Bayesian Learning and Investment Dynamics
Bartholomew Moore - Rutgers University and Huntley Schaller - Carleton University
In this paper we explain anomalies to the theory of investment by
introducing Bayesian learning about an unobserved state variable. We
consider a firm whose problem is to maximize the present value of
expected cash flows. We assume that the logarithm of the gross real
rate of interest exhibits Markov-switching in its mean, and that the firm
uses Bayes' rule to infer the current Markov state. We show that a
high signal-to-noise ratio and high persistence of the Markov state
variable can explain the low correlation between investment and the
optimal-linear forecast of marginal q, and the serial correlation of the
unexplained portion of investment. The model also explains the sluggish
response of investment to changes in the cost of capital.
Consider, for example, how investment will respond to observed movements in the interest rate when there is a persistent low-interest-rate regime and a less persistent high-interest-rate-regime. Suppose that the signal-to-noise ratio is high so that the firm does not often mistake white noise shocks for a change of regime. While the economy is in the persistent low-interest-rate regime, transitory shocks will do little to alter Bayesian firm's forecast of interest rates and will thus have little or no effect on investment. The white noise shocks will, however, cause serially correlated movements in the optimal-linear forecast of marginal q. An econometrician regressing observed investment on the linear q will obtain a low R-squared and serially correlated residuals. Furthermore, since the low-interest-rate regime is known to be persistent, it is only after there has been a large and persistent increase in interest rates that investment will respond. The response to a change in the state variable will be gradual, thus explaining the sluggish response of investment to changes in the cost of capital.
Scheduled for Session 2.2 Modeling Economic Dynamics And Adjustment Costs