Arthur Havenner

State space modeling of multiple time series

Time series methods offer the possibility of making accurate forecasts even when the underlying structural model is unknown, by replacing the structural restrictions needed to reduce sampling error and improve forecasts with restrictions determined from the data. While there has been considerable success with relatively simple univariate time series modeling procedures, the complex interrela- tionships possible with multiple series requite more powerful techniques.Based on the insights of linear systems theory, a multivariate state space methos for both stationary and nonstationary problems is described and related to ARMA models. The states or dynamic factors of the procedure are chosen to be robust in the presence of model misspecification, in constrast to ARMA models which lack this property. In addition, by treating th emidel choice as a formal approximation problem certain new optimal properties of the procedure with respect to specification are established; in particular, it is shown that no other m...

Cointegration and stock prices : The random walk on wall street revisited

Abstract One of the more famous and empirically regular instances of unit roots occurs in stock price series, where they are consistent with rational expectations and efficient markets under certain assumptions. In this paper we investigate the dynamic behavior of five stock prices over the volatile period from January 1972 through December 1979, finding that the series are cointegrated with one dominant common trend, a component with an estimated root of 0.95. In addition, we find a lesser trend and two complex roots in the cycle model. These findings are validated by out-of-sample forecasts beginning in January 1980. In addition to mean squared error criteria, the forecasts are evaluated nonparametrically in two separate applications of a test due to Henriksson and Merton.

The Economics and Econometrics of Damage Control

Concern for the potentially harmful side effects of agricultural chemical inputs, especially pesticides, highlights the need to accurately determine the economic levels of their use. We consider three model specification issues: interaction of direct production inputs with damage control inputs in damage abatement, justification for a priori exclusion of production inputs from the abatement function, and the motivations and consequences of alternative stochastic specifications. Empirical analysis using farm-level data shows that misspecification of the stochastic element in the production function can overestimate the marginal physical productivity of pesticides and grossly underestimate the responsiveness of demand to increases in pesticide prices. Copyright 1997, Oxford University Press.

FOREIGN EXCHANGE RATES: A MULTIPLE CURRENCY AND MATURITY ANALYSIS

Abstract Three dollar exchange-rate series — the German mark, the British pound, and the Japanese yen — are converted to market price revisions by calculating the difference between the price for delivery at a fixed date and the preceding periods price for delivery at the same date. These price revisions are likely to meet the stationary stochastic process assumptions required for time-series modeling even if the original series are non-stationary. Using error covariance assumptions analogous to a pooled time series of cross-sections, a multivariate time-series model is fitted and tested on 168 observations out of sample. Forecasting performance is evaluated using both mean squared errors and a non-parametric test of direction.

Stochastic production function estimation: small sample properties of ML versus FGLS

Just-Pope production functions have been traditionally estimated by feasible generalized least squares (FGLS). This paper investigates the small-sample properties of FGLS and maximum likelihood (ML) estimators in heteroscedastic error models. Monte Carlo experiment results show that in small samples, even when the error distribution departs significantly from normality, the ML estimator is more efficient and suffers from less bias than FGLS. Importantly, FGLS was found to seriously understate the risk effects of inputs and provide biased marginal product estimates. These results are explained by showing that the FGLS criteria being optimized at the multiple stages are not logically consistent.

A method for approximate representation of vector-valued time series and its relation to two alternatives

Abstract A method for constructing a state-space representation of vector-valued time series is discussed and related to two other methods for Markovian model construction from a unifying point of view. The three methods are shown to employ different normalizations and to employ different optimality criteria for summarizing in a finite-dimensional state vector information contained in past data. The method proposed in this paper is shown to lead to stable innovation models for time series. A numerical example is included to illustrate the potential of the proposed method.

Applications of Computer Aided Time Series Modeling

I: Introduction to State Space Modeling.- 1. The SSATS algorithm and subspace methods.- 2. A guide to state space modeling of multiple time series.- II: Applications of State Space Algorithm.- 1. Evaluating state space forecasts of soybean complex prices.- 2. A state space model of monthly US wheat prices.- 3. Managing the heard: price forecasts for California cattle production.- 4. Labor market and cyclical fluctuations.- 5. Modeling cointegrated processes by a vector-valued state space algorithm.- 6. A method for identification of combined deterministic stochastic systems.- 7. Competing exchange rate models.- 8. Application of state-space models to ocean climate varibility in the northeast pacific ocean.- III: Applications of Neural Networks.- 1. On the equivalence between ARMA models and simple recurrent neural networks.- 2. Forecasting stock market indicies with recurrent neural networks.

System Theoretic Time-Series Forecasts of Weekly Live Cattle Prices

Multivariate time-series forecasts of weekly live cattle prices in six different geographic markets are developed using a procedure based on the principles of linear systems theory. These forecasts were found to be informative and superior to those obtained from an alternative model. Following the approach developed for stock prices by Cerchi and Havenner, arbitrage portfolios were constructed from the model parameters. A simulation exercise based on 208 weekly observations withheld from model specification and estimation suggests that these arbitrage activities would have been profitable in practice.

Estimation Analogies in Control

Abstract Least squares estimation and quadratic openloop optimal control share both algebra and intuition; this paper explores the duality to develop the parallels in each. Estimation results suggest control analogs to correlation measures, significance tests, distributed lags and problems of multicollinearity, tests of compatability between restrictions and data, and a version of the Kalman filter that is sometimes computationally expedient. Conversely, the control literature provides methods of incorporating certain types of inequality restrictions and a development of recursive estimation based on the Kalman filter. A by-product is a method of solving control problems using least squares computer programs.

Choosing a monetary instrument The case of supply-side shocks

Abstract This paper examines the monetary instrument choice problem in models with an explicit supply sector and endogenous prices and price expectations. We use the popular linear-quadratic certainty-equivalence framework to obtain analytic solutions. The results indicate that interest rate policies generally are better able to insulate the real sector from unanticipated supply shocks. We also show that at this level of abstraction the optimal choice of an instrument is independent of the specification of expectations, e.g., rational vs. adaptive price expectations. The paper then reports empirical evidence from two experiments on the large nonlinear MIT-PENN-SSRC econometric model. The evidence is consistent with the hypothesis that interest rate policies are preferable if the supply sector is the major source of uncertainty.