Table 4.

Continuity-based RD analysis: predetermined census 1940 covariates using local polynomial regression

Covariates	Coefficient	CER-optimal bandwidth [meters]	Robust inference			Nl	Nr
			95% CIrbc	SE	p-value
Total occupied dwelling units/m2	0.00	589.729	[0.00, 0.003]	0.00	0.439	57	88
Total population/m2	0.001	566.81	[− 0.006, 0.007]	0.00	0.802	57	84
White population	− 0.037	552.109	[− 0.146, 0.087]	0.06	0.621	57	84
“Negro” population	0.04	562.387	[− 0.09, 0.146]	0.06	0.607	57	84
Homeownership rate	− 0.07	411.152	[− 0.17, 0.054]	0.06	0.308	47	75
Median home value	− 706.36	528.97	[− 2494.90, 806.892]	842.31	0.316	56	83
Male employment rate	0.00	573.186	[− 0.02, 0.009]	0.01	0.601	57	85
Male high school completion rate	0.01	396.638	[− 0.01, 0.019]	0.01	0.361	45	73
Vacancy rate	0.01	484.644	[− 0.02, 0.035]	0.01	0.428	54	80

Discrete analysis using cluster standard errors; all models use first-degree polynomial; median home value is rounded to the nearest whole dollar

^*p < 0.10;

^**p < 0.05;

^***p < 0.01