Measures of Dispersion and Variability: Range, QD, AD and SD
The Impact of Unionization Threat on Non-union Wage Rates in Canada
1. THE IMPACT OF UNIONIZATION THREAT ON NON-UNION WAGE
RATES IN CANADA
BY
JOEY YI ZUO
OCTOBER 2007
2. i
Abstract
In an effort to reduce the workers’ benefits from joining the union, employers increase wages
of their non-union workers when facing an increased threat of unionization (Rosen, 1969).
This paper presents novel evidence regarding the effect of the threat of unionization on wage
rates in Canada for the period between 1998 and 2006. Drawing on the insights provided by
nine consecutive annual Canadian Labor Force Surveys, I find that the threat of unionization
has a larger positive effect on the non-union wages compared to the threat’s effect on the
union wages and, hence, has an inverse effect on the union-wage gap. Further, the analysis
by sectors suggests that these results hold for the private sector only and do not extend to the
public sector. Importantly, I find that the results are sensitive to the definition of the
unionization threat and to the list of explanatory variables included in the estimation model.
5. iv
LIST OF FIGURES
Figure 1: Union membership in Canada (1998-2006) - - - - - - - - - - - - - - - - - - - - - -44
Figure 2: Effect of predicted probability of unionization on the non-union/union wages
and the union-wage gap, for the whole sample, by year- - - - - - - - - - - - - - - - - - - - -45
Figure 3: Effect of predicted probability of unionization on the non-union/union wages
and the union-wage gap, for the public sector, by year- - - - - - - - - - - - - - - - - - - - - 46
Figure 4: Effect of predicted probability of unionization on the non-union/union wages
and the union-wage gap, for the private sector, by year - - - - - - - - - - - - - - - - - - - - -47
Figure 5: Effect of industry union density on the non-union/union wages and the union-
wage gap, for the whole sample, by year. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -48
Figure 6: Effect of industry union density on the non-union/union wages and the union-
wage gap, for the private sector, by year- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 49
Figure 7: Effect of industry union density on the non-union/union wages and the union-
wage gap, for the public sector, by year- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -50
6. 1
1. Introduction
Union membership in Canada has declined from approximately 55% to 30%, since the
1980s, according to Blanchflower (2006). The decline can be attributed to decreasing union
membership in the private sector. In contrast, union membership in the public sector has
increased in the past decade. Over the last decade private firms have responded to the threat
of unionization most notably by subcontracting, outsourcing, and even plant-closings. The
retail giant Wal-Mart, for example, recently closed its store in Jonquiere, Quebec after the
store became unionized (Bianco 2006). In this paper I am interested in assessing whether the
unionization threat in Canada justifies such severe reactions on the part of firms.
My analysis draws on the seminal work by Rosen (1969). Rosen suggests that the
ability of unions to negotiate higher wages increases with the extent of unionization; that is,
as the proportion of employed workers who are union members in an industry or occupation
increases. In particular, Rosen notes that as the extent of unionization increases “[the
possibility] for output or product substitution against unionized firms is reduced”. Moreover,
to avoid an increase in wage rates that follow unionization, employers are expected to
increase the wages of their current non-union employees in an attempt to reduce the
employees’ benefits from joining the union. This increase in non-union wages is likely
greater when the non-union employees have similar attributes to those of the union members
or in industries (occupations or cities) with substantial union presence. The literature refers to
this phenomenon as the effect of the unionization threat on non-union wages.
Several studies tested these predictions (e.g., Kahn 1978; Moore et al. 1985; Podgursky
1986). Despite the wide interest, the empirical evidence on the unionization threat effect
remains an unsettled question. Using firm-level data, Leue and Tremblay (1993), for instance,
7. 2
found no significant effect of unionization threat on non-union wages. Freeman and Medoff
(1981) found that in manufacturing, the unionization threat has a strong positive effect on
union wages, but no or a weak effect on non-union wages. While the literature on the
unionization threat effect primarily draws on the U.S. data, Canadian studies tend to focus on
identifying the wage gap between union and non-union workers. The estimates range from
9.5% in Kumar and Stengos (1985), to between 16% and 51% in MacDonald and Evans
(1981), and 34.7% in Chaykowski and Slotsve (2002). To my knowledge, no study has
examined the effect of unionization threat on wage rates in Canada.
This paper attempts to fill this gap in the literature by providing evidence that pertains
to the unionization threat effect on the non-union wages and the union-wage gap. Using data
from the Canadian Labor Force Surveys between 1998 and 2006, I build on prevalent
approach in the literature and construct industry-level union density as my proxy for the
threat of unionization. I expand on this initial approach by constructing an alternative
measure following Farber’s (2005) methodology. This alternative measure for the threat of
unionization is constructed as the predicted probability of union membership, a function of
worker-, job-, and firm-specific attributes. I proceed to regress wages of non-union workers
on these alternate unionization threat variables while controlling for a wide set of observable
worker-, job-, and firm-characteristics. In addition, I extend on Farber’s (2005) study by
examining separately the effect of the threat of unionization in the private and public sectors.
My results are consistent with the hypothesis that the threat of unionization is directly
related to an increase in non-union wages in the private sector, but not in the public sector.
The magnitudes of the threat effects in my study are somewhat similar to those in the related
literature that draws on the U.S. data. The estimated threat effect in the non-union private
8. 3
sector in Canada is between 14.9% and 21.9% (from 1998 to 2006), as compared to 20% in
the Farber’s (2005) study which draws on the U.S. data. Importantly, I find that my results
are sensitive to the list of explanatory variables and to the definition of the unionization
threat. In Section 2, a review of the literature on the unionization threat effect is provided.
Section 3 provides a brief description of the theoretical framework that motivates my
empirical analysis. Section 4 focuses on methodology and Section 5 on data description.
Results are presented in Section 6. Section 7 concludes.
2. Literature Review
Economists have been long concerned with assessing the effect unions have not only on
wages of union workers but also on wages of non-union workers. Some argued that the effect
of unions on the non-union wages (i.e., the unionization threat effect) results from a desire by
the non-union employers to avoid unionization; as higher wages reduce the benefits to
unionization. Rosen (1969), for instance, argued that we should observe a positive
relationship between the non-union wages and the extent of union organization (or the
percent of unionized workers in the industry) at lower levels of union organization. This
prediction is supported by Moore et al. (1985). The authors found that the unionization
density in an industry with fewer union members has a positive effect on non-union wages.
Corneo and Lucifora (1997) and Kahn (1978) reported similar findings.
Podgursky (1986) extended the argument and found that non-union wages at medium-
sized firm increase with the union threat. Pearce (1990) also suggested that the effect of
unionization threat increases with firm size in the non-union sector. Farber (2005)
documented that stronger evidence in favour of the union threat effects could be found in
9. 4
deregulated industries. However, some studies report that no significant threat effect is found.
Freeman and Medoff (1981), for instance, found that in manufacturing, the threat has a
strong positive effect on union wages, but no or weak effect on non-union wages. Leue and
Tremblay (1993) also claimed that no significant effect was found of the effects of either the
percentage organized or the firm-level predicted threat of unionization on non-union wages.
