Prediction ability

My dissertation constructs a micro foundation for one important component of the manager\u27s human capital: the ability to predict future events that affect profits of a firm. Prediction ability is measured by quality of information which a manager observes in order to predict future events. It contributes to 3 different subjects. My static assignment model shows that a manager who has prediction ability goes to a risky industry because risk increases the marginal productivity of prediction ability. This prediction contrasts with that of Lucas (1978), in whose model talented managers simply manage bigger firms. Data support this view: Talented B-school graduates choose to work in risky industries. The correlation between an ability measure and a risk measure is 0.75. I simulated my assignment model. The results fit the B-school placement data quite well. It shows that a 1 percent increase in GMAT of a B-school graduate brings a 158 percent increase in the risk of a firm that the graduate is assigned to. I construct an observable measure of a manager\u27s prediction ability. Although many economists apply the Blackwell theorem (1951) to analyze an economic impact of accurate information, none estimate it. This dissertation shows that, in a particular function, quality of information can be estimated by the correlation coefficient between future profitability and current decisions. Since Tobin\u27s Q reflects the future profitability of capital, I construct a measure of prediction ability from the correlation between a firm\u27s future Q and its current growth rate. Using this measure I show that managers prediction ability has a positive impact on a firms expected Tobins Q with evidence from the COMPUSTAT dataset. My dynamic investment model shows that a manager who expects to receive more valuable information attains a faster expected growth rate: If current investment decreases future adjustment costs, a good manager has an incentive to invest more on average in order to prepare for a future investment opportunity. This result is in contrast to that of Demers (1991), who argued that the more valuable information a manager expects to receive, the less he invests on average. Using my measure of prediction ability, the data conform to my theory as well with evidence from the COMPUSTAT dataset