Partial identification of nonlinear peer effects models with missing data

Swiss Journal of Economics and Statistics

Table 4 Time performance of the Horowitz–Manski (HM) and Manski–Tamer (MT) interval estimators (with non-closed groups and individuals excluded from their own peer group)

Group size	No. of groups	5 missing observations			10 missing observations			7 missing observations
		Prob. of missing	Average time (secs)		Prob. of missing	Average time (secs)		Prob. of missing	Average time (secs)
		(in %)	Logit		(in %)	Logit		(in %)	OLS
		(in %)	HM	MT	(in %)	HM	MT	(in %)	HM	MT
5	50	2.00	0.7	0.6	4.00	23	0.5	2.80	120	0.3
5	100	1.00	0.8	0.6	2.00	27	0.5	1.40	124	0.3
5	200	0.50	1.1	0.8	1.00	40	0.6	0.70	123	0.3
5	500	0.20	2.0	1.4	0.40	76	1.2	0.28	140	0.5
5	1000	0.10	3.7	2.4	0.20	114	2.3
5	2500	0.04	6.6	4.2	0.08	199	4.2
10	50	1.00	0.8	0.6	2.00	24	0.4	1.40	117	0.3
10	100	0.50	1.1	0.7	1.00	28	0.6	0.70	102	0.3
10	200	0.25	1.7	1.0	0.50	44	1.0	0.35	121	0.5
10	500	0.10	3.6	2.3	0.20	111	2.3	0.14	155	0.7
10	1000	0.05	6.1	3.8	0.10	195	3.8
10	2500	0.02	12.6	7.8	0.04	400	8.0
25	50	0.40	1.1	0.7	0.80	38	0.6	0.56	118	0.3
25	100	0.20	1.7	1.1	0.40	70	1.2	0.28	137	0.5
25	200	0.10	3.7	2.2	0.20	118	2.3	0.14	155	0.7
25	500	0.04	6.6	4.4	0.08	200	4.2	0.06	242	1.6
25	1000	0.02	12.5	8.3	0.04	389	7.8
25	2500	0.01	30.6	19.6	0.02	966	19.7