Another strand of literature has provided findings of the unionization threat’s effect on
wage dispersion. Belman and Heywood (1990), for instance, have shown that the percentage
organized in the union reduces union wage dispersion but has a weak effect on non-union
wages. According to findings in Neumark and Wachter (1995), at the industry (city) level, an
increase in the percentage organized in the union reduced (increased) the non-union industry
(city) wage differential. Kahn and Curme (1987), on the other hand, found that an increase in
the percentage organized in unions decreased the dispersions of non-union wages.
The reviewed papers tend to use different measures of the threat of unionization (see
Appendix A Table 2). The most common measure for the unionization threat is an industry-
level or occupation-level union density. For example, Podgursky (1986) uses the proportion
of production workers who are covered by union contracts in an industry as a measure of the
union threat. Kahn and Curme (1987) and Moore et al. (1985) use both industry-level and
occupation-level union membership rates. Neumark and Wachter (1995) employ the city-
level union density rate as an explanatory variable in the wage regression. Farber (2005), on
the other hand, uses the predicted probability of being a union member as a measure of the
threat. Unlike the industry-level union density, Farber’s measure allows the threat of
unionization to differ not only across industries, but also across workers who are employed in
the same industry but differ in their age, attained education, marital status, gender, etc.
10. 5
In this paper, I follow Farber’s (2005) approach. I use repeated cross-sectional data
from 1998 to 2006 for Canada to construct the predicted probability of union membership. I
also construct an alternative measure, industry-level union density, in an attempt to infer how
sensitive the results are to the definition of the union threat measure. Using both measures I
can, therefore, better understand why related literature has found remarkably different
evidence for the effect of unionization threat on non-union wages. In addition, a separate
analysis for the public sector and the private sector is provided. I am interested in the latter
distinction, since it is more likely that employers in a private sector are confronted with
decisions stipulated in a theoretical framework of profit maximization. In contrast, employers
in the public sector may be pursuing other goals such as ensuring stable employment.
Overall, my analysis contributes in four respects to the related literature. Namely, I: (1)
estimate the effect of unionization threat on non-union and union wage rates in Canada over
an extended period of time; (2) identify the threat effect on the wage gap between union and
non-union workers; (3) examine both the union and the non-union wage responses to the
unionization threat separately in the public and private sectors; (4) measure the threat effect
with the predicted probability of union membership and the industry-level union density.
3. Theoretical Framework
Following Farber’s (2005) methodology, I let P (α , β ) denote the probability of
unionization, where α = (WU − W N ) / W N denotes the union wage gap ( 0 < α < 1 ), WU the
union wages, W N the non-union wages, and β the index of the threat of unionization for a
given union-wage gap ( 0 < β < 1 ). I assume that P > 0 , Pβ > 0 , P > 0 , P > 0 , and
α αα αβ
11. 6
Pββ > 0 . Note that Pαβ > 0 implies that, the marginal effect of an increase in the union-wage
gap on the probability of unionization increases in magnitude as the threat β increases.
Let the expected wage be denoted as E (W ) . Hence, the expected wage is a weighted
average of the union wage and the non-union wage rates with weights representing the
probability of unionization ( P ) and the probability of non-unionization ( 1 − P ), respectively.
Using the above introduced notation, the expected wage can be written as:
PWN (WN − WU )
E (W ) = WN + P(WU − WN ) = WN + = WN +PWN α = WN (1 + Pα ) . (1)
WN
Employers who employ non-unionized workers choose WN in order to minimize E (W ) . The
optimal W N solves the first order condition obtained by setting the derivative of
the E (W ) with respect to WN to zero: (1 − P) − P α (1 − α ) = 0 . The effect of the threat of
α
unionization on the non-union wage rate can be obtained by taking the derivative of this first-
order condition with respect to β :
∂WN P + P α (1 − α ) ∂WN ∂WU
= β αβ
+ . (2)
∂β ( P + P )(1 + α ) 2 ∂WU ∂β
α αα
If ∂WU / ∂β ≥ 0 , one gets:
∂WN P + P α (1 − α ) ∂WU
= β αβ
+ ∂WU / ∂β > ≥0. (3)
∂β ( P + P )(1 + α )
α αα
2
∂β
Pβ + Pαβ α (1 − α )
Since Pα > 0, Pβ > 0, Pαα > 0, Pαβ > 0, Pββ > 0 , 0 < α < 1 it follows that > 0.
( Pα + Pαα )(1 + α ) 2
The comparative statistics’ results in (2) and (3) are central to this study that aims to
estimate the effect of the threat of unionization on the non-union wage. The result suggests
that an increase in the likelihood of unionization ( β ) has: (1.) a positive effect on the non-
12. 7
∂WN
union wage ( > 0 ); (2.) a nonnegative effect on the union wage such
∂β
∂WN ∂WU
that > ≥ 0 ; and (3.) a negative effect on the union-wage gap or the union wage
∂β ∂β
premium. This paper tests empirically these three predictions by drawing on the data
collected from nine annual labor force surveys in Canada from 1998 to 2006.
4. Methodological Approach
To test these predictions I use two measures for the threat of unionization ( β ). In the
next two sections I describe how these two measures are constructed. My third approach to
testing the model’s prediction explores one of the model’s assumptions; i.e., that employers
minimize their wage costs. While this assumption may be valid for employers in the private
section, it may not be a good description of the employers’ decisions in the public sector. I
explore this conjecture by examining separately the effect of unionization threat on non-
union wages for workers in the private sector and for workers in the public sector.
4.1. Predicated Probability of Union Membership as a Measure of Unionization Threat
My first approach to estimating the unionization threat’s effect on the non-union wage
rates follows Farber (2005). In the first step, I estimate the predicted probability of union
membership by running a probit regression for each year:
Prob(Unioni = 1| X i ) = φ (η ' X i ) . (4)
In this equation, Φ ( ⋅) denotes a standard normal cumulative distribution function, η is a
vector of coefficients I wish to estimate, and Xi is a vector of worker and firm characteristics,
13. 8
industry and province dummies (See Appendix A, Table 1 for detailed explanation of
variables). The threat of unionization assigned to worker i in my sample is defined as follows:
∧ ∧
threat i = φ (η ' X i ) . (5)
∧
In the second step, I use the threat variable, threat i , as an independent variable in the
wage regression. I estimate separately the wage regression for the sample that consists solely
of non-union workers and in a sample that consists solely of union workers. In particular, an
econometric specification for the union wage equation can be written as:
∧
ln( wageiU ) = δ 0U + δ1U threat iU + γ U ' X iU + ε iU . (6)
Similarly, the wage regression I estimate for the non-union workers is:
∧
ln( wageiN ) = δ 0 N + δ1N threat iN + γ N ' X iN + ε iN . (7)
∧
In equations (6) and (7), threatiU is the predicted probability of being a union member
∧
for a worker in the sample of union workers, and threatiN is the predicted probability of
being a union member for a worker in the sample of non-union workers. Hence, unionization
threat is measured by the extent the non-union employees have similar attributes to those of
the union members. XiU is a vector containing other explanatory variables in the union sample;
XiN is a vector containing other explanatory variables in the non-union sample; γ U and γ N are
vectors of estimated coefficients for the control variables; δ 0U and δ 0 N are the constants, δ 1U
and δ1N are the coefficients on the threat effects; and ε iU and ε iN are the residuals in the
sample of union workers and the sample of non-union workers, accordingly.
