This dissertation investigates the role of misintermediation in macroeconomic fluctuations. A perennial problem of financial markets is that of maturity mismatching, or misintermediation, a situation in which financial intermediaries fund long-term, illiquid loans with short-term liabilities. McCulloch (1981) concludes that misintermediation can be responsible for business cycles and predicts pro-cyclic behavior of surprises in real interest rates over business cycles. In my dissertation, I study the issue from both theoretical and empirical perspectives. An interpolation method for yield curve fitting is also specified to enable the empirical study of misintermediation.
My dissertation consists of four chapters. In Chapter 1 and 2, I theoretically investigate the mechanism for generating a disequilibrium boom or recession in a finite horizon structural model. In each period, agents decide how much to consume and how much to invest in heterogeneous capital for subsequent periods. If demand and supply happen to coincide in each period, the model will have a unique equilibrium term structure of interest rates. Otherwise, unexpected changes in real interest rates will occur accompanied by the realization of a recession or boom because previous plans cannot be completely corrected as a new period starts. The model is then extended to study the changes of real wage rates and heterogeneous capital prices under either a recession or boom regime. It reveals that misintermediation would not only bring about unexpected output fluctuations and surprises in interest rates, but also give rise to unanticipated changes in factor prices.
An empirical examination of the misintermediation hypothesis relies on an empirical estimation of the term structure of interest rates. Chapter 3 hence specifies a multiple exponential decay model to fit both U.S. real and nominal term structures following the approach of interpolation. Several estimation methods, including unconstrained/constrained minimization, quadratic programming and iterative least squares, are introduced to estimate the parameters in the objective function according to different curve-fitting purposes. As a comparison, this chapter also proposes a semi-natural cubic spline model to fit the same data set. The results show that the multiple exponential decay model not only gives a parsimonious functional form, but also smoothes through idiosyncratic variations associated with the forward rate curve. In addition, selection of the number of terms/coefficients in an interpolation function governs the overall goodness of fit, which is optimized by three separate statistical tools.
In Chapter 4, an empirical study examines the relationship between unanticipated changes in real interest rates and unexpected fluctuations in real output over time. The former is derived from the U.S. real term structures estimated by the multiple exponential decay interpolation. Specifically, a monthly series of synthetic real consol prices is constructed as a proxy for the price of all future output, changes in which suggest changes in real interest rates over time. To proxy unexpected output fluctuations, I use either a time series of innovations to real GDP or a series of innovations to factor utilization. Statistical results show a negative correlation between the consol price series and the factor-utilization-based series as well as the real GDP-based series. This empirical evidence is consistent with the misintermediation hypothesis.