Corporate Hedging, Executive Compensation and Commodity Price Prediction

This thesis examines the agency problem surrounding the corporate hedging decision. It gives insight on how managerial incentives impact corporate hedging decisions and on how executive compensation can be used to minimize the agency problem and factors determining the optimal compensation. The model predictions are then tested against empirical data. One of the factors aecting optimal executive compensation is volatility of commodity prices. To explore this, the last chapter develops an empirical model to forecast commodity prices.

Past theoretical and empirical studies found that risk-averse managers tend to overhedge, without analyzing how to align shareholders and managers hedging strategies. In this dissertation I develop a model aligning hedging strategies using executive compensation, incorporating a risk-averse managers utility into the hedging decision. Consistent with standard theories, the model show managers hedge more of the expected production than shareholders. The model shows there is a decrease in corporate hedging with the presence of managerial equity-based incentive pay. It also shows managerial incentives can be used to impact corporate hedging to minimize agency problem. To align and optimize managerial hedging decisions, the optimal managerial incentive should comprise more of the equity-based portion when there is a low risk tolerance, or low price volatility, or a low variable cost. In contrast, when there is high coecient of absolute risk aversion, or low price volatility, or high variable cost, it is best to compensate the manager with a lower equity-based portion in order to optimally align hedging decisions. In other words, by determining and examining the primary factors aecting compensation scheme includes risk aversion, price volatility, and prot margin we can determine the optimal compensation scheme. When there is a low (high) coecient of absolute risk aversion, low (high) price volatility, or low (high) variable cost, then optimal compensation should comprise more (less) equity-based incentives.

Next, using empirical data I test the model predictions from the theoretical framework; (i) when incentive pay increases, the optimal hedge ratio decreases, (ii) when price volatility increases, the optimal hedge ratio decreases, while price volatility have a negative relation with equity-based incentive, (iii) when risk aversion increases, the optimal hedge ratio decreases, while risk aversion have a negative relation with equity-based incentive, and (iv) when variable cost increases, the optimal hedge ratio decreases, while variable cost have a negative relation with equity-based incentive. The predictions are tested against data obtained from oil and gas rms using a standard regression approach. I nd that the model predictions are further supported by empirical evidence from the oil and gas industry showing (i) a negative relationship between incentive pay and hedge ratio, (ii) a negative relationship between price volatility and hedge ratio/incentive pay, (iii) a negative relationship between risk aversion and hedge ratio/incentive pay, and (iv) a negative relationship between price volatility and hedge ratio/incentive pay. Overall, the rst two chapters claries the optimal compensation scheme under varying economic environments in order to mitigate the agency problem associated with hedging decisions.

Last, a new model for the series of West Texas Intermediate (WTI) crude oil prices process is introduced, which accommodates spikes and local trends in its trajectory, as well as the multimodality of its sample distribution. The model relies on the convolution of two stationary processes, causal and noncausal processes, which allows for the estimation of the monthly WTI crude oil prices series. As an alternative specication, the mixed causal-noncausal autoregressive (MAR) models are estimated and used for oil price prediction. Two forecasting methods developed in the literature on MAR processes are applied to the data and compared. In addition, this chapter examines the long-term relationships between the WTI crude oil price, the Ontario Energy Price Index (OEP) and the Ontario Consumer Price Index (OCPI). These relationships are established using the cointegration analysis. The vector error correction (VEC) model allows us to predict the Ontario price indexes and the WTI crude oil prices. This chapter shows an alternative simple method of forecasting Ontario price indexes from stationary combinations of WTI crude oil price forecasts obtained from the mixed causal-noncausal autoregressive (MAR) models. This chapter shows that both method of prediction yields forecasts that are close approximation of the out of sample value.