Theory suggests that non-union firms are expected to increase wage rates for their non-
union workers when faced with the unionization threat. In particular, a positive correlation
14. 9
between the threat of unionization and the non-union wage rates is expected. Specifically, we
expect δ1N > 0 . We also expect that δ1N > δ1U ≥ 0 , due to the results derived in (3).
4.2. Industry Union Density as a Measure of Unionization Threat
In my second approach, I consider an alternative measure of unionization threat,
following the approach prevalent in existing literature (e.g., Podgursky 1986; Kahn and
Curme 1987). This measure for the threat of unionization is the industry-level union density.
In particular, this alternative measure is constructed, for each year, in the following manner:
∧ number of workers employed in industry j who are union members
threat j , alt = . (8)
number of workers employed in industry j
In the second step, I run a regression of the logarithmic value of hourly wage on the
alternative measure of the union threat, controlling for various worker and firm specific
attributes. Hence, I estimate the following equation for the sample of union workers:
∧
ln( wageiU ) = δ 0UAlt + δ1UAlt threatiUAlt + γ ' X iUAlt + ε iUAlt . (9)
And similarly for non-union workers:
∧
ln( wageiN ) = δ 0 NAlt + δ1NAlt threatiNAlt + γ ' X iNAlt + ε iNAlt . (10)
Theory suggests that δ1NAlt > 0 , and δ1NAlt > δ1UAlt ≥ 0 .
4.3. Analysis by Private and Public Sector
The motivation for my third approach stems from the assumption that is necessary to
generate the central prediction—a positive effect of the threat of unionization on non-union
wage rates. Note that in deriving the main results, I assume that employers minimize the
expected wage costs. While this assumption is most likely valid for employers in the private
15. 10
sector it may not be valid for employers in the public sector. I therefore follow my first
approach by using only the sample of workers who were employed in the private sector at the
time of a survey. I then compare the results to those obtained based on the sample of workers
who were employed in the public sector. Due to the abovementioned differences in the
maximization problem across employers in the public and private sectors, I expect that the
relation between the unionization threat and the non-union wages is likely to be stronger for
the sample of workers in the private sector.
4.4. Potential Problems
The first potential problem with the above models is the likely heteroscedasticity and
autocorrelation. When the variance of regression residuals depends on the explanatory
variables, serious consequences may occur for OLS and probit estimators. Although the OLS
estimators remain unbiased, the estimated standard errors are wrong. As a result, inferences
and hypotheses tests cannot be relied on. In the probit models, “the maximum likelihood
estimators are inconsistent and the covariance matrix is inappropriate” (Green, 2003; page
679). Therefore, I choose to compute and report heteroscedasticity-robust standard errors.
In addition, in the probit model, the coefficients cannot be interpreted as marginal
effects. Hence, I choose to compute and report marginal effects, which are a non-linear
combination of the regression coefficients. The marginal effects are obtained by calculating
the derivative of the outcome probability with respect to the control variables.
Autocorrelation might be another problem because nine years of data is used in the second
step. I decided to include indicator variables for each year. I also included interaction terms
that allow for the effect of various explanatory variables on wage to differ across years.
16. 11
5. Description of Data and Variables
My analysis draws on monthly Labor Force Surveys (LFS) conducted by Statistics
Canada. I use data collected every January from the year 1998 to 2006. The overall pooled
sample consists of 432,574 observations after I drop observations because of missing
information on union membership or control variables. Among workers in my final sample
291,596 (67.4%) are non-union members and the rest 140,978 (32.6%) are union members;
112,767 (26.1%) are in the public sector and 319,807 (73.9%) in the private sector.
The choice of the independent variables is based on the review of the literature on
determinants of wages. Most importantly, the human capital theory suggests that the
workers’ productivity increases with the worker’s ability and acquired skills (Becker, 1993).
Wages are thereby expected to be correlated with the workers’ attained education (e.g.,
number of years spent in school) and other components of the workers’ human capital (age,
for instance, may measure acquired work experience). Other variables such as demographic
and industry characteristics which might affect the worker’s wage are also included, as
suggested by Lewis (1986). My choice of explanatory variables draws also on Belman and
Heywood (1990) who used race, gender, marital status, education, employment status,
location or region, and industry in their analysis of the effect of unionization on wage
dispersion. Following Farber’s (2005) model, other variables which might affect the worker’s
decision to join the union, such as firm size are also included.
6. Results
6. 1. Determinants of Union Membership in Canada, 1998-2006
The first step to assessing the effect of the threat of unionization on wages of union and
17. 12
non-union workers entails constructing the measure for the unionization threat. As described
in Section 4.1., I estimate the probit model in order to obtain the predicted probability of
becoming a union member. The estimates are reported in Appendix B, Tables 3, 6, and 9.
Table 3 presents the results for the whole sample, while Tables 6 and 9 report results for
workers employed in the public sector and those employed in the private sector, respectively.
The marginal effects instead of the original coefficients are reported.
The results based on the most recent survey in 2006 suggest that both worker-specific
and employer-specific characteristics significantly affect the likelihood of being a union
member. The characteristics associated with a worker that are positively associated with the
likelihood of union membership are gender (male workers are more likely to be union
members) and age (older workers are more likely to be union members). Workers who are
married or have higher attained education, on the other hand, are less likely to be union
members. Firm characteristics also matter in terms of explaining the worker’s propensity to
join the union. For instance, employment in the public sector increases the probability of
being a union member by approximately 36.7% in 2006. Also, there exists a positive
correlation between the firm size and the probability of being a union member. The larger
the firm is, the higher the probability for the worker to become a union member. For instance,
workers employed in a firm with more than five hundred employees are 35.8% more likely to
be union members compared to those in the firm with less than 20 employees in 2006.
In addition, residents of Quebec, British Columbia, Manitoba, and Saskatchewan are
more likely to be union members, as compared to the Albertans. Moreover, industry
dummies are all significant in determining the probability of unionization. The results
suggest that the workers in industries that require lower skills or more labor work are more
18. 13
likely to be unionized. Exceptions are the industries which are highly unionized from the
early days of unionization, for example, health care, education, and public administration.
Similar results, in terms of magnitude and sign, to those found in 2006 are found across
all nine years. However, differences are also found for some explanatory variables. For
instance, in 2000, workers residing in New Foundland, Nova Scotia, and British Columbia
were more likely to be union members compared to Albertan workers. In 1998, married
workers were more likely to be unionized than the unmarried workers. After 1999, the
marital status was negatively associated with the likelihood of union membership.
