121Unionization in Canada and the United Statesknown, but their treatment in the same fashion by the two surveys should not significantly affect the comparative analysis. As in Farber (1983, 1990), I assume DES; = 1 for all those with US; = 1. Excluded from the analysis are those not currently employed, the selfemployed, managers, and those who did not respond to the questions on union status, desire for union representation, public or private sector employment, and demographic variables common to the two surveys (age, education, sex). After these exclusions and the adjustment to the AFL survey for quota sampling, there are 517 observations from the CFL survey (out of 1,000 interviews) and 890 observations from the AFL survey (out of 1,452 interviews). Some of the key statistics are summarized in table 4.3. The top part of the table shows these statistics for each of the full samples while the bottom part splits the samples into the public and private sectors. These surveys confirm that the probability that a randomly selected Canadian worker is unionized is more than double that of a U.S. worker (0.48 versus 0.22). To some extent this differential is overstated by the inclusion of members of professional associations in the Canadian data but not in the U. S. data. 9 The remainder of this large differential can be accounted for by three factors. First, the desire for union representation in Canada is about 28 percent higher than that in the United States (Prob(DES; = 1) of 0.64 versus 0.50). There is also less unsatisfied demand for union status in Canada; Prob(DES; = IIUS; = 0) is 0.30 in Canada versus 0.36 in the United States. However, the most remarkable difference between the two countries is clearly the greater supply of unionization conditional on desire for union status; Prob(US; = 1IDES; = 1) is 0.76 in Canada versus 0.44 in the United States, that is, 73 percent higher in Canada. These summary statistics indicate that Canada's higher union density is due to both greater demand for and greater supply of union coverage in Canada, but that intercountry differences in the supply of union representation appear to be relatively more significant. In order to assess the relative importance of demand and supply factors, the relationship Prob(US; = 1) = Prob(US; = 1IDES; = 1) * Prob(DES; = 1) can be decomposed into two components corresponding to differences in the demand for and supply of unionization:~Prob(US;= 1) = ~Prob(DES; = 1) * Prob(US; = 1IDES; = 1)(4)+Prob(DES; = 1)0.14 .084* 0.6* ~Prob(US; + 0.565 * 0.32==1IDES;1)+ .1810.265,9. Union membership as a percentage of paid workers was 37 percent in Canada in 1984 and 16 percent in the United States in 1990, a differential of 21 percentage points. Collective agreement coverage as a percentage of paid workers was 42 percent in Canada in 1984 and 18 percent in the United States in 1990, a differential of 24 percentage points. The sources for these statistics are the surveys reported in table 4.2.
