Thomson (2016). Table A31 in the Appendix shows the results from linear regressions with Time-Series Cross-Sectional Data including regime-fixed effects and more extensive control variables such as the rate of urbanization, the size of the agricultural area in a country, and the elevation, all taken from Thomson (2016). The results show that an increase in legislative corruption is indeed associated with more landholding inequality. This effect is only significant at the 0.1 level for the land Gini index but highly significant at the 0.001 level for land inequality. Nevertheless, data missingness is high among the subsample of authoritarian regimes.

In order to provide further evidence for the proposed mechanism, I have also used cross-country data on income inequality from Solt (2020). Results from this analysis can be found in Table A32 in the Appendix. The sample of country-year observations is much larger here and the same model specifications as above show a highly significant effect of legislative corruption on a standardized measure of income inequality. This applies to both, measures of disposable income as well as market income. Taken together, these results provide evidence for the proposed mechanism that corruption enables elites to self-enrich and amass resources.

After having tested the mechanism across authoritarian regimes, I will next highlight some of the scope conditions of the argument. First, the geographical scope of this argument has been tested with the help of V-Dem’s classification of the world into six broad regions. Reiterating the above analyses for each of the regions shows that there are certain scope conditions to the mechanism laid out in this paper. The baseline results are largely confirmed for the regions of the Middle East and North Africa, Sub-Saharan Africa, and some models in Latin America and the Caribbean (Tables A34 to A43 in Appendix). In Eastern Europe and Central Asia as well as the Asia-Pacific legislative corruption is not linked to strengthening authoritarian legislatures.

Prior research has pointed out the ways that elites enrich themselves through public offices and in parliaments, especially in Africa and the Middle East (e.g., Lust-Okar 2006; Demarest 2021; Collord 2022). In Asia, the significant negative results may be due to the fact that several dictators have been avid adversaries of corruption such as Lee Kuan Yew in Singapore, Chiang Kai-shek in Taiwan, Park Chung-hee in South Korea, and most recently, Xi Jinping in China (e.g., Carothers 2022). In Eastern Europe and Central Asia, the findings are in line with the negative coefficients for Communist regimes in Model 1 which comprise a large number of observations in this region. This may be due to parliaments’ different roles in these regimes where they serve the purpose of projecting unity and have fewer distributional consequences (e.g., Schuler 2021).

A broad range of literature has now also established that authoritarian politics can vary significantly by regime type according to Geddes’ classic typology of party-based, personalist, and military regimes as well as monarchies (Geddes 1999). These regime types are significant predictors of different outcomes including regime durability (e.g., Geddes et al. 2018; Lee 2014), economic growth (Wright 2008), and international cooperation and conflict (Weeks 2012; Mattes and Rodríguez 2014). Thus, it may also be expected that institutional development varies by regime type. Tables A44 to A51 in the Appendix, illustrate that the argument advanced here primarily applies to party-based and military regimes and is only confirmed in some models in personalist regimes and monarchies.

These results are in line with expectations from prior research. First, it is unsurprising to see this mechanism not playing out to the same extent in personalist regimes and monarchies. Personalist leaders, to which many monarchs also belong, have by definition removed all constraints on their power and as a consequence, elites are unable to effectively bargain over power-sharing agreements let alone restrain the dictator through legislatures (e.g., Geddes et al. 2018). In contrast, party-based regimes are characterized by their tendency to institutionalize power-sharing (e.g., Geddes et al. 2018) and institutions such as legislatures often occupy central roles in elite bargaining (e.g., Reuter and Robertson 2015; Reuter and Szakonyi 2019). Military regimes are often similar to the extent that their leaders often “have had to retire from active duty” (p. 69 Geddes et al. 2018) and thus, gave up their control over elites’ careers and effectively, became subject to intra-regime bargaining. Both regime types, party-based and military dictatorships also tend to be the autocracies with the lowest levels of corruption overall (Chang and Golden 2010). Yet, due to the small number of observations, the results for military regimes should be interpreted with caution.

Thus, while the baseline results remain robust for different model specifications, the sample specifications reveal some of the scope conditions of the argument. The link between corruption and legislative strength is especially strong in the Middle East and Africa, and primarily applies to party-based and military dictatorships.

This paper has put forward the argument that contrary to conventional wisdom, corruption is an important factor in strengthening nominally democratic institutions in authoritarian regimes. Legislatures under authoritarianism benefit from more resourceful and powerful delegates that are capable of challenging the dictator or fending off autocrats’ attempts to curtail the role of parliament. A key mechanism for elites of the ruling coalition to become important actors in their own right is to accrue economic and political influence through corruption which, in turn, allows them to shift the power between legislature and executive. The empirical results of this study have substantiated these claims and provided evidence that this mechanism is at work in several regions of the world.

As such, this paper is the first to comparatively illustrate mechanisms behind the strength of legislatures under authoritarianism across the globe. It contributes to a growing interest in the evolution of nominally democratic institutions in authoritarian regimes that was in recent years primarily confined to the African context (e.g., Opalo 2019; Collord, 2022; Demarest 2021). The findings provide further evidence for the centrality of elite contestations in determining institutional trajectories in autocracies.

Competitive electoral processes, on the other hand, have long been associated with stronger institutions because they incentivize delegates to invest resources into institution-building (e.g., Barkan 2009; Mayhew 2004). Alternatively, they open channels for regime outsiders to gain access to parliament and use these opportunities to press the dictator for concessions (e.g., Gandhi 2008; Malesky and Schuler 2010). Nevertheless, what may have been missed by prior literature is that more competitive processes add uncertainty which may, in fact, undermine the role of the legislature as a venue for elite deals. Delegates may instead be less likely to use parliaments as a forum for elite bargaining when regime outsiders are also able to enter it so as not to reveal cracks among regime insiders. In fact, elite competition is primarily highlighted in legislatures with little or no competition between the ruling party and an opposition (e.g., Schirmer 2002; Noble 2020; Lü et al. 2020). The findings of this paper also point in this direction.

Nevertheless, the claims presented in this paper must not be interpreted to suggest that corruption is not undermining democratic institutions or that anti-corruption strategies should be abandoned by governments. Instead, this paper contributes to recent efforts to situate concepts such as corruption, patronage, and clientelism adequately in political processes in contexts in which it is relatively more prevalent, that is, authoritarian regimes, without applying a normative lens to them (e.g., Jiang 2018; Jiang et al. 2022; Demarest 2021). In other words, while corruption may not be a normatively “good” practice, it is nevertheless important to understand its causes and political ramifications.

Finally, this research also highlights that authoritarian elites can indeed work as counterweights to the personalization of power in autocracies. Nevertheless, whether the mechanism highlighted here is also conducive to democratization remains debatable. The examples of the Prussian junkers (Ziblatt 2009) and “rotten boroughs” (Cox 2005) suggest that elites entrenched in legislatures are often obstacles to democratization and comparative work also suggests that high levels of corruption may actually stabilize authoritarian regimes (Fjelde and Hegre 2014). The links between institutional evolution and regime transitions may therefore merit further more explicit work in the future.