Separate results for public and private sectors are reported in Tables 6 and 9. For the
private sector, the estimates on major independent variables are similar, in terms of the
magnitudes and the signs, to the ones obtained from the whole sample. For workers in the
public sector, the estimates are different from those in the whole sample; for example, being
male decreases the probability of joining the union, whereas being married increases the
probability of being a union member.
6.2. The Effect of the Unionization Threat on Non-union Wage Rates
I start by analyzing results of the first approach in Section 4. In particular, this approach
uses the predicted probability obtained from the probit model as a measure for the threat of
unionization. The results are presented in Table 4 in Appendix B. Figure 2 in Appendix C
plots the main results reported in Table 4 based on the full sample. In particular, the Figure
plots the estimated coefficients of the threat effect on both the union and non-union wages by
year, organized in the following four panels. Panel A depicts the estimated marginal effects
of the threat effect on wages of union and non-union workers from the regression without the
19. 14
industry and province dummies; Panel B reports the results in which province dummies are
included; in Panel C, controls for the industry are added; and finally, Panel D presents results
from in which the province and industry dummies are included.
As shown in Panel A in Figure 2, Appendix C, the estimates of marginal effects of the
predicted probability of unionization on non-union wages are approximately 30%. The
lowest estimate is at 29.5% in 1998 and the highest is at 34.9% in 2006. Compared to the
effects on non-union wage, those on the union wages are slightly higher, whereas the highest
effect is at 39.5% in 1998 and the lowest at 36% in 2001. In Panel B, in Figure 2, by
including the province dummies to wage regression, the estimated marginal effect of the
predicted probability of unionization on the non-union wages (ranges from 29.3 to 41.8%) is
still smaller compared to the effect on the union wages (ranges from 46.5 to 56.4%); but
effects still have an increasing trend. Figure 2 Panel C adds the industry dummies to the list
of explanatory variables in the wage regressions. The union threat effect on non-union wages
(ranges from 15.1% to 26.9 %) is now higher than that on the union wages (ranges from -2.5
to 23.2%), both showing a decreasing trend.
Finally, in Panel D, Figure 2, by controlling for both province and industry
characteristics, the threat effects are reduced. The estimates of marginal effects of the
unionization threat on the non-union and union wages are decreasing over the period from
27.3% to 24.8%; for instance, an increase in the probability of being a union member will
increase non-union wages by 27.3% in 1998 and by 24.8% in 2006. The threat effect on
union wages is at a lower range, decreasing from 25.9% and 13.2%. The results in both Panel
C and D are consistent with the theory’s predictions.
20. 15
6.3. An Analysis of Unionization Threat in Private and Public Sectors
Estimation results for workers in the public and private sector are presented in Tables 7
and 10, and plotted in Figures 3 and 4. In the private sector, the threat effects on the non-
union wage range from 21.9% in 1998 to 14.9% in 2006. The within- and between-province
variation for the non-union wages is 39.3% to 40.6%, whereas the within- and between-
province and within-industry variation is 33.1% to 12.5%. The threat effects on the union
wages are smaller after controlling for industry. Figure 3 in the Appendix C suggests that
there is a decreasing trend in the threat effects in Panels C and D.
In the public sector, with the full set of control variables, an increase in the probability
of being a union member is estimated to increase non-union wages by 12% to 9%, and 14%
to 11% increase in union wages from 1998 to 2006 (see Figure 4 and Table 7). However,
with fewer controls in the wage regressions, the effects of unionization threat on the wages of
workers in the public sector (both unionized and non-unionized) are ambiguous, because the
signs on the coefficient of the threat are both positive and negative. One explanation for the
observed pattern is as follows. Namely, the private sector aims to minimize the expected
costs of wage payments, so the effects of unionization threat on wages are clear-cut. The
maximization problem for employers in the public sector is more ambiguous. Therefore, the
effect of unionization threat on the public workers’ wage rates cannot be determined.
In conclusion, using the predicted probability of becoming a union member as a
measure of the threat of unionization, I find that the threat effects on union and non-union
wages are both positive. However, the threat effects are higher for non-union workers in the
private sector, but not in the public sector. Importantly, I find this evidence to be sensitive to
the set of control variables that I include when estimating the wage regression.
21. 16
6.4. An alternative Measure for Unionization Threat
In this section, I discuss the results obtained by using as an alternative measure of the
unionization threat. In particular, I use as a measure for the unionization threat an industry-
level union density. The results are reported in Tables 5, 8, and 11, for the full sample, for
workers in the public sector, and for workers in the private sector, respectively. The main
results as they pertain to the unionization threat effect on the union and the non-union wage
rates are depicted in Figures 5 through 7, for the full sample, the sample of workers in the
public sector, and the sample of workers in the private sector, respectively.
The alternative measure for the threat of unionization gives different results in terms of
the magnitude and sign of the threat effects over the years, as compared to the results
obtained when the threat measure was inferred from the probit model. The difference is
particularly pronounced for Panel D (see Figures 5 through 7 in the Appendix). For the whole
sample in Panel D, the threat effects on the non-union wages are greater than the effects on
the union workers only after 2002. Moreover, the threat effects turn negative after 2002. This
finding suggests that with an increase in the unionization threat, the wage rates actually
decrease. For the private sector, the threat effects on the non-union wage are greater then the
effects on the union wage only for 1998 and 1999 fro Panel D. The effects become negative
after 1999. The threat effects on non-union wage rates of public workers are only greater than
that on the union workers in 2000. The effects are negative throughout the nine year period.
7. Conclusion
Using both the predicted probability of being a union member and the industry-level
union density as measures for the threat of unionization, I provide evidence that suggests that
22. 17
the threat of unionization can have a positive impact on wages of non-union workers in
Canada. In particular, an increase in the probability of being a union member is estimated to
result in a 14.9% increase in wages of non-union workers in the private sector, and 10.6% for
non-union workers in 2006 in Canada. The difference between these two percentage terms
measures the effect of the unionization threat on the union-wage gap.
The result supports the theoretical prediction that the threat of unionization has a
positive effect on the non-union wages in private sectors, and a positive or close to negligible
effect on that in public sectors. Further, the results can help explain why the threat of
unionization may tend to result in plant closures mostly in private sectors. More importantly,
though, my findings are shown to be very sensitive to the list of explanatory variables
included in the wage regressions as well as to the definition of the threat of unionization.
Upon restricting the sample to public and private sector, the results show that the threat
effects on the private sector are driving the results reported for the whole sample.
The results in this paper are of importance for several reasons. To my understanding of
the literature, this paper is the first study of the union threat effects on non-union wage rates
in Canada. While a growing literature has explored the union threat effect for the United
States and several European countries, studies using Canadian data have been thus far
restricted to estimating the union wage premium. The results obtained in this paper are
consistent with the theory for certain specifications but not for others. Hence, the conflicting
results reported in related literature regarding the union threat effect on the non-union wage
rates are reaffirmed in this study as well. Overall, the main conclusion I draw from my
analysis is that the effects of unionization threat are not clear cut.
