Abstract
A large theoretical and empirical literature has studied the response in prices to changes in demands and costs. Three of the essays in this thesis are empirical studies of price-setting. The most important reason for studying the properties of price adjustments is the possible link between pricing and the business cycle. One family of models deals with price rigidities. If prices adjust slowly to changes in costs or demand, it will magnify fluctuations in output. A second family of models examines price changes due to changes in intensity of competition. Many of these models build on the idea that changes in demand affect the possibility of maintaining implicit collusion. Cyclical changes in the intensity of competition may cause a cyclical pricing pattern that magnifies fluctuations in output. Another line of research models the effects of liquidity constraints on the business cycle. One possible effect is that prices in markets where consumers have switching costs may increase, if the firms are hit by increased liquidity constraints. As increased liquidity constraints are common in recessions, prices will get a counter-cyclical tendency, which magnifies fluctuations in output. Another reason for studying price-setting patterns is that they can give an indication of the form of interaction between competing firms, which are of importance for competition policy. The fourth essay models sex discrimination formally. It shows that it is possible that sex discrimination in the labor market might be due to self-fulfilling sex stereotypes on the distribution of time out for child care between men and women. Both an even and an uneven distribution of time out for child care are possible equilibria in the model. The four essays can be read separatelyEssay 1 Price Responses to Seasonal Demand Changes in the Swedish Gasoline MarketA common idea in a number of papers on cyclical pricing is that implicit price collusion may be affected by changes in demand. In these models, tacit collusion is modeled as a game where agents balance gains from deviating from the collusive price, thereby gaining a short-run profit, against the gains from maintaining collusion in future periods. If demand fluctuates, the gains from deviating is high in periods of high demand; collusion may still be sustainable, however, if the collusive price is allowed to vary with demand. A lower price in high demand states reduces the gains from deviating, since it reduces the gain from each unit sold in these demand states. Hence, in order to equalize the profit from deviating and the profit from sticking to the implicit agreement, the price must move in the opposite direction to demand. Changes in demand can, naturally, be correlated with other factors affecting the price, such as cost changes. Thus, prices do not necessarily fall when demand increases in the data. According to the theory the, however, increase in demand should add a tendency to lower prices in order to make implicit collusion sustainable. The basic idea that changes in demand should affect prices if firms are engaged in implicit collusion has been employed in a number of papers, with different models for different assumptions on the pattern of demand changes. The demand for gasoline in Sweden follows a seasonal cycle, with demand being 42 percent higher in July than in January. Haltiwanger and Harrington provide a theory for implicit collusion over deterministic cycles, such as the seasonal cycle. Borenstein and Shepard (1996) tested Haltiwanger and Harrington’s model on the American gasoline retail market and found support for the theory. In this essay the model is tested for Swedish data, but no support for this theory is found. It is also investigated whether the effects on margins of the demand fluctuations induced by tax increases are compatible with theories of implicit collusion, but this is found not to be the case.Essay 2 Price Adjustments by a Gasoline Retail ChainEssay 2 is joint with Marcus Asplund and Richard Friberg. Stickiness of prices is an important building block in many business cycle models. This has spurred an empirical literature on price adjustments. Different types of price rigidities have different policy implications. Price setting can be state dependent, e.g. prices are adjusted when costs have changed by at least some minimal amount since the last price adjustment, or time-dependent e.g. adjusted once a week or once a year. Another issue is whether prices are equally rigid downwards as upwards. The second essay examines price responses in the Swedish gasoline retail market to changes in the Rotterdam spot price of gasoline, exchange rates and taxes. The main results are that cost changes are not fully passed through in the short run, but gradually moves towards the long-run equilibrium. Prices are stickier downwards than upwards. There is a minimum absolute size of price changes. Only very limited evidence of time-dependent price setting is found.Essay 3 Prices, Margins and Liquidity Constraints: Swedish Newspapers 1990-1996Essay 3 is written with Marcus Asplund and Niklas Strand. Chevalier and Scharfstein (1996) provide a model where consumer switching costs in combination with liquidity constraints give rise to a counter-cyclical tendency in prices. Customer stocks can be viewed as an investment, when consumers have switching costs when changing suppliers. Firms can exploit captured customers by setting a high price to raise short-run profits. However, a high price will induce consumers to search for alternatives, and customers once lost are costly to win back. Liquidity constraints, for instance in recessions, may force firms to sacrifice long-run profits for short-run gains. In this case, firms may have to cut back on investments in customer stocks by raising prices. Using firm level data from the Swedish daily newspaper industry, we test the effects of liquidity constraints on prices in markets with consumer switching costs. The newspaper industry is of particular interest, since firms set prices in two markets, the subscription market, where switching costs are high, and the advertising market, where switching costs are low. With accounting data from newspaper firms we can, by solvency, broadly categorize them as being more or less liquidity constrained. When Sweden enters a recession at the beginning of the nineties, we find a relative increase in subscription prices and margins for liquidity constrained firms. This is not the case for advertising prices, however. The results support the theory.Essay 4 Statistical Discrimination and Sex Stereotypes in the Labor MarketTime out for child care is unevenly distributed between the sexes. Parental leave benefits are usually exclusively given to the mother or distributed to both parents according to their choice. In Sweden, one month is reserved for each parent, however. One reason for this is to give the child better contact with both parents, but increased equality between the sexes in the labor market has also been put forward as an argument. This argument implicitly rests on the idea that sex stereotypes create sex discrimination, and that sex discrimination affects the distribution of time out for child care between the sexes. Essay 4 investigates if the uneven distribution of time out for child care can be explained by self-fulfilling sex stereotypes. It provides a model of distribution of time out for child care based on statistical discrimination and human capital investments. The model has three equilibria. In one equilibrium, time out for child care is evenly distributed between the sexes. In the second equilibrium, there is full specialization. The third equilibrium is an intermediate case, where time out for child care is unevenly distributed without full specialization There are no differences in ability or variance of ability between the sexes, the only differences between the equilibria are the self-fulfilling expectations of firms and workers.