A Stochastic Dynamic Programming Analysis of Farmland Investment and Financial Management

This dissertation investigates farm firm growth using a multiperiod investment portfolio problem that includes farmland, nonfarm assets, and debt financing on farmland. The investment portfolio problem is formulated as a stochastic continuous-state dynamic programming model. Since this dynamic programming model lacks a closed-form solution, it is solved numerically with collocation methods. We develop a test for checking the accuracy of a stochastic continuous-state dynamic programming model. Using this accuracy test, we examine the accuracy of the solution for the investment portfolio problem. We compare the accuracy of collocation methods with Chebyshev and linear spline interpolations. We propose techniques for improving the accuracy and efficiency in solving a large-scale dynamic programming model. We solve the investment portfolio problem that includes risky farmland and a riskless nonfarm asset or debt financing on farm land in the presence of transaction costs, credit constraints, stochastic land prices and farm returns. We explore how the optimal portfolio is adjusted in a dynamic and stochastic environment. Results show that the optimal portfolio depends on farm returns, farmland price, and liquid assets. We explore the effect of initial farm size, initial wealth levels, length of the planning horizon, interest rate, riskiness of returns, and risk aversion. Observed risk avoiding behavior in investment decisions is often attributed to risk aversion. We find that risk avoiding behavior in investment decisions can also be attributed to the length of the decision maker’s planning horizon. Also, unlike in a static model, changes in the riskiness of returns affect the optimal portfolio in the dynamic model, even when the decision maker is risk neutral. The above portfolio problem is extended by adding a risky mutual fund investment. We find that it is optimal for farmers to include the mutual fund in the portfolio along with farmland investment. Furthermore, higher debt financing on farmland is optimal when mutual fund investment is included in the model. Finally, we find that the probability of exiting farming increases in the model with the mutual fund investment opportunity