23. 18
References
Belman, D., and Heywood, J. (1990) “Union Membership, Union Organization and the
Dispersion of Wages”, The Review of Economics and Statistics, Vol. 72, No. 1, pp.
148-153.
Becker, G. S. (1993) Human capital: A theoretical and empirical analysis, with special
reference to education, Chicago and London: University of Chicago Press. Third
Edition, pp. 390.
Bianco, A. (2006) “No Union Please, We Are Wal-Mart”, Business Week, Feb.13, 2006.
Accessed at www.businessweek.com/magazine/content/06_07/b3971115.htm.
Blanchflower, D. (2006) “A Cross-Country Study of the Union Membership”, IZA
Working Paper. Accessed at http://ideas.repec.org/p/iza/izadps/dp2016.html.
Chaykowski, R. P., and Slotsve, G. A. (2002) “Earnings Inequality and Unions in
Canada”, British Journal of Industrial Relations, September 2002, Vol. 40, No. 3,
pp. 493-519.
Corneo, G., and Lucifora, C. (1997) “Wage Information Under the Union Threat Effects:
Theory and Empirical Evidence”, Labor Economics, Vol. 4, No. 3, pp. 265-392.
Farber, H. (2005) “Non-union Wage Rates and the Threat of Unionization”, Industrial
and Labor Relations Review, Vol. 58, No. 3, pp. 335-352.
Freeman, R., and Medoff, J. (1981) “The Impact of the Percentage Organized on Union
and Non-union Wages”, Review of Economics and Statistics, Vol. 63, No. 4, pp.
561-572.
Greene, W. H. (2003) Econometric Analysis. New Jersey: Prentice Hall. Fifth Edition.
24. 19
Kahn, L., and Curme, M. (1987) “Union and Non-union Wage Dispersion”, The Review
of Economics and Statistics, Vol. 69, No. 4, pp. 600-607.
Kahn, L. M. (1978) “The Effect of Unions on the Earnings of Non-union Workers”,
Industrial and Labor Relations Review, Vol. 31, No. 2, pp. 205-216.
Kumar, P., and Stengos, T. (1985) “Measuring the Union Relative Wage Impact: A
Methodological Note”, Canadian Journal of Economics, Vol. 18, No. 1, pp. 182-
189.
Lewis, H. (1986) Union Relative Wage Effects: A Survey. Chicago: University of
Chicago Press, 1986.
Leue, C., and Tremblay, C. H. (1993) “A New Econometric Model of the Union Threat
Effects”, Applied Economics, Vol. 25, No. 10, pp. 1329-1336.
MacDonald, G., and Evans, J. C. (1981) “The Size and Structure of Union-Non-union
Wage Differentials in Canada”, Canadian Journal of Economics, Vol. 14, No. 2,
pp. 216-231.
Moore, W. J., and Newman, R. J., and Cunningham, J. (1985) “The Effect of the Extent
of Unionism on Union and Non-union Wages”, Journal of Labor Research, Vol.
6, No. 1, pp. 21-44.
Neumark, D., and Wachter, M. (1995) “Union Effects on Non-union Wages: Evidence
form Panel Data on Industries and Cities”, Industrial and Labor Relations Review,
Vol. 49, No. 2, pp. 20-38.
Pearce, J. (1990) “Tenure, Unions, and the Relationship between Employer Size and
Wages”, Journal of Labor Economics, Vol. 8, No. 2, pp. 251-269.
25. 20
Podgursky, M. (1986) “Unions, Establishment Size, and Intra-Industry Threat Effects”,
Industrial and Labor Relations Review, Vol. 39, No. 2, pp. 277-284.
Rosen, S. (1969) “Trade Union Power, Threat Effects and the Extent of Organization”,
The Review of Economic Studies, Vol. 36, No. 2, pp. 185-196.
26. 21
Appendix A
Table 1: Description of variables
Variable Name Description
Hourly wage before taxes and other deductions, including tips, commissions and
Hrlyearn
bonuses (inferred from questions 205-209 in the Labor Force Survey).
Union status dummy is set to 1 if the worker is a union member and 0 otherwise
Unionmbr
(inferred from question 220 in the Labor Force Survey).
Public sector dummy is set to 1 if the worker works in the public sector and 0
Public
otherwise (inferred from question 115 in the Labor Force Survey).1
A set of age dummies indicating which age group the worker (i.e., the survey’s
Age15-24, Age 25-34,
respondent) belongs to at the time of the survey (inferred from question ANC_Q03
Age 35-44, Age 45-54
in the Labor Force Survey).
Gender dummy is set to 1 if male and 0 otherwise (inferred from question Q01in the
Male
Labor Force Survey).
Marital status dummy is set to 1 if married and 0 otherwise (inferred from question
Married
MSNC_Q01 in the Labor Force Survey).
A set of dummy variables that identify the highest attained level of schooling at the
time of the survey: 1 if no high school or grades<12 is the excluded group; 1 if high
Hisch, Post, Univg
school graduates ( Hisch), some postsecondary (Post), university graduate (Univg)
(inferred from question EDQ01-04 in the Labor Force Survey)
A set of provincial dummies that identify survey respondent’s residence: New
Nfld, pei, ns, nb, que,
Foundland, P.E.I, Nova Scotia, New Brunswick, Quebec, Ontario, Manitoba,
ont, man, sask, bc
Saskatchewan, British Columbia.
A set of dummy variables that identify the size of a firm (in number of employees)
at which a survey respondent worked at the time of the survey (inferred from
Firmsize question Q240 in the Labor Force Survey). The number of employees at all
locations, in four categories: it is 1 if less than 20 employers, 2 if 20 to 99
employers, 3 if 100 to 500 employers, 4 if more than 500 employers.
A set of industry dummy variables indicating the industry the worker works in:
accordingly, the dummy variables represent Forestry, Fishing, Mining, Oil and Gas;
In02, In03, In04,
Utilities; Construction; Manufacturing - durables; Manufacturing - non-durables;
In05, In06, In07,
Wholesale Trade; Retail Trade; Transportation and Warehousing; Finance,
In08, In09, In10,
Insurance, Real Estate and Leasing; Professional, Scientific and Technical Services;
In11, In12, In13,
Management, Administrative and Other Support; Educational Services; Health Care
In14, In15, In16,
and Social Assistance; Information, Culture and Recreation; Accommodation and
In17, In18
Food Services; Other services; Public Administration (inferred from question 115 in
the Labor Force Survey).
1
The public sector includes employees in public administration at the federal, provincial and municipal levels, as well
as in Crown corporations, liquor control boards and other government institutions such as schools (including
universities), hospitals and public libraries. The private sector comprises all other employees and self-employed
owners of businesses, and self-employed persons without businesses.
