Because transactions costs are smaller for allocating new cash flows than for reallocating existing asset holdings, financial flow variables are important determinants of investors' short-run asset demands. The demand-for-bonds equations implied by the resulting "optimal marginal adjustment" model of portfolio behavior constitute the demand side of a structural supply-demand model of the determination of the long-term interest rate. Empirical results, based on demand-for-bonds equations estimated using U.S. data for six major categories of bond market investors, support the optimal marginal adjustment model and show that the associated structural model of interest rate determination, which is restricted by the underlying demand-for-bonds equations, fits the data about as well as do previously developed unrestricted reduced-form term-structure equations.

This paper uses a model of the valuation of bonds bearing call options, together with observed market yields on callable bonds, to infer information about the uncertainty associated with interest rate expectations. A dynamic programming solution of the model simultaneously determines both the bond price and the issuer's optimal refunding strategy, given the relevant data describing the bond and the market's expectations of future interest rates. Application of the valuation model in reverse, for quarterly average data for 1969-76, generates a time series representing the uncertainty which the market associated with its expectations of future interest rates during this interval, given the then-prevailing yields on new issues of utility bonds and industrial bonds callable after 5 years and 10 years, respectively. This uncertainty, parameterized as the standard deviation of a truncated normal distribution, fluctuated between 1/2 percent and 3/4 percent between 1969 and early 1974, then rose to sharply higher levels from mid-1974 through mid-1975, and has fluctuated between 3/4 percent and 1 percent since late 1975.

Substantial shifts in wealth ownership from individuals to pension funds are currently taking place in the United States and also are in prospect for the foreseeable future. Moreover, pension funds typically exhibit portfolio preferences that are markedly different from those of individuals. In a world of heterogeneous investors, redistributions among wealth holders with different portfolio preferences will in general alter the structure of asset yields. Partial-equilibrium simulation experiments based on a model of the U. S. long-term bond market indicate that redistributions of saving flows from individuals to pension funds, in plausible magnitudes, can have major effects on the term structure of interest rates. In a world in which wealth holders ' risk aversion renders different assets less than perfect substitutes, the interaction between investors' portfolio preferences and existing asset supplies determines the structure of asset yields. Using a model in which the only explicitly traded assets are money and bonds, for example, Patinkin [19651 showed explicitly how either a shift in the exogenously determined

The results presented below demonstrate that the structural modeling approach to interest rate determination not only stands apart from the sectoral disaggregation question conceptually but also performs fairly well without sectoral disaggregation empirically. This paper presents estimation and dynamic simulation results for an aggregated equivalent to the disaggregated model of the determination of bond yields developed in Friedman (1977; 1979). Instead of six bond demand and two bond supply equations, here there are but one demand and one supply equation. The empirical results show that, while disaggregation is of value in structural interest rate modeling (that is, the disaggregated model outperforms the aggregated one), even the aggregated structural model performs very well in comparison with familiar unrestricted reduced-form term structure equations.

The use of iterative algorithms, based on the Gauss-Seidel method or a similar approach, to solve systems of nonlinear simultaneous equations may lead to problematical situations which in theory are not surprising, but in practice are unexpected by the user. In particular, such situations may arise in the solution of econometric models for simulation purposes. One source of the problem lies in the failure of these algorithms, which repeatedly solve single equations according to some sequential ordering,1 to deal with the interaction properties of specific higher-order subsystems of closely related equations. One illustration of such a subsystem is the set of equations which determines unemployment and labor force in the Wharton Econometric Forecasting Model [1].

Including data from the 1980's sharply weakens the postwar time-series evidence indicating significant relationships between money (however defined) and nominal income or between money and either real income or prices separately. Focusing on data from 1970 onward destroys this evidence altogether. Evidence indicating cointegration of real income and real money balances, with due allowance for the effect of interest rates, also deteriorates when the sample extends through the 1980's. A positive finding is that the spread between the commercial paper rate and the Treasury bill rate consistently contains highly significant information about future movements in real income.