Four essays in applied economics

Abstract

The four papers constituting this thesis cover a range of different topics, thus reflecting the breadth of issues I have worked on in the course of my PhD. All seek to rely as little as possible on prior assumptions, and to approximate as far as the context allows the ideal of a fully randomized experiment. The papers are listed in the chronological order in which I have worked on them, and are now briefly discussed in turn. Chapters 1 and 2, which are joint work with Frank Betz, investigate the extent to which the Protestant Reformation has influenced respectively preferences for leisure (chapter 1) and preferences for redistribution and state intervention (chapter 2) in present-day Switzerland. Both focus on a region in the South-West of Switzerland, part of which was forced in the 17th century to convert to Protestantism whereas the other part was to remain Catholic. 15th century population figures along with a wide range of qualitative evidence allow us to demonstrate the equality of the two regions before the event, and hence the causal nature of the present-day differences found. By complementing our Instrumental-Variable analysis with a Spatial Regression Discontinuity Design that focuses on those municipalities situated nearby the historical religious border, we are furthermore able to address potential concerns about remaining confounders in the domains of geography, proximity to France, and proximity to the nearby city of Geneva. The first paper, on preferences for leisure, speaks to the very active recent literature that seeks to empirically test Weber [1904]'s famous hypothesis whereby the Protestant Reformation has led to a more pronounced "work ethic" and thereby increased economic prosperity. It stands out from prior work in the literature for employing direct measures of "work ethic" as well as for focusing specifically on those "Ascetic" strands of Protestantism which Weber [1904]'s hypothesis is based on. We find that traces of a "Protestant work ethic" can be found until the present-day, in attitudes as well as actual working behavior, but we do not find significant effects on present-day income per capita. Looking beyond the Weber debate, this evidence is interesting also as an example of how a historic event, through the channel of culture, can have effects that persist until the present day. The second paper moves from self-regarding to political preferences, analyzing attitudes on the extent to which the government should intervene in the market, and the extent to which it should redistribute the returns generated therein. We confirm the hypotheses implied by the works of the sociologists Esping-Andersen [1990] and Manow [2002] whereby ceteris paribus "Ascetic Protestantism" has led to weaker preferences for redistribution and government intervention than Roman Catholicism. These differences are then found to manifest themselves also in a range of economic outcomes, such as measures of the inequality of pre-tax income within each municipality. They can be seen to refute Marx [1904]’s view whereby existence determines consciousness rather than consciousness existence. Chapters 3 and 4, joint work with Andreas Fagereng and Kjetil Telle, focus on the labor market and household finance, with very direct implications for public policy. Both make use of Norwegian administrative data, which are based on the highly reliable and comprehensive Norwegian tax registers, and cover the universe of Norwegian tax payers. Chapter 3 uses these data to track the paths of income, wealth and holdings in different asset classes around the year of job loss. Panel data techniques and the complementary use of information on plant downsizings allow us to find clear evidence that households start to accumulate additional savings and to reallocate these to less risky and more liquid asset classes from about 2 years before their job loss, subsequently deplete these and further savings to cushion consumption while unemployed, and finally start rebuilding their savings to the initial level, all within +/- 4 years of their job loss. We also find that the extent to which households employ such consumption smoothing strategies is smaller for those with lower initial holdings and for the young, that it does not seem to vary significantly with unemployment spell length, and that it is higher (sic) for those whose income shock is more permanent. The above-mentioned evidence on the importance of prior holdings is evidence in favor of liquidity constraints, but is only tentative due to the likely endogeneity of these prior holdings. This limitation is addressed in the 4th chapter. This investigates the responsiveness of job search duration, which can be interpreted as a form of consumption, to quasi-randomly assigned lump-sum severance payments. Identification is achieved by exploiting discontinuities in the available amounts along the age and time dimensions. We thus find that that each NOK 1,000 of severance pay lowers the propensity to start a new job on any day within the first 24 months of job loss by about 1.4% relative to the group not receiving any severance pay. Such an effect has previously only been investigated and found for Austria, where the generosity of unemployment insurance is low relative to most other OECD economies other than the US. Our finding that such an effect exists even in a country with as high and equitable private wealth and as generous unemployment insurance as Norway suggests that liquidity constraints are likely to be a relevant phenomenon also in many other OECD economies.