27. 22
Table 2: Review of literature on the union threat effects
Author Period Country Survey Proxy for the Threat of Unionization Findings
- The union threat effect on non-union wage rate
- Predicted probability of union
found in private sector only
Yi Zuo 1998- Labor Force membership as a function of worker,
Canada - Results are sensitive to definition of the
(this paper) 2006 Surveys job, and firm characteristics
unionization threat and to a set of explanatory
- Industry union density
variables
- Predicted probability of being a union - The effect of the threat of unionization on the non-
1978-
Farber U.S. CPS member as a function of worker, job, union wages is sensitive to set of explanatory
2002
and firm characteristics variables
- Industry-level percentage of employed - The effect of industry-level unionization increases
Pearce 1990 U.S. CPS
with union membership with firm size in the non-union sector
- The percentage of three-digit industry
- No significant effect found of percentage organized
Leue and 1979- employment organized in unions
U.S. EOPP and the predicted threat of unionization on non-
Tremblay 1980 - The probability that a firm is organized
union wages
by a union
- At the industry level, an increase in the percentage
- Industry-level percentage of employed organized reduces the non-union industry wage
Neumark 1973- organized in unions differential
U.S. CPS
and Wachter 1989 - City-level percentage of employed - At the city level, an increase in the percentage
organized in unions organized in union increases the non-union city
wage differential
- Unionization in an industry with fewer union
Moore,
- Industry-level union membership rate members has a significant positive wage effect on
Newman, 1973-
U.S. CPS - Occupation-level union membership non-union workers
and 1979
rate - Unionization within an occupation has no wage
Cunningham
effect on non-union workers
- Large and small non-union employers tend to
- Proportion of an industry’s production respond less to the union threat
Podgursky 1979 U.S. CPS
workers covered by union contracts - Wage at medium-sized non-union employers
increases with the union threat
28. 23
Table 2 (Continued)
Author Period Country Survey Proxy for the Threat of Unionization Findings
- Non-union workers in highly organized markets receive
Freeman 1973- - Industry-level percentage of employed higher wages than those in less unionized industries
U.S. CPS and EEC
and Medoff 1975 covered by collective agreement - In manufacturing, the threat has a strong positive effect on
union wages, but no or a weak effect on non-union wages
- For occupations which are not organized unionization
- Industry-level union membership rate threat has strong impact on non-union wages
Kahn 1967 U.S. SEO - Occupation-level union membership - For occupations which are highly unionized, the within
rate occupation-industry union effect on non-union wages is
negative
- Non-union workers with below-median earnings receive
Census of - Three union dummies based on the
Rosen 1958 U.S. higher wages with unionization, except for managers and
Manufactures percentage organized in union
professionals
- Non-union workers with below-median earnings receive
Heywood Labor Force
1997 U.K. - Industry-level union coverage higher wages with unionization, except for managers and
and Belfield Survey
professionals
- Threat effects are strongly correlated with union density
Corneo and Fedemecanica
1990 Italy - Firm-level union density - Threat effects on wages are significant with an
Lucifora Survey
intermediate level of union density
Abbreviation used in Table 2: CPS - Current Population Survey; EOPP - Employment Opportunity Pilot Project; EEC - Expenditures for Employee Compensation
Surveys; SEO - Survey of Economic Opportunity.
29. 24
Appendix B Estimation Results
Table 3: Determinants of union membership in Canada, 1998 – 2006 (all sample)
Dataset: Canadian Labor Force Survey
Sample: All observations: Analysis by year
2006 2005 2004 2003 2002 2001 2000 1999 1998
Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal
effect effect effect effect effect effect effect effect effect
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8) (9)
Public 0.367 0.399 0.361 0.387 0.393 0.323 0.332 0.319 0.303
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Male 0.045 0.030 0.046 0.035 0.046 0.050 0.052 0.060 0.051
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)***
Married -0.013 -0.011 0.001 -0.014 -0.005 -0.006 -0.013 0.000 0.011
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000) (0.000)***
Age base category: Age 55 + … … … … … … … … …
… … … … … … … … …
Age between 15 and 24 -0.117 -0.098 -0.101 -0.119 -0.121 -0.130 -0.153 -0.120 -0.136
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.001)***
Age between 25 and 34 -0.028 -0.047 -0.029 -0.039 -0.050 -0.050 -0.050 -0.030 -0.028
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.001)*** (0.001)***
Age between 35 and 44 0.001 -0.008 -0.005 -0.012 -0.002 -0.006 -0.008 0.008 -0.001
(0.000) (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.001)*** (0.001)*** (0.001)**
Age between 45 and 54 0.032 0.018 0.022 0.026 0.015 0.022 0.018 0.030 0.039
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
High school dropout (excluded group) … … … … … … … … …
… … … … … … … … …
11 to 13 years of schooling/graduate -0.009 -0.020 -0.020 -0.022 -0.005 -0.022 -0.014 -0.012 -0.027
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)***
At least some postsecondary diploma -0.043 -0.034 -0.041 -0.043 -0.020 -0.032 -0.026 -0.036 -0.041
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)***
University: bachelors or graduate degree -0.122 -0.122 -0.141 -0.132 -0.123 -0.130 -0.118 -0.123 -0.139
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)***
30. 25
Table 3 (Continued)
Dataset: Canadian Labor Force Survey
Sample: All observations: Analysis by year
2006 2005 2004 2003 2002 2001 2000 1999 1998
Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal
effect effect effect effect effect effect effect effect effect
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8) (9)
