Oxford Research Proposal Example: Antitrust economist to competition enforcement (Score 93)

Merger control rests on a deceptively simple premise: authorities assess whether a transaction substantially lessens competition and intervene when it does. In practice, enforcement outcomes diverge sharply across jurisdictions and over time within the same jurisdiction, in ways that industrial organisation models of harm alone cannot explain. Remedies are negotiated under political pressure, phase-two investigations are abandoned before formal prohibition, and market definition decisions appear sensitive to the identity of the merging parties rather than to underlying economic facts. This proposal asks: to what extent do political economy factors — specifically, the degree of agency independence, the structure of political accountability, and the intensity of lobbying by merging parties — explain cross-jurisdictional and intertemporal variation in merger enforcement outcomes, after controlling for case-level economic characteristics? Two subsidiary questions follow. First, does the effect of political constraints differ between horizontal mergers in concentrated markets and vertical or conglomerate transactions, where the economic case for intervention is more contested? Second, do enforcement gaps widen during electoral cycles, consistent with a political business cycle mechanism operating through competition policy? The rationale is not that enforcement is uniformly captured but that the standard account, which treats decisions as outputs of a welfare-maximising agency, systematically under-predicts the variance in observed outcomes. A framework treating the agency as an actor embedded in a principal-agent chain from legislature to executive to regulator offers a more complete positive model and has direct implications for institutional design. Two bodies of work bear on this question without yet being integrated at the level of empirical identification. The industrial organisation literature on merger control — in the tradition of Werden, Farrell and Shapiro, and the post-Chicago school — provides the welfare benchmark and tools for measuring competitive harm: upward pricing pressure indices, diversion ratios, and simulation models. This literature treats enforcement decisions as approximately correct on average, with errors attributable to information asymmetry or methodological disagreement rather than political interference. It is strong on the economics of harm but largely silent on the institutional environment in which assessments are made. The political economy of regulation literature — drawing on Stigler's capture theory, Peltzman's extension, and more recent work on bureaucratic independence by Gilardi and Maggetti — provides a framework for understanding how agencies respond to political principals. Applied to competition policy, this literature has produced qualitative case studies and cross-country independence indices, but has rarely connected those indices to case-level enforcement outcomes while controlling for the economic characteristics of the merger. The result is a gap: good theory about why political constraints should affect enforcement, and good tools for measuring competitive harm, but no systematic empirical study combining both to identify the independent effect of political economy variables on enforcement decisions. A smaller adjacent literature on lobbying and competition policy — including work by Ades and Di Tella on corruption and competition, and empirical studies of revolving-door effects in US antitrust — suggests the mechanism runs partly through the quality of the agency's information set and partly through direct pressure on remedies. This proposal treats lobbying intensity as a mediating variable rather than a competing explanation. The gap addressed is therefore methodological as much as substantive: the absence of a study using a merged dataset of case-level economic characteristics and agency-level political economy variables to estimate the enforcement gap attributable to political constraints, with identification from variation in electoral timing and changes in agency governance rules. The study proceeds in three phases. Phase one constructs a case-level dataset covering merger decisions by the European Commission, the UK Competition and Markets Authority, and the German Bundeskartellamt over a fifteen-year window. For each decision, I code the enforcement outcome (unconditional clearance, clearance with remedies, prohibition), investigation phase, reported market concentration, sector, and identity of the notifying parties. Decision documents are publicly available from each authority's case register, yielding an estimated sample of 1,200 to 1,500 decisions. Phase two links case-level data to authority-level political economy variables: a formal agency independence index derived from founding statutes and amendments, government ideology at the time of decision, electoral cycle position, and a lobbying intensity proxy constructed from third-party submissions and, where available, disclosed lobbying expenditure. The independence index draws on existing comparative datasets, extended where coverage is incomplete. Phase three estimates ordered probit models in which the enforcement outcome is the dependent variable, economic harm proxies are controls, and political economy variables are the regressors of interest. The key identification challenge is that political economy variables correlate with unobserved authority-level characteristics. I address this with authority fixed effects, absorbing time-invariant characteristics, and an instrumental variable strategy exploiting exogenous variation in electoral timing plausibly unrelated to the economic characteristics of mergers notified in any given quarter. For the electoral cycle hypothesis, I test whether the probability of a phase-two investigation or prohibition falls in the twelve months preceding a general election, conditional on the economic harm proxy. This is a falsifiable prediction requiring no strong prior about the direction of political pressure. All primary data sources are publicly available administrative records. No personal data are involved, and no ethics approval beyond standard departmental registration is anticipated. The main feasibility risk is coding reliability: outcome classification requires reading decision summaries, and concentration measures are not always standardised. I will develop a coding protocol in the first term, pilot it on fifty decisions, and calculate inter-rater reliability before full coding. A second risk is sample attrition: some decisions are redacted, and the lobbying proxy will be incomplete for earlier years. Robustness checks will restrict the sample to decisions with full covariate coverage and test whether attrition correlates with the political economy variables. Provisional timeline: term one, literature review and coding protocol; term two, data collection and cleaning; term three, estimation and first draft. The scope is calibrated to a one-year MSc dissertation; the ambition is a clean identification of the electoral cycle effect rather than a comprehensive theory of regulatory capture. Oxford's Department of Economics hosts research in industrial organisation and political economy that maps directly onto both components of this proposal. The department's public research profile includes work on market power, regulatory economics, and the political economy of institutions, and the wider university environment — including the Blavatnik School of Government and research on regulatory governance — provides broader intellectual context for the institutional design questions the project raises. The data infrastructure required is modest: access to publicly available case registers, standard econometric software, and library access to the comparative regulation literature. No proprietary datasets or fieldwork are required, making the project feasible within the resource constraints of a one-year programme. I have not approached any faculty member regarding supervision. My expectation is that the admissions process will determine whether the proposal falls within current supervisory capacity. The methodological approach — applied microeconometrics with a political economy identification strategy — is consistent with the department's published research methods, and the question is tractable within the dissertation word limit and timeline of the MSc Economics programme.