Changing local political boundaries by splitting local governments to create more localized governance is a common, yet contested reform pursued worldwide. Proponents argue smaller local government units can improve services, boost local economic activity, tailor policy to local preferences, and achieve greater accountability as they are easier to monitor for citizens. Critics counter that splitting produces local government units that are too small to effectively deliver services, leaves them vulnerable to capture by special interests, and creates localities that are too small to finance themselves without subsidies from the rest of the country.
A new researcher paper, Decentralizing Development: The Economic Impacts of Government Splits by Ricardo Dahis (Monash University) and Christiane Szerman (LSE), published in the American Economic Journal: Applied Economics, examines this trade-off using one of the largest within-country splitting episodes on record (in Brazil in the 1990s), tracing the short- and long-run effects of creating new municipalities on public services, the labor market, economic activity, and fiscal performance.
The Brazilian Setting
Brazil’s municipalities hold substantial administrative, fiscal, and political power, and until the mid-1990s the rules for creating new ones were unusually lenient. Generous federal subsidies and permissive redistricting regulations led the number of municipalities to jump 34 percent, from 4,124 to 5,507, between 1988 and 1996, as districts previously lacking decision-making power split off to become municipalities in their own right. Concerned about this pace, Brazil’s Congress passed a 1996 constitutional amendment that effectively halted new splits by centralizing approval and requiring evidence of fiscal sustainability.
This makes an unusually good setting for evaluating decentralization, since municipalities carry real decision-making authority and comprehensive data exist on public services, the labor market, fiscal performance, and satellite-recorded nighttime light as a proxy for economic activity.
Isolating the Effect of Splitting
Splits are rarely random. Applicant districts tend to be less developed and feel neglected by their parent municipalities, and become eligible for larger federal transfers once independent. Municipalities with an applicant district are also larger in population and area, and receive a smaller share of federal transfers, than those where no district ever applied. Comparing split municipalities to those that never applied would confound splitting’s effect with these pre-existing differences.
To address this, the authors built a new dataset of split requests, including those never approved, from historical archives across 11 states. Their research design compares municipalities whose requests were ratified to “almost split” municipalities whose requests failed, typically for reasons unrelated to local economic conditions (political vetoes, failed referenda, or being caught by the 1996 reform before the process could finish). Both groups show similar pre-existing trends, supporting a causal interpretation. As a further check, a difference-in-discontinuities design in Minas Gerais compares districts that narrowly passed the referendum threshold to those that narrowly failed, with qualitatively similar results.
What Splitting Delivers
The study’s results suggest that as a result of splitting, both bureaucracy and public services expand. New municipalities increase capital expenditures (machinery, buildings) by 27 percent and current expenditures (payroll, administration) by 17 percent, financing an expansion of local bureaucracy: municipal public jobs rise by around 16 percent, with no offsetting decline in state or federal public employment. Service gains follow: trash collection access rises 4.4 percentage points and sewage access one point, concentrated in services municipalities control exclusively, with no equivalent gains in shared services.
Results further suggest that younger cohorts benefit most. Since education is a service that Brazilian municipalities provide on their own, individual-level census data show younger, more exposed cohorts see the largest gains in school attendance (2 to 5 percentage points) and literacy (up to 4 points), consistent with new investment in schooling after splitting.
The private sector responds unevenly. Using matched employer-employee records, the authors find no significant aggregate expansion of private-sector jobs or establishments, though new establishments emerge disproportionately in the service sector. Satellite nighttime-light data, capturing the informal economy alongside the public and private sectors, show a clearer picture: luminosity rises quickly in the first five years and stabilizes at an 8 percent increase in the long run.
The gains are concentrated in the districts that asked to split. Disaggregating to the district level, applicant districts see luminosity rise by roughly 40 percent 15 years on; headquarters districts (the original town-hall seat) see a much smaller 6 percent gain; and districts that neither applied nor split see a negative, statistically insignificant change. This growth is not confined to the new town hall’s vicinity; it is widespread across the applicant district.
Why Does Splitting Work? Money, Autonomy, or Both?
New municipalities see a 15 percent rise in fiscal revenues, driven mainly by higher federal transfers under Brazil’s population-based formula, which favors smaller jurisdictions. But the authors show this is only part of the story: holding revenues constant reduces, but does not eliminate, the estimated effects, pointing to a second channel, decentralized decision-making itself. Supporting this, service gains concentrate in activities under exclusive local control; gains are largest in districts farther from the parent town hall and less urbanized at baseline, consistent with autonomy helping the most neglected areas; new municipalities frequently elect mayors from different parties than their parent, suggesting a better fit with local preferences; and alignment with the state governor does not predict the results.
Does Splitting Come at Someone Else’s Expense?
Because new municipalities draw a larger share of a fixed pool of federal transfers, non-splitting municipalities in the same state see their transfer share fall by 13.7 percent on average, even as splitting municipalities gain 20.3 percent. Yet the authors find no evidence this reallocation harms non-splitting municipalities: estimates for public jobs, private jobs, establishments, and luminosity are statistically indistinguishable from zero. One interpretation is that the lost transfers may have funded lower-value spending, so their loss carries no measurable economic cost.
The authors also quantify cost-effectiveness: each public job created through splitting costs around US$3,635 per year in federal transfers, well below the roughly US$8,000 per job estimated for federal transfers alone. The implied output multiplier ranges from 2.06 to 4.34, above the 1.9 median in the fiscal-spending literature, and in line with the 1.1 to 2.6 range estimated for federal transfers alone.
Policy Implications
The paper reaches cautious conclusions rather than broad generalizations. It does not identify an optimal size for local government but suggests that in settings with unusually large administrative units, like Brazil, where the average municipality in 1988 covered over 2,000 square kilometres (far larger than comparable units in Italy, the US, or India), subsidized, voluntary splits can generate meaningful development gains. These gains are driven by a combination of greater fiscal resources and more responsive local decision-making, with the largest benefits accruing to peripheral, remote, and historically neglected districts. Importantly, the study finds no measurable evidence that these gains came at the expense of municipalities that did not split.
The authors caution that these findings should not be interpreted as an endorsement of splitting in all contexts. While the evidence points to substantial economic and service delivery benefits, the study cannot assess longer-term governance risks. Other research suggests that revenue windfalls may weaken accountability, increase corruption, or reduce the quality of political leadership. Understanding whether such trade-offs emerge after government splits, and how new local governments evolve over time, remains an important priority for future research.
Read the full paper:
Dahis, Ricardo, and Christiane Szerman. 2026. “Decentralizing Development: The Economic Impacts of Government Splits.” American Economic Journal: Applied Economics 18 (3): 220–253.
An pre-publication version of the study is available here (open access).
Note: The Feature Image for this blog post was generated using AI.