Firm with less than 20 employees (base) … … … … … … … … …
… … … … … … … … …
Firm with 20-99 employees 0.187 0.226 0.221 0.208 0.228 0.207 0.216 0.233 0.234
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Firm with 100-500 employees 0.343 0.368 0.382 0.376 0.351 0.364 0.384 0.385 0.371
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Firm with more than 500 employees 0.358 0.374 0.383 0.394 0.394 0.382 0.399 0.409 0.396
(0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.000)***
Additional control variables:
Industry X X X X X X X X X
Province X X X X X X X X X
Observations 50085 49634 46846 49461 48268 48485 46963 47036 45798
Pseudo R-squared 0.316 0.316 0.313 0.323 0.326 0.314 0.303 0.318 0.305
Robust standard errors in parentheses
*significant at 10%; ** significant at 5%; *** significant at 1%
31. 26
Table 4: Effect of predicted probability of unionization on union/non-union wage rates, 1998 – 2006 (all sample)
Dataset: Canadian Labor Force Survey
Panel D-Control for
Panel A-No control for Panel B-Control for Panel C-Control for
both industry and
industry or province province industry
province
Sample: Non-union Union Non-union Union Non-union Union Non-union Union
workers workers workers workers workers workers workers workers
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8)
The threat of unionization in 2006 0.349 0.376 0.418 0.564 0.151 -0.025 0.248 0.132
(0.017)*** (0.019)*** (0.017)*** (0.021)*** (0.022)*** (-0.026) (0.025)*** (0.049)***
The threat of unionization in 2005 0.338 0.375 0.398 0.548 0.182 0.003 0.264 0.142
(0.015)*** (0.016)*** (0.016)*** (0.018)*** (0.019)*** (-0.022) (0.021)*** (0.040)***
The threat of unionization in 2004 0.370 0.394 0.415 0.553 0.210 0.041 0.278 0.165
(0.014)*** (0.015)*** (0.014)*** (0.017)*** (0.017)*** (0.019)** (0.019)*** (0.035)***
The threat of unionization in 2003 0.326 0.376 0.360 0.514 0.197 0.068 0.249 0.171
(0.013)*** (0.014)*** (0.013)*** (0.015)*** (0.015)*** (0.017)*** (0.016)*** (0.029)***
The threat of unionization in 2002 0.315 0.366 0.341 0.486 0.214 0.085 0.257 0.168
(0.013)*** (0.013)*** (0.013)*** (0.014)*** (0.015)*** (0.016)*** (0.016)*** (0.027)***
The threat of unionization in 2001 0.321 0.360 0.345 0.480 0.236 0.112 0.274 0.192
(0.013)*** (0.014)*** (0.013)*** (0.015)*** (0.015)*** (0.017)*** (0.016)*** (0.029)***
The threat of unionization in 2000 0.329 0.380 0.340 0.485 0.261 0.161 0.283 0.222
(0.014)*** (0.015)*** (0.014)*** (0.016)*** (0.017)*** (0.019)*** (0.018)*** (0.035)***
The threat of unionization in 1999 0.324 0.374 0.333 0.456 0.268 0.193 0.287 0.235
(0.015)*** (0.016)*** (0.015)*** (0.018)*** (0.018)*** (0.022)*** (0.020)*** (0.040)***
The threat of unionization in 1998 0.295 0.395 0.293 0.465 0.269 0.232 0.273 0.259
(0.016)*** (0.019)*** (0.016)*** (0.021)*** (0.021)*** (0.026)*** (0.023)*** (0.049)***
Public sector -0.003 -0.099 0.024 -0.092 -0.026 -0.047 -0.003 -0.026
-0.009 (0.008)*** (0.009)*** (0.009)*** (0.011)** (0.011)*** (-0.012) (-0.019)
Male 0.250 0.174 0.252 0.170 0.217 0.137 0.219 0.138
(0.003)*** (0.004)*** (0.003)*** (0.004)*** (0.003)*** (0.005)*** (0.003)*** (0.005)***
Married 0.099 0.045 0.105 0.048 0.085 0.038 0.091 0.042
(0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)***
Age base category: Age 55 + … … … … … … … …
… … … … … … … …
32. 27
Table 4 (Continued)
Dataset: Canadian Labor Force Survey
Panel D-Control for
Panel A-No control for Panel B-Control for Panel C-Control for
both industry and
industry or province province industry
province
Sample: Non-union Union Non-union Union Non-union Union Non-union Union
workers workers workers workers workers workers workers workers
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8)
Age between 15 and 24 -0.367 -0.39 -0.358 -0.364 -0.321 -0.374 -0.313 -0.358
(0.008)*** (0.012)*** (0.008)*** (0.012)*** (0.008)*** (0.011)*** (0.008)*** (0.013)***
Age between 25 and 34 -0.105 -0.123 -0.095 -0.110 -0.103 -0.129 -0.094 -0.118
(0.008)*** (0.008)*** (0.007)*** (0.008)*** (0.007)*** (0.008)*** (0.007)*** (0.008)***
Age between 35 and 44 0.017 -0.033 0.025 -0.022 0.008 -0.039 0.016 -0.030
(0.008)** (0.007)*** (0.007)*** (0.007)*** (-0.007) (0.007)*** (0.007)** (0.007)***
Age between 45 and 54 0.042 -0.001 0.050 0.005 0.037 -0.003 0.045 0.004
(0.008)*** (-0.007) (0.008)*** (-0.007) (0.008)*** (-0.007) (0.007)*** (-0.007)
High school dropout (excluded group) … … … … … … … …
… … … … … … … …
11 to 13 years of schooling/graduate 0.153 0.123 0.141 0.111 0.136 0.123 0.125 0.109
(0.005)*** (0.006)*** (0.004)*** (0.006)*** (0.004)*** (0.006)*** (0.004)*** (0.006)***
At least some postsecondary diploma 0.251 0.206 0.241 0.203 0.217 0.203 0.209 0.197
(0.004)*** (0.006)*** (0.004)*** (0.005)*** (0.004)*** (0.006)*** (0.004)*** (0.006)***
University: bachelors or graduate degree 0.554 0.463 0.531 0.458 0.495 0.453 0.476 0.443
(0.006)*** (0.007)*** (0.006)*** (0.007)*** (0.007)*** (0.008)*** (0.007)*** (0.011)***
Firm with less than 20 employees (base) … … … … … … … …
… … … … … … … …
Firm with 20-99 employees 0.067 -0.081 0.058 -0.092 0.062 -0.029 0.053 -0.031
(0.005)*** (0.012)*** (0.005)*** (0.012)*** (0.005)*** (0.013)** (0.005)*** (0.015)**
Firm with 100-500 employees 0.087 -0.094 0.075 -0.113 0.072 -0.029 0.062 -0.033
(0.006)*** (0.013)*** (0.006)*** (0.013)*** (0.006)*** (0.014)** (0.007)*** (-0.021)
Firm with more than 500 employees 0.123 -0.057 0.107 -0.09 0.114 0.012 0.100 -0.001
(0.005)*** (0.013)*** (0.005)*** (0.013)*** (0.006)*** (-0.015) (0.007)*** (-0.025)
33. 28
Table 4 (Continued)
Dataset: Canadian Labor Force Survey
Panel D-Control for
Panel A-No control for Panel B-Control for Panel C-Control for
both industry and
industry or province province industry
province
Sample: Non-union Union Non-union Union Non-union Union Non-union Union
workers workers workers workers workers workers workers workers
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8)
Additional control variables:
Industry X X X X
Province X X X X
Year X X X X X X X X
Interaction terms with year X X X X X X X X
Constant 2.039 2.453 2.082 2.464 2.001 2.403 2.043 2.425
