A Comment on the Appropriate Estimation of Intrametropolitan Firm Location Models
评论了多篇论文中剔除禁止工业用地辖区后税收变量显著影响企业选址的结论，指出这种做法可能导致截断偏差，且是否实际禁止企业入驻是实证问题。
The analysis in several papers-Wasylenko (1980), Fox (1981), Charney (1983), and McHone (1986)-of intrametropolitan firm location is performed on a data set that excludes those jurisdictions that zone out firms. The general conclusion from these studies is that when these jurisdictions are removed from the sample, the fiscal variables, especially taxes, become statistically significant determinants of firm location choices. This is in contrast to the conclusions about the tax variables when all jurisdictions are included in the analysis. The argument for the removal of these jurisdictions is that if jurisdictions zone out firms, then it is inappropriate to include these jurisdictions in the sample as they are not places that are supplying sites for firm locations. Thus including such places in the sample may bias the estimation results.1 Peddle (1987) maintains that the case for eliminating places that zone out industry is weak, because, according to his view, this involves many fewer jurisdictions than researchers typically exclude from their sample. Thus, removing these jurisdictions from the sample leads to truncation bias in the estimated coefficients. Whether or not jurisdictions actually zone out firms is an empirical issue and can not be settled on the basis of his