SAMPLE 1 30 * Beer demand data set from * Griffiths, Hill and Judge (1993, Table 11.1, p. 372). * The variables are quantity demanded (Q), price of beer (PB), price of * other liquor (PL), price of other goods (PR) and income (Y). READ Q PB PL PR Y 81.7 1.78 6.95 1.11 25088 56.9 2.27 7.32 0.67 26561 64.1 2.21 6.96 0.83 25510 65.4 2.15 7.18 0.75 27158 64.1 2.26 7.46 1.06 27162 58.1 2.49 7.47 1.10 27583 61.7 2.52 7.88 1.09 28235 65.3 2.46 7.88 1.18 29413 57.8 2.54 7.97 0.88 28713 63.5 2.72 7.96 1.30 30000 65.9 2.60 8.09 1.17 30533 48.3 2.87 8.24 0.94 30373 55.6 3.00 7.96 0.91 31107 47.9 3.23 8.34 1.10 31126 57.0 3.11 8.10 1.50 32506 51.6 3.11 8.43 1.17 32408 54.2 3.09 8.72 1.18 33423 51.7 3.34 8.87 1.37 33904 55.9 3.31 8.82 1.52 34528 52.1 3.42 8.59 1.15 36019 52.5 3.61 8.83 1.39 34807 44.3 3.55 8.86 1.60 35943 57.7 3.72 8.97 1.73 37323 51.6 3.72 9.13 1.35 36682 53.8 3.70 8.98 1.37 38054 50.0 3.81 9.25 1.41 36707 46.3 3.86 9.33 1.62 38411 46.8 3.99 9.47 1.69 38823 51.7 3.89 9.49 1.71 38361 49.9 4.07 9.52 1.69 41593 * Consider a log-linear model GENR LQ=LOG(Q) GENR C=1 GENR LPB=LOG(PB) GENR LPL=LOG(PL) GENR LPR=LOG(PR) GENR LY=LOG(Y) * OLS estimation OLS LQ C LPB LPL LPR LY / NOCONSTANT LOGLOG * * Estimation by generalized entropy methods * Specify the parameter and error support matrices Z and V DIM Z 5 5 V 30 3 SAMPLE 1 5 GENR Z:1=-5 GENR Z:2=-2.5 GENR Z:3=0 GENR Z:4=2.5 GENR Z:5=5 SAMPLE 1 30 GENR V:1=-1 GENR V:2=0 GENR V:3=1 * Solve the dual problem and get the optimal Lagrange multipliers GME LQ C LPB LPL LPR LY / NOCONSTANT ZENTROPY=Z VENTROPY=V LOGLOG * Try another model DIM ZD 4 4 SAMPLE 1 4 GENR ZD:1=-2.5 GENR ZD:2=-0 GENR ZD:3=2.5 GENR ZD:4=5 * DEVIATION - Measure variables as deviations from the mean; * this option automatically forces NOCONSTANT. SAMPLE 1 30 GME LQ LPB LPL LPR LY / DEVIATION ZENTROPY=ZD VENTROPY=V LOGLOG STOP