This is the second in this year’s series of posts by PhD students on the job market.
Sixteen developing countries have experienced giant oil or gas discoveries since 2010 (Zhang et al., 2019). While some discoveries lead to vast revenue windfalls, many take longer than expected to produce or never produce at all. This uncertainty creates challenges for governance: policymakers may spend, borrow, or engage in rent-seeking in anticipation of resource windfalls. If revenues fail to materialize, policymakers may struggle to adapt, leading to a “Presource Curse.” Governments that do enjoy windfalls may struggle to invest effectively in public goods provision or economic diversification.
In my Job Market Paper, I explore the dynamics of local governance after offshore oil and gas discoveries in Brazil. I exploit quasi-experimental subnational variation in discovery announcements and subsequent production realizations to identify the effects of news and revenue shocks on municipal public finances, public goods provision, and political competition, selection, and patronage. I compare outcomes in discovery-treated places with control municipalities that experience exploratory offshore drilling but no discoveries during the study period, under the assumption that–conditional on drilling–discoveries are as-if random (Cavalcanti et al., 2016).
To measure governance outcomes, I build a municipality-level panel (2000-2017) that includes data on public finances and hiring, elections and campaign donations, public goods provision, and municipality characteristics. For robustness, I re-estimate event study specifications using pre-matched controls. I implement Callaway and Sant’Anna’s (2021) did estimator to address concerns over TWFE with differential treatment timing.
Offshore discoveries and formulaic sharing rules create quasi-experimental variation
Brazil experienced a wave of offshore oil and gas discoveries between 2000-2017. I compile a comprehensive list of 179 discovery announcements filed by oil companies with Brazil’s Securities and Exchange Commission during this period. Formulaic sharing rules allocate revenues from offshore oil and gas production to geographically aligned coastal municipalities, allowing local governments to predict whether they will benefit from a discovery. I reconstruct projections of coastal boundaries (Figure 1) and overlay geo-referenced wells to link discoveries to aligned municipalities.
Discoveries are exogenous to municipalities–they are made by multinational corporations operating hundreds of kilometers offshore, servicing installations from distant ports, and responding to global prices and technologies. Subsequent production realizations are also driven by extra-municipal factors. A fall in global prices can make a promising field commercially unviable; reserves may turn out smaller or more difficult to extract than initially estimated. Companies may leave discoveries undeveloped due to internal financial difficulties or shifts in strategy.
48 Brazilian municipalities were affected by offshore discovery announcements between 2000-2017. To quantify heterogeneity in discovery realizations, I develop a forecasting model based on standard offshore production assumptions, announced reserve volumes, average production delays, and Brazil’s royalty distribution rules. Comparing forecast and realized revenues in each discovery-treated municipality reveals that some places eventually enjoy revenue windfalls in line with expectations, while others are disappointed. I leverage this exogenous variation in forecast error to categorize municipalities into two treatment arms (“satisfied” and “disappointed”). I estimate effects on each separately relative to never-treated controls.
Successful oil discoveries deliver revenue windfalls but limited development benefits
Municipalities do not react immediately to discovery announcements, likely due to constraints imposed by a Fiscal Responsibility Law that limits deficit spending. As production ramps up 5-10 years after a discovery, per capita oil revenues increase dramatically in “satisfied” places that receive windfalls, growing by 5,441% after ten years relative to never-treated controls (Figure 2). This represents an increase from a baseline control mean of R$129 per capita to R$7,121, or from 2% to 109% of baseline average income. Total per capita revenues in satisfied places increase by 75% after ten years (from R$1,084 to R$1,898). Total per capita spending increases 20% (from R$874 to R$1,055), and spending on education and health increases 27%. Despite these increases, public goods provision falls relative to controls following a discovery announcement. This may be due to limited capacity to spend windfalls effectively, leakage of oil rents into corruption, or lags in improving outcomes. Municipalities that receive windfalls do not increase spending to promote non-extractive sectors.
Disappointed municipalities are left worse off than never-treated controls
Oil revenues remain unchanged in municipalities that experience discovery announcements but never enjoy windfalls (by 2017). Nevertheless, per capita revenues in these “disappointed” places decline 27% after ten years relative to never-treated controls (from R$1,084 to R$795), largely due to falling tax revenues (-37%) and federal/state transfers (-9%). Consequently, per capita spending declines 24% (from R$874 to R$664), per capita investment 57% (from R$101 to R$44), and education and health spending 26% (from R$536 to R$397). Public goods provision falls. Spending on promotion of non-extractive sectors is weakly positive after discoveries, suggesting disappointed places increase efforts to diversify their economies when oil discoveries fail to deliver.
Discovery announcements increase electoral competition and reduce schooling of leaders
Since mayors and councilors control oil revenues, the expected value of holding office increases when a discovery is announced. I estimate a difference-in-differences specification where treatment is defined as announcement of an offshore discovery in the four years preceding an election. I find discoveries increase competitive candidates running for council (+4.8%) and decrease schooling levels of candidates and winners (-3%). Discoveries weakly increase number and value of campaign donations. I find no effects on patronage, defined as the rate at which winning mayors hire campaign donors to discretionary public jobs. Incumbents are significantly less likely to be reelected (-5.2%) when oil revenues are below expectations at the time of the election.
Why are disappointed places worse off?
If municipalities did not react preemptively to discovery announcements, why do disappointed places suffer negative outcomes relative to never-treated controls? I propose three mechanisms:
1. Private sector activities relocate to booming places, eroding the tax base
Construction firms are beneficiaries of public infrastructure spending and may relocate from disappointed to booming places. Formal employment in construction falls 25% in disappointed places ten years after a discovery, and local tax revenues–collected on real estate transactions and urban properties–fall 37%.
2. Federal and state transfer revenues fall
Formulaic revenue transfers, which make up 75-95% of municipal budgets, decline by 9% per capita in disappointed municipalities ten years after a discovery. I show transfers pegged to population, school enrollment, and exports account for this decrease, and discuss reasons for each in my paper.
3. Rent-seeking and political turnover bite
Discoveries cause lower-education individuals to run for and win municipal office. These individuals may have lower private-sector opportunity costs and potentially greater interest in rent-seeking (Caselli and Morelli, 2004). Baragwanath (2020) shows oil revenues increase entry of corrupt candidates. Increased political turnover in disappointed places may disrupt public service delivery (Akhtari et al., 2021).
What are the policy takeaways?
At the country-level, rules that concentrate resource revenues in specific places exacerbate uncertainty and make local governance more difficult. Spreading risk over a wider exploration portfolio would smooth idiosyncratic outcomes in individual fields, dilute disappointment, and avoid overwhelming local administrative capacity. Leaders should manage expectations after discoveries and work between levels of government to build capacity in preparation for booms.
At the local level, policymakers in Brazil are developing innovative policies to manage oil revenues, including a universal basic income program and sovereign wealth funds. The rocky beginnings of Brazil’s oil boom are raising awareness of the risks and opportunities that extractive resources create for development.
Erik Katovich is a PhD Candidate in Agricultural and Applied Economics at the University of Wisconsin-Madison. Erik uses large administrative datasets and econometric and geospatial tools to study institutions of natural resource governance, industrialization and industrial policies, and the effects of resource booms and busts on governments, firms, and workers. Learn more about his research at: ekatovich.github.io