This thesis is a collection of three essays. The first two study how ideas spread through a network of individuals, and how it an advertiser can exploit it. In the model I develop, users choose their sources of information based on the perceived usefulness of their sources of information. This contrasts with previous literature where there is no choice made by network users and thus, the information flow is fixed. I provide a complete theoretical characterization of the solution and define a natural measure of influence based on choices of users. I also present an algorithm to solve the model in polynomial time on any network, regardless of the scale or the topology. I also discuss the properties of a network technology from a public economic standpoint. In essence, a network allows the reproduction of ideas for free for the advertiser. If there is any free-riding problem, I show that coalitions of users on the network can solve such problem. I also discuss the social value of networks, a value that cannot be captured for profit. The third essay is completely distinct from the network paradigm and instead studies funding rules for public universities. I show that a funding rule that depends solely on enrolment leads to "competition by franchise" and that such behavior is sometimes inefficient. I suggest instead an alternate funding rule that allows government to increase welfare without increasing spending in universities.

This dissertation is composed of three chapters. All three deal with topics in development economics. The first chapter examines the effects on village institutions of introducing formal financial institution options into the village. The second addresses the effects of government policy on educational investment and crime. The third tests the explanatory power of various explanations of the gender gap in math test scores. The first chapter examines the effects of a transition from a ``traditional'' economy based on an uncertain source of income, with risk fully insured away by one's neighbors in a social network through costly network ties, to a ``modern'' economy in which some agents have access to partial insurance at a lower cost. A theoretical model is used to show that village social networks can break down as some members of the village no longer need the insurance the social network provides, producing a reduction in welfare (if the costs of reducing moral hazard are not too high) for at least some individuals and possibly the village as a whole. This loss of welfare can occur even when networks provide other benefits to those belonging to them and is likely to be heterogeneous, depending on the opportunities and networks available to individuals. This paper tests these predictions using Indonesian data to examine the effect of a change in the banking institutions available to a community on the strength of social networks (measured by community participation) and welfare (measured by household expenditure and by child health). The analysis finds that changing financial institution availability in general does not influence community participation or welfare, but that financial institutions that primarily serve certain groups do relatively reduce the welfare of households not in those groups, which is consistent with the hypotheses generated by the model. Crime is an important feature of economic life in many countries, especially in the developing world. Crime distorts many economic decisions because it acts like an unpredictable tax on earnings. In particular, the threat of crime may influence people's willingness to invest in schooling or physical capital. The second chapter explores the questions "What influence do crime rates and levels of investment have on one another?" and "How do government policies affect the relationship between investment and crime?" by creating a simple structural model of crime and educational investment and attempting to fit this model to Mexican data. A method of simulated moments procedure is used to estimate parameters of the model and the estimated parameters are then used to carry out policy simulations. The simulations show that increasing spending on police or increasing the severity of punishment reduces crime but has little effect on educational investment. Increased educational subsidies increase educational investment but reduce crime only slightly. Thus, one type of policy is insufficient to accomplish the goals of both reducing crime and increasing education. The third chapter is joint work with Prashant Bharadwaj, Giacomo De Giorgi, and Christopher Neilson. Boys tend to have better performances than girls in mathematical testing; in particular, there are significantly more boys than girls among high achievers and the score distribution appears to have a longer right tail for boys. We confirm such results on several low- and middle-income countries. In particular we find that the gender gap is already present by age 10 and substantially increases by age 14 and 15. We propose and try to test a series of explanations for such a gap: (i) parental investment, (ii) ability, (iii) school resources, (iv) individual investment and effort (not tested directly), (v) competitive environment, and (vi) cultural norms. We conclude that none of our proposed explanations can account for a substantial portion of the gap.