Since the early eighteenth century, scholars have suggested that there is a trade-off between political representation and government efficiency. On the one hand, larger legislatures offer greater opportunities for ethnic and political minorities to elect their candidates, which can democratise access to public goods. On the other hand, adding representatives to a legislature may exacerbate collective action problems, thus hindering service provision. Thus, what is the actual effect of larger legislatures on public service delivery? Do legislators improve public services or are additional lawmakers only a burden to taxpayers?
In a recent study, titled “Legislature Size and Welfare: Evidence from Brazil”, Danilo Freire and his colleagues tackle this question using an exogenous change in the number of municipal legislators in Brazilian cities. In March 2004, the Brazilian Electoral Court established a series of population cut-offs for all city councils in the country. The number of seats was allocated as follows. Each city started with nine councillors, then it should add one legislator for every 47,619 inhabitants until their councils reached 21 members. Conversely, cities between one and four million inhabitants should have at least 33 councillors, and this number increases up to 41 members following the previous formula of one additional legislator per 47,619 inhabitants.
This set of electoral rules makes Brazil the ideal case to test whether one additional legislator has a positive or negative impact on public services. Although an endogenous decision motivated the Electoral Court ruling, the population cut-offs created sharp discontinuities in the distribution of local council seats. Cities close to the cut-offs had very similar characteristics, but those just above the threshold gained a new legislator. Moreover, cities could not self-select into any group, as their population estimates were calculated using the 2003 projections by the Brazilian Census Bureau (IBGE). This change in council size also happened only seven months before the election, making it impossible for candidates to adjust their service provision strategies for the 2004 election. Therefore, the ruling allowed the authors to study the effect of larger legislatures on public service provision during the 2005-2008 term while holding all else constant.
Danilo and his co-authors find that the increase in legislature size significantly improves municipal education and healthcare. Increasing the city council size by one legislator lowers infant mortality by 2.01 children per one thousand infants born and reduces postneonatal mortality by 0.90 children per one thousand infants who survived their first 28 days. Larger councils also increase enrolment by 2.58 children in elementary education classrooms without reducing school quality. These are substantial figures for a developing country.
The authors also provide further evidence to test the validity of their results. First, they created a novel dataset with 346,553 laws passed in 63 municipalities within ten thousand inhabitants away from the cut-offs. They find that while all municipalities mention public goods in their legislation, those with larger city councils had 15 per cent more proposals about public service delivery. Second, the authors surveyed 174 former councillors who served during the 2005-2008 term, the period they analyse in the article. The councillors confirmed that mayors indeed used bureaucratic appointments and political favours to secure legislative support, which is in line with the other findings reported in the paper.
But what drives this improvement in public service provision? The answer is political partisanship. In Brazil, larger city councils changed the composition of mayoral coalitions and the number of appointed bureaucrats in the municipality. The extra city councillor had a 91 per cent chance of belonging to the mayor’s pre-electoral coalition, which allowed mayors to reduce bargaining costs and boost public investments with fewer constraints. Also, Brazil adopts a unique formula to allocate seats and it strongly favours larger parties. As a result, small parties often join electoral coalitions to improve their chances of winning legislative seats. As mayors are usually members of the largest political coalitions, candidates from small parties have a strong incentive to side with the mayor before and after the election. This explains the additional support mayors receive from the newly-elected city councillors.
Furthermore, each additional legislator leads to 104 politically appointed bureaucrats in their municipalities. In Brazil, a higher number of political appointees is associated with better service provision, as their tenure in office depends directly on the survival of the politician supporting them. Thus, political appointees tend to be more productive than career bureaucrats, which also improves social welfare.
The results described in the article have implications that extend well beyond Brazil. Several countries have separate executive and legislative powers, such as the federal level of presidential systems, state governments, and local city councils. As these institutions are influenced by political allegiances, the dynamics exposed in the article be generalisable. In particular, those results may also help scholars understand why other countries in Latin America and Asia can generate effective governance despite having strong executives and large coalitions.
Finally, the article opens several questions for future analysis. First, future research could evaluate whether additional legislators help mayors gain access to federal or state transfers in countries where resources are centrally provided, such as Japan. Although the article shows that city councillors change the dynamics of local spending, it is unclear whether this is true in unitary governments. Second, scholars might want to investigate whether those findings remain valid under different conditions. It is possible that other electoral characteristics can reduce or reverse the impact of additional legislators on welfare. Lastly, we still do not know how other legislative features affect citizen well-being. For example, city councils may vary in size, monetary compensation for councillors, committee structure, and internal power structure. Understanding how legislature size interacts with other features would improve our knowledge about how local and national legislatures provide public goods. These questions are crucial for institution design and welfare in both developed and developing democracies.
The article is available at https://osf.io/x97gf and is forthcoming in the American Journal of Political Science.
Senior Lecturer in Criminology