Enlightened Pork: Special Interest Politics and Nighttime Light Emission (with Roland Hodler, University of St. Gallen and Paul Raschky, Monash University)
Every budget period legislators negotiate over the allocation of funds to government programs. The programs might target the overall population or specific industries, societal or geographic groups, etc. In political economics it is well established that many programs extensively benefit well-organized groups while the cost are disbursed all over society. While there is an extensive theoretical literature it proved to be very difficult to quantify the extent of such “pork barrel” spending because any spending might be associated with a well-founded and pressing need. In our investigation we define pork barrel spending as regional favoritism and try to identify geographically targeted transfers that only arise because the jurisdiction’s representative is member of a relevant spending committee. The US congress is organized in numerous specific committees that hold proposal power over spending programs within their jurisdiction. Committees have extensive control over the allocation of public funds and membership in committees is mainly governed by seniority. The seniority principle makes it less likely that member selection is primarily based on objective needs. We aim at identifying the influence of congressional committee membership on transfers to the federal electoral districts of its members. Due to the number of budgetary items and complexity of spending programs one major difficulty is to quantify the overall amount of transfers and benefits to specific electoral districts. Because there is no standard information on economic activity (e.g. GDP) at the district level we resort to a recently advanced alternative measure: satellite recordings of nighttime light emission. This measure of human activity is strongly correlated with GDP measures, while it includes activities beyond the pure economic sphere (including potential activity due to transfers) and can be precisely calculated according to district boundaries. Our identification strategy relies on discrete changes of membership in relevant spending committees and its hypothesized changes in nighttime light emission. To make causal inference we take advantage of the regular redistricting procedures following the decennial census, which reallocates geographical areas from one district to another.

The size and source of the incumbency advantage: A regression discontinuity approach (with Lukas Schmid, University of St. Gallen)
The electoral advantage of incumbent policymakers is well established. The literature, however, mainly focuses on the US electoral system. It is characterized by majority voting and a resulting two party system and the incumbency advantage amounts to an electoral margin of roughly 8 to 10 percentage points. We extend the analysis to the Swiss electoral system with its proportional electoral rule for most elections to the Swiss National Council. We propose a new causal identification strategy by developing a regression discontinuity approach for proportional electoral systems. We compare the electoral results (at time t) of candidates who in the previous election (at time t-1) were elected with only a narrow margin, with candidates who lost by that same narrow margin. Focusing on such close elections introduces quasi-random variation that can be exploited for causal inference. The difference to the common regression discontinuity design (RDD) for majority systems is that we can exploit two sources of quasi-random variation. First, proportional systems need precise rules to allocate seats to parties based on electoral support (Hagenbach-Bischoff), where continuous vote shares are translated into discrete parliamentary seats. Some parties narrowly gain an additional seat while others narrowly lose. Second, the within party allocation of seats is according to the absolute vote share of the individual candidates on the party list. One candidate will be assigned the marginal seat while the following candidate misses out. We concentrate again on narrow margins for the candidate assignment to the marginal seat and use this information as our second source of quasi-random variation. With this procedure we can eliminate the influence of individual characteristics and causally estimate the incumbency advantage for elections to the Swiss National Council. It is possible that the size of the incumbency advantage varies with e.g. institutional, economic or demographic characteristics of the respective canton. Hence, we should not only be capable of determining the size of the incumbency advantage in different cantons, but also its source.

Institutions, Regulation and Taxation (with Simon Lüchinger, University of Lucerne and Andrei Shleifer, Harvard University)  
How do political institutions affect the emergence and evolution of regulation? We approach the question in two ways: First, we establish the basic link between political institutions and regulation. Secondly, we acknowledge the well-known fact that regulatory measures are often functionally equivalent to fiscal policy measures. Hence, we take into account that regulatory and fiscal decisions are not independent and shed light on the influence of political institutions on the interplay of regulation and taxation. We study the influence of the electoral regime, political ideology and limits to legislative output (popular referendum). We collected data on regulatory activity of the Swiss cantons from 1908 to 2010 and exploit this long-term panel data by using difference-in-differences strategies and instrumental variable techniques.