Knowledge that Transforms

Numerous authors, including Shiller, LeRoy and Porter, and Singleton, have reported empirical evidence that stock prices and long interest rates are more volatile than can be justified by standard asset-pricing models. This paper shows that in small samples the "volatility" or "variance-bounds" tests tend to be biased, often severely, toward rejection of the null hypothesis of market efficiency. Thus the apparent violation of market efficiency may be reflecting the sampling properties of the volatility measures, rather than a failure of the market efficiency hypothesis itself. The paper also reports some unbiased estimates of the bounds on holding period yields and long interest rates. Much of the evidence of excess volatility disappears when the tests are corrected for small sample bias.

It is often argued that separation of public spending and taxing decisions engenders in the voter overoptimistic hopes that someone else will bear the cost of public services, thereby inducing an increase in the size of government. But are perceived tax prices systematically and persistently reduced by separation? Although in stock markets all agents' expectations may be unbiased, they may or may not be unbiased in political markets. This paper analyzes the separation created in federal states when the central government finances local expenditures. Evidence from a dynamic decisive-voter model is presented that indicates that the separation introduced by federal grants to Canadian provinces did in fact reduce the perceived tax price of provincial public services and raise provincial expenditures. The results suggest also that the effect of separation diminished over time.