Table 1 Canadian provinces—DID estimates

Outcome: log weekly vaccine first doses, V_it
	(1)	(2)	(3)
Mandate announced, P_it	0.504***	0.506***
	(0.002)	(0.001)
Week 0			0.359***
			(0.005)
Week 1			0.543***
			(0.001)
Week 2			0.498**
			(0.010)
Week 3			0.705***
			(0.001)
Week 4			0.713**
			(0.018)
Week 5+			0.651*
			(0.056)
log weekly cases, C_it		0.002	0.002
		(0.967)	(0.958)
log weekly deaths, D_it		0.048	0.046
		(0.365)	(0.415)
R²	0.817	0.820	0.821
Sample size, N	920	920	920
Province fixed effects	X	X	X
Date fixed effects	X	X	X

Sun and Abraham²³ treatment effect heterogeneity robust estimates. Columns 1 and 2 report the average mandate effect estimate over the post-announcement period, without and with controlling for weekly cases and deaths. Column 3 decomposes the effect by post-announcement week. The sample period is 15 June to 14 September 2021, using Alberta, New Brunswick, Saskatchewan, Prince Edward Island and Newfoundland as the control group (latest treated). Week n, where n = 0, 1, 2, …, is a binary variable that takes value 1 for the days in the nth week immediately after the announcement date (week 0 is the week starting at the announcement date) and value 0 otherwise. The C_it and D_it variables are log weekly totals for dates t − 6 to t. P values from wild bootstrap (boottest) standard errors clustered by province with 4,999 repetitions are reported in parentheses; *, ** and *** denote 10%, 5% and 1% significance (two-sided), respectively. A coefficient estimate x is equivalent to 100x log points or a 100(e^x − 1) per cent increase in weekly doses. The results in column 3 are not directly comparable to Fig. 3 because in Table 1 the reference point is the entire pre-announcement period. The column 3 estimates for week 4 and onwards are based on post-mandate vaccinations in Quebec, the province with the earliest announcement. The smaller T = 0 estimate for week 0 reflects the fact that the outcome variable is a backward-looking weekly sum.

Quick links

Search