Knowledge that Transforms

In his 1983 paper, Jeremy Siegel derives a seemingly implementable policy rule involving optimal responses to interest rates. The existence of such a rule would be of tremendous interest to central banks whose monetary policies place heavy weight on responses to interest rates. The Siegel rule is especially appealing because it is (i) an optimal combination policy in the sense of William Poole (1970), and (ii) the proposed implementation of the rule does not require detailed knowledge of the structure of the economy. All that is required is a calculation of the covariance between innovations in prices and interest rates. Within the confines of a rational expectations equilibrium model, in which Siegel assumes agents do not make use of information embodied in the current nominal interest rate, he is able to design an optimal combination policy that does not require detailed information about the economy. That such an optimal policy exists is not new, but that it can be easily implemented is novel.' The policy rule depends solely on the covariance between innovations in the aggregate price level and innovations in the nominal rate of interest, normalized by the variance of innovations in the interest rate. When this index is zero, policy has been set optimally. When the index is positive the feedback term on interest rates in the money supply rule is too large, and when the index is negative the feedback term is too small. Given that one can obtain reduced-form expressions for prices and interest rates, the index is easily computed. Unfortunately, Siegel's proposal violates Robert Lucas's (1976) critique. That is, he implicitly treats as invariant certain aspects of economic behavior that will generally change when one moves to an operational interest rate This note shows in detail that in a model where prices are flexible and agents observe local market prices (i.e., the model at least employed verbally by Siegel), that the coefficients in the aggregate supply and demand functions are not invariant to the form of the money supply rule. This lack of invariance will cause Siegel's rule to be nonoperational. The sensitivity of parameters in aggregate supply functions to policy is not restricted to equilibrium models with flexible prices. This property also extends to contracting models with endogenous indexing (see, for example, Jo Anna Gray, 1976). Therefore, Siegel's rule will not be implementable in a wide variety of commonly used macro models.

Since the publication of Bridger Mitchell's article on the Optimal Pricing of Local Telephone Service (1978), it has been assumed that social welfare can usually be increased by moving from a flat monthly rate for local calls to a two-part tariff with a price per call that is somewhat in excess of marginal cost. While a fixed monthly charge for local calls can be considered a regressive head tax (A. M. Henderson, 1947), it does not follow that a two-part tariff will resolve the equity problem. In this paper I use a simple diagram and a two-person revenuemaximizing formula to illustrate one of the more important limitations of usage-sensitive pricing. In the following analysis, it is assumed that there are two types of telephone users. The first type of consumer, D1, is assumed to have a net demand for calls or message units, represented by the linear equation:

Technical conditions give the entrepreneur a much freer hand about the acquisition of inputs (which are largely needed to start a new process) than about the completion of outputs (whose process of production-in the ordinary business sense-may be already begun). If forward markets consisted entirely of hedgers, there would always be a tendency for a planned weakness on the demand side; a smaller proportion of planned purchases than of planned sales would be covered by forward contracts. [1946, p. 137]

In two important studies, Charles Plott (1982) and Vernon Smith (1982) assess the current state of the literature about laboratory experiments in economics. As a profession, we are becoming aware that experimental methods can be applied to our models, with cautious but growing confidence that these procedures can help us evaluate alternative theories. These are significant developments whose potential ramifications are only beginning to be explored. Given the importance of laboratory testing, I would like to discuss a key feature of past experiments, one that has not been fully appreciated because of its central role in standard economic theory. In particular, these experiments depend on inducing agents to respond according to a prespecified value structure. This usually amounts to starting with a known and fully determinate set of demand and supply value schedules for all transacting agents. Smith, for example, specifies four major principles about how preferences are to be experimentally induced (nonsatiation, saliency, dominance, and privacy; see pp. 931-35).