Forecast of Virginia coal production

This thesis provides a model for forecasting coal production rates in southwest Virginia. A multiple linear regression model is developed for the forecasting process. The model includes six independent variables: Virginia coal price times Virginia coal mining productivity (x₁), Virginia mining company production levels (x₂), Virginia coal reserves (x₃), U.S. domestic electricity consumption (x₄), U.S. coal exports (x₅), and U.S. domestic industrial (includes coke) coal consumption (x₆).

Historical values for the six variables from 1979 to 1993 were used in generating the multiple linear regression model coefficients. The model captures 86.87% of the variation in Virginia coal production over this period. The forecast from 1994 to 2010 was generated by using forecasted values for the six independent variables. The sensitivity of the model was tested by slightly changing the values of selected independent variables. The results indicate a decline in Virginia coal production to approximately 32 million tons in 2010. Under more favorable conditions, the model results indicate that the coal production levels will remain approximately steady. Under less favorable economic conditions, the model results indicate that Virginia coal production levels will be reduced by more than 50% by the year 2010.

These results were also supported by a curve fitting exercise based on the work of M. King Hubbert (1969 and 1973). Hubbert states that the production of a non-renewable natural resource follows a normal or lognormal curve. The peak production level will occur when approximately have of the reserves have been mined or slightly before. Approximately 2 billion tons of coal has been mined in southwest Virginia. Current recoverable reserves are estimated at 1.5 billion tons.