Smaller Measurement Errors, Better Economic Policies?
Johannes Binswanger (University of St. Gallen)
Manuel Oechslin (University of Lucerne)

Abstract : More and more economic transactions leave a "digital footprint", a trend that will improve the precision with which key economic indicators, such as GDP, inflation, or unemployment, can be estimated. We analyze the consequences of this trend for economic policy and performance in a political-agency model that includes fundamental uncertainty about the optimal design of growth-promoting reforms. We demonstrate that more precise economic statistics can inhibit--rather than stimulate--reform attempts. By improving the assessment of ongoing reform processes, better statistics more clearly expose reform designs that do not work and need realignment. For the incumbent government, this implies a higher risk of losing power in an upcoming election: when confronted with negative news on the economy, voters are less likely to give the government the benefit of the doubt. As a result, the government may prefer to play it safe and adhere to the less risky status quo. A basic implication of our model is that the effect of better statistics on economic performance is not unambiguously beneficial but depends on institutional factors, such as the degree to which politicians are held accountable to the voters.

Public Good Under Appointed Versus Elected Mayors: Policing and Crime in Belgium
Andrea Colombo (ECARES, ULB)
Ilan Tojerow (ULB, CEB, DULBEA, IZA)

Abstract : This paper studies how the method by which mayors gain their positions impacts the provision of local public goods and, most specifically, policing. Our identification strategy exploits a natural experiment provided by the introduction in 2005 of direct election to mayorship in only one region of Belgium, Wallonia; while mayors from elsewhere are still appointed by the locally elected City Councils. Particularly, we compare crime incidence under mayors that are directly elected by voters and those that are appointed by an elected body. Conducting a difference-in-difference analysis with a rich dataset registering locally-reported crimes from 2000 to 2012, our results show a post-reform decrease in overall crime from 5% to 8%, depending on the specification. We provide evidence that the reallocation of efforts towards fighting specific types of crimes by directly-elected mayors drives this effect, rather than a general increase in police efficiency. Moreover, the “increased accountability” effect for each directly-elected mayor di- lutes when the management of local police has to be coordinated with other neighboring peers.

Political Cognitive Biases Effects on Fund Managers' Performance
Marian Moszoro (George Mason University)
Michael Bykhovsky (Columbia Engineering BOV)

Abstract : Who does a better job of managing money: Republicans or Democrats? Under rational agent hypothesis, financial industry practitioners should not be affected by political discourse, and investors cannot realize abnormal returns on publicly available information. Rare events, however, may silence rationality and potentiate cognitive dissonance on a spectrum of agents. We assembled a comprehensive dataset of equity hedge funds performance and matched the managers' political affiliation by their partisan contributions. We document higher returns of funds managed by Democrats for 10 subsequent months---from December 2008 to September 2009. This result is unique and robust to placebo time windows and random partisan affiliation shuffling. We conjecture that the conjunction of the financial crisis, Obama's election, and politically polarized interpretation of the US central bank policy during that period had an asymmetric impact on hedge fund managers' perception. In other periods, when the political discourse did not involve central bank policy, there was no statistically significant difference in fund managers’ performance depending on their political beliefs.

Separation of Powers and New Public Spending
Kaj Thomsson (Maastricht University)
Alessandro Bonatti (MIT)

Abstract : We develop a model of American legislative politics, and we use this model to explain the shape of new public spending programs. In the model, the distribution of federal funds across regions of the country is the outcome of a bargaining game in which the President acts as the agenda-setter and Congress bargains over the final shape of the spending bill. The model highlights the importance of the American president being independently elected, which leads to a more balanced spending outcome than if proposal power was allocated to some member of Congress. In addition, we use our model to demonstrate the importance of the institutional rules within Congress, in particular its bicameral structure and the sequentiality of the legislative process. We then apply the model to the New Deal; in this context the model suggests, among other things, that both economic and political concerns were behind the shape of the New Deal spending, and that a less politically minded President would have been less constrained by Congress.