Expected Inflation and Interest Rates: Reply
回复Tanzi的评论，指出名义利率受多种因素影响，负实际利率在通胀期并不异常，且投资者对税收的认知不能仅凭税后实际利率为负来判断。
Vito Tanzi's comment on my earlier study (1982) is based upon his belief that the Mundell effect alone could not likely account for the ex ante negative real rates observed in his simulations. His suggestion is that Perhaps a Mundell effect, together with a fiscal illusion, provides the best explanation (p. 502). Nowhere did I argue that the Mundell effect was all-encompassing. nominal rate is determined by many factors besides expected (X). My reducedform nominal rate equation includes uncertainty, the current and lagged money stock, and the lagged price level as well as expected as arguments. Thus, at any point in time, the nominal rate will shift with a movement in any of these variables. One way of looking at this is that holding s constant, i will vary with the other predetermined variables so that the real rate changes. Since Tanzi and previous researchers have used partial equilibrium analysis to highlight the relation between 1T and i, we tend to forget that the level of the real rate depends upon all the predetermined variables of the system. To concentrate on the partial derivative of i with respect to sr, we must take as given some initial i, holding constant the other factors influencing i. Tanzi's simulations set the real rate of (r) initially equal to .03. Then by assuming different values for di/d7r and v, he generates implied values for the after-tax real rate (r*). Since the after-tax real rate turns negative at quite low rates of 7r, Tanzi concludes that fiscal illusion is present as interest rates did not adjust for the effect of taxes (p. 502). simple fact that the after-tax real rate turns negative need imply nothing regarding whether or not investors are cognizant of taxes. There is nothing inherently unusual about negative real rates of in the short run during inflationary periods. Clearly, if money can be costlessly stored, the nominal rate of cannot be negative but this in no way rules out negative real rates. a world of incomplete markets, there may be inflationary periods when a negative real rate of is the best that optimizing agents can earn. If there was a suitable asset available that guaranteed a zero return, we would not expect to observe ex ante negative real rates. Benjamin Friedman (1980) emphasizes the limited nature of portfolio substitution possibilities by arguing that if holding goods was a feasible portfolio alternative, then the expected rate of could be viewed as simply the return from holding such real assets. such a case, higher expected would reduce the lenders supply of loans directly via portfolio substitutions from loans (bonds) to real goods. problem with this scenario, as stated by Friedman is: In practice, however, the available opportunities for such portfolio substitutions involving consumption or other real commodities are usually extremely limited; purchasing the Consumer Price Index basket of goods is not a feasible portfolio alternative (p. 34). lack of such a real asset allows market instruments to offer a negative yield. Of course, with no (and costless storage), one could hold money so that we don't expect to observe negative rates in noninflationary periods. Frederic Mishkin provides corroborating evidence in this area as he finds The real rate, whether adjusted or unadjusted for taxes, is negatively correlated with inflation (1981, p. 191, emphasis added) and The real rate appears to have been positive in the 1950s and 1960s but has since turned negative in the midand late-1970s (p. 192). Regarding Tanzi's simulations, I do not believe that they correctly assess the path of the real rate following an exogenous change in inflationary expectations (nw). A major theme of my earlier paper was the emphasis *Arizona State University. I thank Don Schlagenhauf, John Schroeter, and Richard L. Smith for helpful discussions.