Count data models with variance of unknown form: an application to a hedonic model of worker absenteeism

We examined an econometric model of counts of worker absences due to illness. The underlying theoretical model is of a sluggishly adjusting hedonic labor market. We compared results fromı three parametric estimators, nonlinear least squares plus Poissonand negative binomial pseudo maximum likelihood, to generalized least squares using nonparametric estimates of the conditional variance. Our data support the hedonic model of worker absenteeism. Semiparametric generalized least squares coefficients are similar in sign, magnitude, and statistical significance to their econometric analogs where the mean and variance of the errors were specified ex ante. Overdispersion test reject the Poisson specification. Robustness checks confirm that in our dataı parameter estimates are sensitive to regressor list but are not sensitive to econometric technique, including how we corrected for possible heteroskedasticity of unknown form