/* This program conducts a duration analysis of the lengths of strikes as a function of the deviation of output from its trend level, an indicator of the business cycle position of the economy. The data was downloaded from the CD provided in the Greene textbook, Econometric Analysis, 4th Ed. Table A20.1 in his data appendix. The data was originally analyzed by J. Kennan (1985) in his paper "The Duration of Contract Strikes in U.S. Manufacturing," Journal of Econometrics, 28, 5 - 28. */ data strike; input dur eco; datalines; 7.00000 .0113800 9.00000 .0113800 13.0000 .0113800 14.0000 .0113800 26.0000 .0113800 29.0000 .0113800 52.0000 .0113800 130.000 .0113800 9.00000 .0229900 37.0000 .0229900 41.0000 .0229900 49.0000 .0229900 52.0000 .0229900 119.000 .0229900 3.00000 -.0395700 17.0000 -.0395700 19.0000 -.0395700 28.0000 -.0395700 72.0000 -.0395700 99.0000 -.0395700 104.000 -.0395700 114.000 -.0395700 152.000 -.0395700 153.000 -.0395700 216.000 -.0395700 15.0000 -.0546700 61.0000 -.0546700 98.0000 -.0546700 2.00000 .00535000 25.0000 .00535000 85.0000 .00535000 3.00000 .0742700 10.0000 .0742700 1.00000 .0645000 2.00000 .0645000 3.00000 .0645000 3.00000 .0645000 3.00000 .0645000 4.00000 .0645000 8.00000 .0645000 11.0000 .0645000 22.0000 .0645000 23.0000 .0645000 27.0000 .0645000 32.0000 .0645000 33.0000 .0645000 35.0000 .0645000 43.0000 .0645000 43.0000 .0645000 44.0000 .0645000 100.000 .0645000 5.00000 -.104430 49.0000 -.104430 2.00000 -.00700000 12.0000 -.00700000 12.0000 -.00700000 21.0000 -.00700000 21.0000 -.00700000 27.0000 -.00700000 38.0000 -.00700000 42.0000 -.00700000 117.000 -.00700000 ; proc lifereg data=strike; model dur = / dist = weibull; output out=out quantiles=.1 .5 .9 std=std p=predtime; title 'Weibull Model with no explanatory variables'; proc lifereg data=strike; model dur = eco / dist = weibull; output out=out quantiles=.1 .5 .9 std=std p=predtime; title 'Weibull Model with eco explanatory variable'; proc print data=out; run; /* Trying out the other duration distributions: Exponential, Loglogistic, and Lognormal. Let's see if the estimates match what Greene reports in his 4th edition book in Table 20.10, p. 942. */ proc lifereg data=strike; model dur = eco / dist = exponential; run; proc lifereg data=strike; model dur = eco / dist = llogistic; run; proc lifereg data=strike; model dur = eco / dist = lnormal; run;