Corruption for competence

Persson and Tabellini (2000), for instance, make this explicit in models in which politicians cannot contract future performance, making promises by a ‘benign’ opposition non-credible, or in which citizens cannot know ex ante the type of politician they are selecting. Contractual incompleteness and imperfect information thus afford the incumbent politician an inherent advantage. A similar reasoning is found in agency theory, e.g. Holmström (1979); Holmstrom (1982); Holmström (1999), in which firm managers engage in moral hazard and why reputation or career concerns alone cannot solve this.

See, also, Desierto (2021a) in which the political resource curse is modeled through theft and bribery, and Desierto (2021b) for an empirical measurement of Philippine mayors’ rents from theft and from bribery.

Rose-Ackerman (1978, 1999, 2007) makes a similar categorization – low level corruption that is committed by bureaucrats in the performance of their administrative duties, and high level corruption committed by politicians in their choice of intervention. For a survey of bureaucratic corruption, see Bardhan (1997); Aidt (2003), and Banerjee et al. (2012). For grand corruption, see Rose-Ackerman (2011b, 2011a); Rose-Ackerman and Palifka (2016), and Fisman and Golden (2017).

For example, revenues might be allocated toward government buildings that are vacant and unused. See https://abcnews.go.com/WN/taxpayer-owned-crumbling-vacant-government-buildings/story?id=10198415.

See, e.g., Desierto (2018a, 2018b); Brollo et al. (2013); Ahmed (2012); Abdih et al. (2012), and Robinson et al. (2006).

A simple example is when the politician allocates revenues towards a highway. Contractors bid to construct a highway, and the winning bid is the lowest cost because it will use substandard materials. Thus, if the winning bid is \(\tau =100\), the contractor spends that much to construct the highway. But on top of that, she pays the politician who awarded the contract a bribe of, say, \(r=10\). So the value of the highway must be above 100, i.e. \(\theta \tau =\theta (100)\), with \(\theta >1\). Indeed, if the contractor uses substandard materials, she can use the remaining materials for other projects, thereby gaining extra profit. The value to the contractor of that highway is, then, more than 100, which is why she can pay the bribe. The net value of the highway is thus \(g=\theta \tau -r=\theta (100)-10\).

An implication is that the rents from theft are obtained prior to the allocation of funds, while those from bribery occur after allocation. A corrupt politician can, for instance, steal some revenues from oil before spending the rest on public goods, or she can first spend revenues on developing oil fields and then appropriate some of the oil for herself or her cronies. In the first case, a political resource curse is generated through theft, in the latter case through bribery.

Coate and Morris (1995), in particular, show that politicians can prefer to distribute rents through public projects, which can be more disguised, than outright transfers. Applying the “Virginia School" perspective to our context would imply that bribe-rents are better hidden than outright theft of revenues.

Thus, the composition of the coalition changes depending on who the leader is. However, its size W is fixed — the disadvantage of using the selectorate framework is that it cannot explain how W becomes the stable coalition size.

This assumption is standard in selectorate theory. An exception is found in Desierto and Koyama (2020), where W is composed of subgroups, and each selectorate member has a different probability of being in W, depending on which subgroup she belongs to.

Of course, regime type and institutions may be co-evolving in many other ways. In this extension, I assume in effect that regime type is slower moving and determines the institutional bias toward theft or bribery. To the extent that the same democratic regime can adopt varying anti-corruption strategies, such assumption may be justifiable. Brazil, for instance, had a new democratic Constitution ratified in 1989, but the random auditing of municipal government revenues – which appears to have decreased the embezzlement or theft of funds (see Avis et al. (2018), and Ferraz and Finan (2011)), only begun in 2003.

It is telling that in spite of the fact that Brazil is a signatory to the convention, it took a long time to uncover the full extent of the Petrobras bribery scandal and to prosecute and convict the guilty parties.