Market Behavior and Market Interaction under Incomplete Information

Abstract

Chapter 1 deals with the question of how buyers behave in the presence of uncertainty about the distribution of prices. To this end, we consider a model of a big, decentralized auction market in which two types of agents are active: buyers and sellers. Both types of agents can successively participate in more than one auction, but they face some probability of “dying" after every auction. The idiosyncratic feature of our model is that no buyer knows how many other buyers there are. Hence, there is some uncertainty about how intensely the buyers compete against each other. Our main result is that the buyers in our dynamic model behave the same as the buyers in a standard (static) common value auction (i.e. an auction for a good with unknown characteristics). Thus, much of what is known about common value auctions also applies to environments in which goods with <em>known </em>characteristics are for sale, but buyers do not know the distribution of winning bids.<br /> In chapter 2 we consider an economy in which the agents care about their perceived rank in the income distribution. Income is assumed to be unobservable, but can be signalled by means of consuming a “positional” good. We analyze the relationship between income inequality and spending on the positional good. Hopkins and Kornienko (2004) have argued that redistribution heats up the competition between rich and poor agents, thereby increasing conspicuous consumption. Adopting the natural assumption that an agent's consumption does not fully reveal her income, we show that redistribution also reduces the reliability of consumption signals, thereby lowering incentives to send them. This second effect matters little when income differences are large, but dominates in egalitarian societies. Thus, conspicuous consumption is highest at intermediate levels of income inequality.<br /> In chapter 3 we examine one way of estimating the returns to schooling, due to Belzil and Hansen (2002). The basic idea behind their approach is to exploit the information that is contained in individual schooling decisions. This can be done as follows: If an individual incurs certain costs in order to go to school, we can conclude that, for this individual, the returns to schooling must have been at least as high as the costs, for otherwise the individual would not have chosen to go to school. Conversely, every decision to exit school indicates that the costs of remaining in school would have been higher than the benefits. While the identification strategy of Belzil and Hansen sounds appealing, we show it to rely on the assumption that individual schooling decisions are partly driven by observable random wage shocks. Thus, while the approach of Belzil and Hansen is an interesting one, it has much more in common with the traditional “random components" method than one might think at first glance.