This thesis consists of three stand-alone chapters studying theoretical models concerning a range of issues that take place within the context of political delegation: tax enforcement, political selection, electoral campaigning. First chapter studies the problem of a small electorate of workers who cannot influence tax rates but can influence their local politicians to interfere with tax enforcement. It develops a two-candidate Downsian voting model where voters are productivity-heterogenous workers who supply labour to a local firm that can engage in costly tax evasion while facing an exogenously given payroll tax collected at the firm level. Two purely office motivated local politicians compete in a winner-takes-all election by offering fine reductions to take place if the firm gets caught evading. Two results stand out. First, equilibrium tax evasion is (weakly) increasing in the productivity of the median voter as a result of the latter demanding a weaker enforcement regime through more aggressive fine reductions. Second, if politicians were able to propose and commit on tax rates as well, then the enforcement process would be interference-free and the tax level would coincide with the median voter's optimal level. These two results underline the fact that from voters' perspective, influencing enforcement policy is an imperfect substitute for influencing tax policy in achieving an optimal redistribution scheme due to tax evasion being costly. In other words, a lax enforcement pattern in a given polity can be indicative of a political demand arising as an attempt to attain a redistributive second-best when influencing tax policy is not a possibility. Second chapter turns attention to the role and incentives of media in the context of ex ante political selection, i.e. at the electoral participation level. It constructs a signalling model with pure adverse selection where a candidate whose quality is private information decides on whether to challenge an incumbent whose quality is common knowledge given an electorate composed of voters who are solely interested in electing the best politician. Electoral participation is costly and before the election, a benevolent media outlet which is assumed to be acting in the best interest of voters decides on whether to undertake a costly investigation that may or may not reveal challenger's quality and transmit this information to voters. The focus of the chapter is on studying the selection and incentive effects of changes in media's information technology. The setting creates a strategic interaction between challenger entry and media activity, which gives rise to two main results. First, an improvement in media's information technology, whether due to cost reductions or gains in investigative strength always (weakly) improves ex ante selection by increasing minimum challenger quality in equilibrium. Second, while lower information costs always (weakly) make the media more active, an higher media strength may reduce its journalistic activity, especially if it is already strong. The intuition behind this asymmetry is simple. While both types of improvements increase media's expected net benefits from journalism, a boost to its investigative strength also makes the media more threatening for inferior challengers at a given level of journalistic activity. Combining this with the first result implies that the media can afford being more passive without undermining selection if it is sufficiently strong to begin with. In short, a strong media might lead to a relatively passive media, even though the media is "working as intended". Third chapter is about electoral campaigns. More precisely, it is a theoretical investigation into one possible audience-related cause for diverging campaign structures of different candidates competing for the same office: state of political knowledge in an electorate. Electorate is assumed to consist of a continuum of voters heterogenous along two dimensions: policy preferences and political knowledge. The latter is assumed to partition the set of voters into ignorant and informed segments, with the former consisting of voters who are unable to condition their voting decisions on the policy dimension. Political competition takes place within a probabilistic voting setting with two candidates, but instead of costless policy proposals as in a standard probabilistic voting model, it revolves around campaigning. Electoral campaigning is modelled as a limited resource allocation problem between two activities: policy campaigning and valence campaigning. The former permits candidates to relocate from their initial policy positions (reputations or legacies), which are assumed to be at the opposing segments of the policy space (i.e. left and right). The latter allows them to generate universal support via a partisanship effect and can be interpreted as an investment into non-policy campaign content such as impressionistic advertising, recruitment of writers capable of producing emotionally appealing speeches, etc. The chapter has two central results. First, a candidate's resource allocation to valence campaigning increases with the fraction of ignorant voters, ideological (non-policy) heterogeneity of informed voters and proximity of candidate's initial position to the bliss point of the informed pseudo-swing voter. The last one results from decreasing relative marginal returns for politicians from converging to pseudo- swing voter's ideal position. Second, even if candidates are otherwise symmetric, a monotonic association between policy preferences and political knowledge can induce divergence into campaign structures. For instance, if ignorance and policy preferences are positively correlated (e.g. less educated preferring more public good) then the left candidate would conduct a campaign with a heavier valence focus and vice versa. Underlying this result is again the decreasing relative marginal returns argument: a candidate whose initial position is already close to that of the informed pseudo-swing voter would benefit more from a valence oriented campaign. An implication of this is that a party that is known having a relatively more ignorant voter base can end up conducting a much more policy focused campaign compared to a party that is largely associated with politically aware voters.

Prior research has shown that the parties in the Swedish Riksdag demonstrate differences in party-culture, constituted by collective self-images and norms. The aim of this study is to investigate if politicians in the Swedish municipalities, from seven different parties, express the same party-culture as their counterparts in the Riksdag previously have. In addition, the study seeks to find out if adherence to the respective party-cultures have implications for the selection of politicians to top positions. Three themes of party-culture is studied, namely political ability, feelings of political responsibility and the party’s strength in relation to individual party members. The themes are investigated by studying a selection of data from the survey KOLFU 2012, which was addressed to all politicians in Sweden’s municipal and county councils. These data were studied mainly through regression analysis. The results demonstrate that there appears to be accordance between the party-cultures on the municipal level and the national level in several respects, but that adherence to party-culture does not generally matter for political selection to top positions.

