I am analysing what are the determinants for the adoption of bonus compensation for a firm (i) headquartered in a country (j). I have more than 700 firms headquartered in 16 countries; my dependent variables is a dichotomous indicating whether bonus compensation is adopted (1) or not (0), and my key independent variables are the first two in the output of the model displayed below.
The first one indicates whether the country in which the firm is headquartered discourage (-1), encourage (1) or does not have a clear position on bonus compensation (0). The second one indicates whether the firm is listed in US.
As you can see, these two variables are measured on two different levels. Therefore, similarly to what happens with continuous dependent variable with mixed command, I'd want to understand how much of my variance is explained by firm- and country-level variables.
Code:
melogit FIRM_INCENTIVE_POLICY CodesIncentivesRECOMM2 USListing *** all remaining indep vars *** || CountryC:, or Fitting fixed-effects model: Iteration 0: log likelihood = -163.67629 Iteration 1: log likelihood = -136.00745 Iteration 2: log likelihood = -135.00212 Iteration 3: log likelihood = -134.96207 Iteration 4: log likelihood = -134.962 Iteration 5: log likelihood = -134.962 Refining starting values: Grid node 0: log likelihood = -139.52018 Fitting full model: Iteration 0: log likelihood = -139.52018 (not concave) ******I omitted all iteration for the sake of brevity****** Iteration 54: log likelihood = -134.9648 Iteration 55: log likelihood = -134.96271 Iteration 56: log likelihood = -134.96218 Iteration 57: log likelihood = -134.962 Iteration 58: log likelihood = -134.962 Mixed-effects logistic regression Number of obs = 727 Group variable: CountryCODED Number of groups = 16 Obs per group: min = 3 avg = 45.4 max = 365 Integration method: mvaghermite Integration pts. = 7 Wald chi2(23) = 200.88 Log likelihood = -134.962 Prob > chi2 = 0.0000 ---------------------------------------------------------------------------------------------- FIRM_INCENTIVE_POLICY | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -----------------------------+---------------------------------------------------------------- CodesIncentivesRECOMM2 | 9.843344 3.403505 6.61 0.000 4.998332 19.38475 USListing | 10.54212 6.53276 3.80 0.000 3.129346 35.5142 FAMILY | 2.478101 1.416788 1.59 0.112 .8081043 7.599248 Outsider5_LN | 1.005343 .1950676 0.03 0.978 .6873139 1.47053 MEETINGS_LN | .4335576 .2472911 -1.47 0.143 .1417559 1.326027 RepRisk_1Y_AVG | 1.003941 .0181223 0.22 0.827 .9690433 1.040096 BETA9802 | .7564532 .3403387 -0.62 0.535 .313195 1.827045 ASSETSUSD_LN | 1.074118 .225251 0.34 0.733 .7121118 1.620151 TOTALDEBTTOTALASSETS8 | 2.979734 3.803166 0.86 0.392 .2442032 36.3583 TSR | 1.001388 .0037582 0.37 0.712 .994049 1.008781 RETURNONASSETS8326 | .3602494 1.129024 -0.33 0.745 .0007743 167.5997 | SectorENCODED | Consumer Discretionary | 2.762074 1.600434 1.75 0.080 .8872009 8.599013 Consumer Staples | 2.734421 1.905945 1.44 0.149 .6975269 10.71938 Energy | 2.054449 1.460637 1.01 0.311 .5099447 8.276902 Health Care | 1.901764 1.115238 1.10 0.273 .6025489 6.002343 Information Technology | 7.974622 5.946515 2.78 0.005 1.84921 34.39014 Materials | 1.237006 .7601254 0.35 0.729 .3709527 4.12501 Telecommunication Services | 2.581441 2.960618 0.83 0.408 .2726662 24.43954 Utilities | 1.341628 1.07053 0.37 0.713 .2808211 6.409653 | Prot_min_inv_Ext_Dir_Liab_LN | 12.63175 17.98257 1.78 0.075 .7756964 205.7003 SayONPaySwitzNB | .9826529 .9250025 -0.02 0.985 .1552877 6.218179 RuleOfLaw | 3.383666 5.149616 0.80 0.423 .1713722 66.80896 CountryMKTCap2GDP | 1.004822 .0069745 0.69 0.488 .9912444 1.018585 _cons | .0000205 .0001294 -1.71 0.087 8.75e-11 4.81151 -----------------------------+---------------------------------------------------------------- CountryCODED | var(_cons)| 1.37e-34 5.45e-18 . . ---------------------------------------------------------------------------------------------- Note: Estimates are transformed only in the first equation. Note: _cons estimates baseline odds (conditional on zero random effects). LR test vs. logistic model: chi2(0) = 5.7e-14 Prob > chi2 = . Note: LR test is conservative and provided only for reference.
Thank you for your time,
Luigi
0 Response to How to interpret variance on different levels in melogit
Post a Comment