ESTIMATING THE AGRICULTURAL DEMAND FOR NATURAL-GAS AND LIQUEFIED PETROLEUM GAS IN THE PRESENCE OF MEASUREMENT ERROR IN THE DATA

The paper begins by discussing the importance of accurate estimates of the price elasticity of demand and some of the problems frequently encountered in obtaining these estimates. To these problems is added that associated with inaccuracy in the measurement of the dependent variable and one or more of the independent variables that affect the quantity demanded. Two diagnostics, i.e. the regression coefficient bounds and the bias correction factor, have been introduced to assess the effect that such measurement error has on the estimated coefficients of demand relationships. The use of these diagnostics will aid in assessing the integrity of the estimates obtained. In considering the demand for natural gas and the demand for liquefied petroleum gas by farmers in the USA, both the quantity demanded and the price data available for demand model estimation purposes contain measurement error. The regression coefficient bounds diagnostic was used to indicate a range over which the true price responsiveness of farmers to changes in energy prices lies. The results suggest that each 1% increase (decrease) in the price of energy will result in a decrease (increase) of between 0.41 and 0.17% in the quantity of natural gas demanded and a decrease (increase) of between 0.48 and 0.07% in the quantity of liquefied petroleum gas demanded. The bias correction factor was computed to evaluate the magnitude of the underestimation of the responsiveness of the quantity of natural gas and liquefied petroleum gas demanded to a change in the number of acres irrigated. For natural gas, the under-estimation was 26.5%, whereas, for liquefied petroleum gas, it was 9.5%.