Table 4.

Sensitivity

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
PcV P4P	0.0176***	0.0159**	0.0212	0.0192*	0.0153**	0.0252***	0.0189
	(0.000)	(0.0073)	(0.0157)	(0.0112)	(0.0068)	(0.0063)	(0.0120)
CI lower	0.00698
CI upper	0.0282
Municipality FE	X	X	X	X	X	X
Year FE	X	X	X	X	X	X	X
Municipality trends	X	X			X	X
Covariates		X		X			X
Population weights	X	X	X	X		X	X
Observations	2,312	2,304	2,312	2,304	2,312	2,016	2,312
Municipalities	289	288	289	288	289	289	289
R 2	0.914	0.020	0.027	0.095	0.004	0.059

Note. Sensitivity tests for the main dependent variable, that is, the PcV share (= penicillin V prescriptions' share of total consumption of antibiotics prescribed for respiratory tract infections (ATC codes: J01CE02/J01AA02/J01CA04/J01CR02/J01DB‐DE/J01FA), children 0–6 years; mean = 0.69). (1) Wild cluster bootstrap on the preferred specification (999 replications, p‐value in parenthesis; CI lower/upper = 95% confidence interval); (2) preferred specification + time‐varying covariates (share of children and elderly, population share with secondary/tertiary education, log population and mean income in municipality, I (control county with other antibiotic‐related P4P), I (implementation year of entry/choice reform), I (drug cost responsibility of care centers)); (3) preferred specification but excluding municipality linear trends; (4) preferred specification, excluding municipality linear trends, and including covariates; (5) preferred specification not using population size weights; (6) preferred specification using entropy balancing weights (2007–2013 sample; see the online Appendix for entropy balancing results); (7) fractional response model. Columns 2–7: standard errors clustered by county council in parentheses (cluster‐bootstrapped with 999 replications in col. 7). P4P = pay‐for‐performance; FE = fixed effects. ***p<.01, **p<.05, *p<.1.