Presentation on theme: "1 SHORT-RUN DENSITY FORECAST FOR ETHANOL AND MTBE PRICES Michael H. Lau Joe L. Outlaw James W. Richardson Brian K. Herbst Texas A&M University Department."— Presentation transcript:
1 SHORT-RUN DENSITY FORECAST FOR ETHANOL AND MTBE PRICES Michael H. Lau Joe L. Outlaw James W. Richardson Brian K. Herbst Texas A&M University Department of Agricultural Economics Agriculture as a Producer and Consumer Energy Conference Arlington, VA, June 24-25, 2004
2 Introduction Rapidly increasing ethanol production. –The National Energy Act exemption and the Clean Air Act of 1990 aided ethanol growth. Gallagher, et. al. (2003) and Lidderdale (2000) looked at long run effects of renewable fuel standards. Ethanol price determined by wholesale gasoline price and the excise tax exemption on blended gasoline (Coltrain, 2001; CFDC, 2004; NDLC, 2001).
3 Objectives and Methods Create short-run probabilistic (density) forecasts for ethanol and MTBE prices for a 24-month period from October 2003 to September 2005. –Why probabilistic (density) forecasts? The results from this study will provide interested parties an unbiased analysis and forecast of ethanol and MTBE prices as production and demand for ethanol continues to grow.
4 Historical Ethanol and MTBE Price Nov. 1994 to Sept. 2003 Ethanol and MTBE prices are available from Hart’s Oxy-Fuel News. Wholesale gasoline price is available from the Energy Information Agency (EIA) of the U.S. Department of Energy (DOE).
5 Summary Statistics for Ethanol and MTBE Prices Nov. 1993 to Sept. 2003 Ethanol PriceMTBE PriceWholesale Gasoline Price Mean1.220.900.76 Standard Deviation0.180.220.18 95 % LCI1.190.860.72 95 % UCI1.260.940.79 CV14.8824.2023.27 Min0.940.470.43 Median1.200.870.73 Max1.711.561.15 Autocorrelation Coefficient0.920.860.93 Source: Energy Information Agency, U.S. Department of Energy.
6 Other Exogenous Variables Considered for Forecasting Retail gasoline price, ethanol fuel demand, gasoline production, diesel price, diesel production, gasoline stock, diesel stock, oil production, oil imports, oxygenated gasoline production, reformulated gasoline production, oxygenate stock, corn prices, and corn production. None were statistically significant and did not improve the forecasting abilities of the VEC model based on the mean absolute percent error (MAPE)
7 Augmented Dickey Fuller and Johansen Co-integration Tests Co-integrating Equation Augmented Dickey Fuller Test for Stationarity. VariableLevels*First Difference** Ethanol Price-2.48-4.98 MTBE Price-1.95-10.12 Wholesale Gasoline Price-1.55-9.93 * Fail to reject "Ho: Data series is non-stationary" at 1 % significance level. ** Reject "Ho: Data is non-stationary at 1% significance level. Johansen Unrestricted Co-integration Rank Test HypothesizedTrace5 Percent1 Percent No. of CE(s)EigenvalueStatisticCritical Value None **0.1622.3715.4120.04 At most 1 *0.044.103.766.65 *(**) denotes rejection of the hypothesis at the 5%(1%) level Trace test indicates 2 co-integrating equation(s) at the 5% level Trace test indicates 1 co-integrating equation(s) at the 1% level
11 PDF Graphs for Ethanol and MTBE Prices Jan. 2004 and July 2005
12 Fan Graph for Ethanol Price Oct. 2003 to Sept. 2005
13 Fan Graph for MTBE Price Oct. 2003 to Sept. 2005
14 Conclusions and Questions VEC is successfully for forecasting Ethanol and MTBE Prices from Oct. 2003 to Sept. 2005. Simulation is beneficial for predicting density forecasts of ethanol and MTBE prices. Limitations of Study –Dependent on forecasted values of wholesale gasoline price. –Perpetual forecast should be used where the coefficients are updated as realized prices occur for ethanol, MTBE, and wholesale gasoline.