Electoral Rules and Leader Selection: Experimental Evidence from Ugandan Community Groups. Despite a large body of work documenting how electoral systems affect policy outcomes, less is known about their impact on leader selection. We study this by comparing two types of participatory decision making in Ugandan community groups: (i) vote by secret ballot and (ii) open discussion with consensus. Random assignment allows us to estimate the causal impact of the rules on leader types and social service delivery. Vote groups are found to elect leaders more similar to the average member while discussion group leaders are positively selected on socio-economic characteristics. Further, dropout rates are significantly higher in discussion groups, particularly for poorer members. After 3.5 years, vote groups are larger in size and their members save less and get smaller loans. We conclude that the secret ballot vote creates more inclusive groups while open discussion groups favor the already economically successful. Preparing for Genocide: Community Meetings in Rwanda. How do political elites prepare the civilian population for participation in violent conflict? We empirically investigate this question using data from the Rwandan Genocide in 1994. Every Saturday before 1994, Rwandan villagers had to meet to work on community infrastructure. The practice was highly politicized and, according to anecdotal evidence, regularly used by the political elites for spreading propaganda in the years before the genocide. This paper presents the first quantitative evidence of this abuse of the community meetings. To establish causality, we exploit cross-sectional variation in meeting intensity induced by exogenous weather fluctuations. We find that an additional rainy Saturday resulted in a five percent lower civilian participation rate in genocide violence. Selection into Borrowing: Survey Evidence from Uganda. In this paper, I study how changes to the standard credit contract affect loan demand and selection into borrowing, using a representative sample of urban micro enterprises, most with no borrowing experience. Hypothetical loan demand questions are used to test whether firm owners respond to changes in loans' contractual terms and whether take-up varies by firms' risk type and other firm owner characteristics. The results indicate that contracts with lower interest rates and less stringent collateral requirements attract less risky borrowers, suggesting that there is scope for improvement of standard financial contract terms. Credit Contract Structure and Firm Growth: Evidence from a Randomized Control Trial. We study the effects of credit contract structure on firm outcomes among small and medium sized firms. A randomized control trial was carried out to distinguish between some of the key constraints to efficient credit use connected to the firms' business environment and production function, namely (i) backloaded returns (ii) uncertain returns and (iii) indivisible fixed costs. Each firm was followed for the 1-year loan cycle. We describe the experiment and present preliminary results from the first 754 out of 2,340 firms to have completed the loan cycle. Firms offered a grace period have higher profits and higher household income than firms receiving a rebate later on as well as the control group. They also increased the number of paid employees  and reduced the number of unpaid employees, an effect also found among firms that received a cash subsidy at the beginning of the loan cycle. We discuss potential mechanisms behind these effects.

L’objectif de cette thèse est d’introduire de manière explicite les caractéristiques personnelles des décideurs publics dans l’analyse de processus politiques français. Trois cas sont successivement analysés, soulevant chacun une problématique distincte. Le premier chapitre s’intéresse à l’échelon municipal, et se base sur un jeu de données original comportant des informations sur l’ensemble des maires des municipalités de plus de 10000 habitants entre 2000 et 2012. L’objectif est de décrire comment l’influence idiosyncratique d’un maire sur la politique d’investissement municipal impacte sa probabilité de réélection. Les résultats indiquent que plus la taille de la municipalité augmente, moins les électeurs se basent sur ce type d’information. Le second chapitre a pour objet la production législative française, et s’appuie également sur une base de données originale. Un double cycle de production législative émerge, généré par les élections présidentielles et législatives. Il apparaît également que les caractéristiques personnelles des ministres influencent la stratégie du gouvernement, notamment l’âge et l’expérience. Enfin, le troisième chapitre se focalise sur l’impact de la compétition électorale sur le processus de recrutement politique. Un important travail de collecte de données concernant la production parlementaire de chaque député de la Ve République permet d’étudier cette relation ainsi que son évolution au cours de la période 1959-2012. Il en ressort que les députés élus dans des circonscriptions compétitives ont une activité parlementaire plus importante, toutes choses égales par ailleurs. Cependant, ce lien entre activité et compétition est en constante diminution depuis les années 1980. / The aim of this thesis is to explicitly introduce the decision-maker into the empirical analysis of different political processes in the French context. Three cases are successively analyzed,each raising a specific problematic. The first chapter focuses on French municipalities. A new original dataset, covering mayors of municipalities of more than 10,000 inhabitants over the period 2000-2012, underpins the study. The objective is to investigate to what extentthe mayor’s personal influence on the investment policy affects his/her reelection probability. Results show that this information plays a significant role only in small municipalities. Chapter twostudies the French legislative production, exploiting another original dataset. A dual cycle of the production of laws emerges, connected to both the presidential and the legislative elections. A link between government members’ personal characteristics and legislative output is established. Finally, chapter three investigates the relationship between electoral competition and political selection. A third important dataset providing data about the individual parliamentary work of the deputies over the period 1958-2012 allows such an analysis. Resultsindicate that deputies elected in contested districts have a higher overall productivity. The intensity of the relationship reached its peak in the 1980’s, but is continuously decreasing since then.