To scare a politician

brazil

Does judiciary presence affect the politician’s management or possible graft of public resources? In their Barcelona GSE Working Paper (No. 796) “Judicial Presence and Rent Extraction,” Stephan Litschig and Yves Zamboni estimate the effect of the presence of judges and prosecutors in the counties of Brazil on the management of public resources, as measured by external government auditors.

The role of the judiciary

One role of the judiciary is to resolve disputes between private parties. Another role that is enshrined in constitutions throughout the world is to monitor and constrain the executive branch. Previous work based on cross-country comparisons has shown that the executive power unchecked imposes detrimental effects on political and economic freedoms. But as with all cross-country work it is hard to tell whether the associations found are indeed causal. Moreover, there is a debate whether the measures of judicial independence used in this line of work reflect permanent constraints on the executive, rather than policies (which have a tendency to shift with political winds).

Litschig and Zamboni study whether judicial presence in counties in Brazil affects rent extraction by local executive officials. They define rents as infractions of public management regulations by local officials as revealed by auditors conducting randomly assigned audits. They posit that judicial presence increases the cost of extracting rents by increasing the probability of detection, as the public faces lower transaction costs of reporting irregularities. On the flip side, the presence of a court may allow local politicians to capture the judiciary and render the judicial check on executive actions mute. This leads to an ambiguity of the effect of judiciary presence on rent extraction at the local level.

As usual, that pesky endogeneity….

Reverse causality is an obvious issue when studying the topic at hand as judiciary presence in a municipality is itself a choice (even if this choice may be collective). Or to put it differently, not only is it possible that judiciary presence has an effect on rent extraction (the effect the authors want to study), it is possible that the existence of rent extraction affects whether a court is present in the first place. To study the effect in the desired direction, an instrument is needed that breaks this reverse causality, something that is related to judiciary presence but not related to rent extraction per se.

The authors have some knowledge about how it is that a county in Brazil receives the physical presence of a court. The law in Brazil stipulates the minimal conditions for the creation of a judicial district – and with it the creation of a court with judges and prosecutors – in terms of population size, along with some other characteristics (like geographical size, county fiscal revenue, judicial caseload). However, it is not usually the case that a single county satisfies these conditions. They show that 75% of counties do not become their own judiciary district but are grouped with contiguous neighbors so that only one of them receives the judicial seat for the district. The authors discovered that the (mostly informal) assignment rule is to locate the judicial seat in the most populous county within the district in order to reduce transaction costs for the citizens. It is this rule that the authors use as an instrument by ranking counties within a district according to population size and determining the probability the highest ranked county receives the judicial seat.

So what is the effect of being “treated” with a judicial seat?

First, the authors need to determine the effect of the instrument on the presence of the judiciary in a county. Their estimates, as shown in Figure 1, indicate that the probability of obtaining the judicial seat when a county is top-ranked in its district increases by 80% points – where the striking feature is the near constant 80-90% point increase in likelihood of obtaining the seat irrespective of the population level (the x axis). This probability drops to about 73% when other county observables (which are significant) are included. The authors note that because population rank has such a large effect on the chance of hosting the court, this estimate, although a local effect, may be fairly representative of the average effect for small and medium sized counties.

figure

Using the above results, the authors find that judicial presence reduces the share of inspections with an irregularity by 10%, where the effect is driven almost exclusively by a reduction of substantive rather than procedural irregularities. This makes intuitive sense as these are the most likely to involve corruption and hence the infractions that are likely to lead to not only detection by the public but also of being prosecuted by the judiciary (even if only at the lesser charge of mismanagement, since corruption is often too difficult to determine). These findings are also quite robust. Figure 2 below shows that, on average, irregular inspections decrease by 4% points for the top ranked counties in the district compared to the lower ranked counties.

figure

The authors go one step further and split the sample by the mayor’s term in office. They find that among first-term mayors, judicial presence reduces the share of inspections with an irregularity by 8-10% points. On the other hand, second-term mayors – who cannot be re-elected immediately after their term expires – do not seem to be affected at all by judicial presence. This indicates that what is driving the result is a higher perceived probability of detection, which disciplines an incumbent with re-election incentives (since this would signal his quality during an election); and not the probability of prosecution, which would also discipline the incumbent without re-election incentives.

So should there be judges and prosecutors in every county?

Since the effect they estimate is likely close to the average treatment effect due to the large first stage, small and medium sized counties currently without judicial presence would likely see a reduction in irregularities if the court came to town. This yields the policy perspective that as long as the net benefit is positive, judicial presence should be scaled up. The authors note that more work is needed to make a policy recommendation because there is still the open question of quantifying the benefit of judicial presence in monetary terms. Last but not least, they point out that their results may be interpreted as qualitative evidence that helps explain why state district attorneys in the US are present in every county, which was not historically the case.