* OLS ANOVA Model for Airlines' Production Efficiency * * Keywords: * regression, ols, log, anova, dummy, airlines, efficiency, production, f-test, * plot * * Description: * We illustrate how to estimate a log Multiple OLS model for the Production * Efficiency of Airlines, conduct Analysis of Variance in order to underpin * potential time and firm effects, construct F-tests to assess the significance * of these effects and create a plot * * Author(s): * Noel Roy * Skif Pankov * * Source: * William H. Greene, Econometric Analysis - 7th Edition * Pearson International Edition, Chapter 6, Example 6.4 (page 193) * * Reading the datafile and naming the variables, specifying to ignore the * first line of the file read (TableF6-1.shd) i t c q pfuel loadfact / skiplines=1 * Generating new variables genr lnc=log(c) genr lnq=log(q) genr lnq2=lnq**2 genr lnpf=log(pfuel) * Creating dummies that represent the time and firm effects. A do command * creates 14 time dummies, named d1, d2, ... d14, using the dum function, * which has a value of 1 when its argument is positive and 0 otherwise. * Similarly, five firm dummies f1, f2, . . .f5 are created set nodoecho do #=1,14 genr d#=dum(1-(t-#)**2) endo do #=1,5 genr f#=dum(1-(i-#)**2) endo * Running an OLS model of lnc on lnq, lnq2, lnpf, loadfact and all generated * dummies (please note the syntax used to include all numbered dummies), * specifying to save estimated coefficients as a vector b ols lnc lnq lnq2 lnpf loadfact d1-d14 f1-f5 / coef = b * Testing the time effects model; our null is that there are no firm effects * present test test f1 test f2 test f3 test f4 test f5 end * Testing the firm effects model; our null is that there are no time effects * present test test d1 test d2 test d3 test d4 test d5 test d6 test d7 test d8 test d9 test d10 test d11 test d12 test d13 test d14 end * Testing for the absence of either firm or time effects test test d1 test d2 test d3 test d4 test d5 test d6 test d7 test d8 test d9 test d10 test d11 test d12 test d13 test d14 test f1 test f2 test f3 test f4 test f5 end * Replicating figure 6.1; the year dummy variables are stored as elements * 5-18 of vector b, the vector of coefficients saved in the ols command * (elements 1-4 are the coefficients of the variables lnq lnq2 lnpf and * loadfact sample 5 18 genr year=time(-4) graph b year /nokey stop