ACRN Blog: Audits and Anti-Corruption

Please note that views expressed in this blog post are those of the author alone and do not represent the position of The Anti-Corruption Research Network, Transparency International, or any of its affiliates.

Centralised Auditing as an Anti-Corruption Mechanism^[1]

By Paul Lagunes, Contributing Editor, The Anti-Corruption Research Network

Many parents, after reaching a point of desperation, use the warning that Santa is watching to get their children to finish a meal. More often than not, the warning that Father Christmas is making a list and checking it twice works. There is a similar strategy available to governments around the world. A supervisory figure can be used as a means of maintaining discipline and honesty among civil servants. The supervisor’s role is that of monitoring.

In the context of a liberal democratic government, the act of monitoring is meant to ensure that those to whom authority is granted remain accountable to the public interest by ensuring that policy outcomes are reached.[2] Toward this end, there are different kinds of oversight mechanisms available. This blog post focuses on one of these oversight mechanisms–centralised auditing. 

Audits as a Form of Monitoring

McCubbins and Schwartz coined the terms “fire alarms” and “police patrols” in reference to two distinct modes of oversight.[3] Both terms are particularly popular in the Political Science literature on congressional oversight. On the one hand, fire alarms are a reactive, sporadic, and non-systematic method of uncovering malfeasance. This particular form of supervision relies on third party actors (e.g., citizens, interest groups, or the media) to examine administrative decisions and bring attention to deviant behavior. On the other hand, police patrols represent a centralised, continuous, and direct approach by regulatory actors of uncovering policy violations. A good example of police patrol oversight is an audit, which can involve the analysis of work-flow data, execution of rigorous spot-checks, and the study of documents, diaries, and time logs.

The United States Government Accountability Office defines an audit as “an objective and systematic examination of evidence to provide an independent assessment of the performance and management of a program against objective criteria….”[4] This particular mechanism is indispensable in deterring corruption because it raises the probability that a corrupt official will get caught committing a dishonest act and, as a result, will be punished.[5] This, at least, is the theory—a theory that, as several scholars have noted, demands stronger empirical backing.[6] 

Testing the Watchful Eye

There are a few empirical studies that examine an audit’s capacity to modify and improve civil servants’ behavior. Di Tella and Schargrodsky used Buenos Aires, Argentina, of the mid-1990s as their testing ground.[7] Corruption was rampant in the city’s public hospitals at the time. In response, Buenos Aires’ newly elected government orchestrated a crackdown on corruption between 1996 and 1997. During this period, they ran audits to collect information on the amount paid by all public hospitals in the city for a number of basic, homogenous supplies (e.g., saline) for which market prices were known and stable. Examining the effect of the audits, the authors found that after the introduction of the monitoring policy the prices of these supplies dropped significantly - around 15%.  

Centralised audits also appear have a disciplining impact on elected officials. In another relevant study, Ferraz and Finan analyse electoral results in conjunction with corruption data obtained from audits run by Brazil’s federal government.[8] They show that empowering citizens with objective information about their local elected officials reduces the likelihood that corrupt mayors get reelected. In a subsequent paper that also relied on Brazil’s federal government’s audit data on municipal corruption, Ferraz and Finan provide empirical evidence to confirm that the threat of being kicked out of office significantly reduces mayors’ proclivity to misbehave.[9] Thus, if we consider the findings from Ferraz and Finan’s two studies, it appears that the combined effect of oversight and punishment can have a restraining effect on public officials’ behavior. 

Some experts, however, have questioned the value of centralised audits. Concerns have been raised about an audit’s time and monetary costs. They also question whether centralised auditing is worthwhile considering the risk that the auditor himself could also become corrupt. Furthermore, they claim that other forms of oversight, such as grassroots monitoring (i.e., a mechanism that empowers citizens to monitor and discipline public service providers) are preferable to centralised auditing. Responding to this last set of critics, Olken ran a field experiment in over 600 Indonesian villages to test whether grassroots versus top-down monitoring is more effective at reducing corruption.[10]

The villages that participated in Olken’s study were involved in a nationwide road-building project funded by the World Bank. The projects were implemented by the village governments and villages were randomly assigned to one of three groups. The first group of participating villages was told that their project would be audited by the central government. The second group was told that locals, who have a vested interest in honest and prompt service delivery by the government, would participate in supervising the project. The third group served as control. To measure corruption in the three groups of participating villages a team of professionals generated an independent estimate of the amount that each project in each village cost to build. This objective estimate was then compared to the amount that each participating village reported it spent on the project. The gap between the two sums indicated the total amount of funds that were likely swindled. The results of the study showed that, between the two forms of oversight, centralised auditing was associated with a more significant reduction in missing expenditures, especially where the village heads planned to run for reelection.

 What We Still Don’t Know about Monitoring

The studies summarised above all have one important and consistent finding: audits can work. Nevertheless, these studies also have shortcomings that demand consideration. For one, even as they confirm the power of oversight, they do not offer equally reliable conclusions on what type of (or even whether) punishments need to be combined with oversight for them to be effective. Furthermore, the findings may be case specific. The studies mentioned took place in Buenos Aires, Argentina, Indonesia, and Brazil’s districts. But would audits work equally well in other parts of the world? Replication in additional settings is necessary to develop a fully reliable theory on the effects of audits.

In fact, if we wish to fully understand an audit’s impact on officials’ behavior, then scholars must help us address a number of lingering questions. For one, can we truly distinguish an audit’s effects on officials’ decisions? Could it be that other observable or unobservable factors are also playing an influential role? In their observational study, Di Tella and Schargrodsky claim that the monitoring scheme orchestrated by the Buenos Aires’ new administration explains improvements observed in procurement officials’ behavior. However, the study’s results could be conflated with the impact that transitioning to a new administration can have on a bureaucracy’s rank-and-file. This is a particularly important point considering that, in 1996, a politician who had campaigned on the promise to fight corruption was elected as mayor. Audits or no audits, it is reasonable to expect that most officials put their best foot forward in light of the new mayor and his team. To disentangle the audits’ exact effects on agents’ behaviour, Buenos Aires’ central government would have had to impart the probes at random. This is precisely what Brazil’s federal government did under Luiz Inacio Lula da Silva’s regime—a fact Ferraz and Finan astutely exploited for their research.

To conclude, though there are studies that are beginning to assess the disciplinary powers of audits there is still plenty that we do not know or that we do not know with certainty. Meanwhile, communities around the world are desperately seeking reliable advice on how they can fight corruption. So, what can we tell them? For the time being, we can say that centralised audits appear to work, even though our knowledge about them remains imperfect.

[10] B. Olken. "Monitoring Corruption: Evidence from a Field Experiment in Indonesia" Journal of Political Economy, 115(2): 200-249, 2007