(0.007)*** (0.012)*** (0.008)*** (0.013)*** (0.013)*** (0.038)*** (0.013)*** (0.039)***
Observations 291596 140978 291596 140978 291596 140978 291596 140978
Adjusted R-squared 0.412 0.328 0.443 0.362 0.472 0.375 0.499 0.407
Robust standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%
34. 29
Table 5: Effect of union density on union/non-union wage rates, 1998 – 2006 (All sample)
Dataset: Canadian Labor Force Survey
Panel A Panel B Panel C Panel D
Sample: Non-union Union Non-union Union Non-union Union Non-union Union
workers workers workers workers workers workers workers workers
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8)
Industry-level union density in 2006 0.234 0.373 0.279 0.337 -0.167 -0.206 -0.132 -0.204
(0.001)*** (0.001)*** (0.013)*** (0.017)*** (0.153) (0.145) (0.148) (0.142)
Industry-level union density in 2005 0.251 0.340 0.291 0.320 -0.101 -0.153 -0.071 -0.145
(0.001)*** (0.001)*** (0.013)*** (0.015)*** (0.127) (0.117) (0.124) (0.114)
Industry-level union density in 2004 0.271 0.331 0.313 0.325 -0.069 -0.118 -0.046 -0.107
(0.001)*** (0.001)*** (0.012)*** (0.016)*** (0.103) (0.101) (0.100) (0.098)
Industry-level union density in 2003 0.235 0.350 0.277 0.301 -0.065 -0.070 -0.046 -0.060
(0.001)*** (0.001)*** (0.012)*** (0.014)*** (0.082) (0.081) (0.080) (0.079)
Industry-level union density in 2002 0.224 0.324 0.282 0.308 -0.023 -0.020 -0.009 -0.009
(0.001)*** (0.001)*** (0.012)*** (0.014)*** (0.064) (0.069) (0.062) (0.066)
Industry-level union density in 2001 0.235 0.331 0.299 0.286 0.016 0.020 0.028 0.035
(0.001)*** (0.001)*** (0.012)*** (0.015)*** (0.055) (0.065) (0.052) (0.062)
Industry-level union density in 2000 0.260 0.326 0.297 0.317 0.034 0.057 0.038 0.084
(0.001)*** (0.001)*** (0.012)*** (0.016)*** (0.055) (0.074) (0.052) (0.071)
Industry-level union density in 1999 0.273 0.284 0.296 0.239 0.077 0.086 0.081 0.108
(0.001)*** (0.001)*** (0.012)*** (0.016)*** (0.069) (0.084) (0.066) (0.081)
Industry-level union density in 1998 0.254 0.347 0.270 0.295 -0.167 -0.206 -0.132 -0.204
(0.001)*** (0.001)*** (0.013)*** (0.016)*** (-0.088) (0.103)* (-0.085) (0.099)**
Public sector 0.020 -0.002 0.039 0.005 0.068 0.034 0.087 0.056
(0.000)*** (0.000)*** (0.007)*** -0.005 (0.008)*** (0.006)*** (0.008)*** (0.006)***
Male 0.237 0.183 0.262 0.192 0.228 0.152 0.230 0.153
(0.000)*** (0.000)*** (0.003)*** (0.004)*** (0.003)*** (0.004)*** (0.003)*** (0.004)***
Married 0.091 0.049 0.104 0.050 0.086 0.037 0.092 0.042
(0.000)*** (0.000)*** (0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)*** (0.004)***
Age base category: Age 55 + … … … … … … … …
… … … … … … … …
Age between 15 and 24 -0.399 -0.432 -0.381 -0.438 -0.351 -0.412 -0.342 -0.400
(0.000)*** (0.001)*** (0.008)*** (0.011)*** (0.007)*** (0.011)*** (0.007)*** (0.010)***
Age between 25 and 34 -0.107 -0.137 -0.103 -0.129 -0.113 -0.138 -0.103 -0.129
(0.000)*** (0.001)*** (0.007)*** (0.008)*** (0.007)*** (0.008)*** (0.007)*** (0.007)***
35. 30
Table 5 (Continued)
Panel A Panel B Panel C Panel D
Sample: Non-union Union Non-union Union Non-union Union Non-union Union
workers workers workers workers workers workers workers workers
Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8)
Age between 35 and 44 0.020 -0.028 0.024 -0.024 0.007 -0.039 0.015 -0.029
(0.000)*** (0.000)*** (0.007)*** (0.007)*** -0.007 (0.007)*** (0.007)** (0.007)***
Age between 45 and 54 0.046 0.014 0.055 0.017 0.044 0.006 0.051 0.012
(0.000)*** (0.000)*** (0.008)*** (0.007)** (0.008)*** (-0.007) (0.007)*** (0.007)*
High school dropout (excluded group) … … … … … … … …
… … … … … … … …
11 to 13 years of schooling/graduate 0.161 0.117 0.136 0.098 0.132 0.117 0.121 0.104
(0.000)*** (0.000)*** (0.004)*** (0.006)*** (0.004)*** (0.006)*** (0.004)*** (0.006)***
At least some postsecondary diploma 0.255 0.196 0.232 0.188 0.211 0.194 0.204 0.188
(0.000)*** (0.000)*** (0.004)*** (0.006)*** (0.004)*** (0.006)*** (0.004)*** (0.005)***
University: bachelors or graduate degree 0.524 0.385 0.496 0.384 0.467 0.416 0.449 0.404
(0.000)*** (0.000)*** (0.006)*** (0.007)*** (0.006)*** (0.007)*** (0.006)*** (0.007)***
Firm with less than 20 employees (base) … … … … … … … …
… … … … … … … …
Firm with 20-99 employees 0.083 -0.034 0.080 -0.017 0.087 0.014 0.079 0.016
(0.000)*** (0.001)*** (0.004)*** (-0.012) (0.004)*** (-0.012) (0.004)*** (-0.011)
Firm with 100-500 employees 0.129 -0.013 0.124 0.018 0.124 0.046 0.115 0.048
(0.000)*** (0.001)*** (0.005)*** (-0.011) (0.005)*** (0.011)*** (0.005)*** (0.011)***
Firm with more than 500 employees 0.193 0.055 0.180 0.075 0.186 0.108 0.172 0.103
(0.000)*** (0.001)*** (0.004)*** (0.010)*** (0.004)*** (0.011)*** (0.004)*** (0.010)***
Additional control variables:
Industry X X X X
Province X X X X
Year X X X X X X X X
Interaction terms with year X X X X X X X X
Constant 2.06 2.399 2.029 2.358 2.004 2.362 2.031 2.346
(0.000)*** (0.001)*** (0.009)*** (0.014)*** (0.013)*** (0.038)*** (0.014)*** (0.038)***
Observations 291596 140978 291596 140978 291596 140978 291596 140978
Adjusted R-squared 0.403 0.340 0.446 0.360 0.472 0.375 0.498 0.406
Robust standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%
36. 31
Table 6: Determinants of union membership in Canada, 1998 – 2006 (Public Sector)
Dataset: Canadian Labor Force Survey
Sample: All observations: Analysis by year
2006 2005 2004 2003 2002 2001 2000 1999 1998
Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal Marginal
effect effect effect effect effect effect effect effect effect
(S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.) (S.E.)
Variable name: (1) (2) (3) (4) (5) (6) (7) (8) (9)
Male -0.046 -0.052 -0.037 -0.053 -0.046 -0.027 -0.039 -0.005 -0.035
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Married 0.016 -0.005 0.005 -0.012 0.003 0.000 -0.011 -0.005 0.047
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (-0.001) (0.001)*** (0.001)*** (0.001)***
Age base category: Age 55 + … … … … … … … … …
… … … … … … … … …
Age between 15 and 24 -0.234 -0.212 -0.168 -0.244 -0.295 -0.287 -0.354 -0.26 -0.317
(0.001)*** (0.002)*** (0.002)*** (0.002)*** (0.002)*** (0.002)*** (0.002)*** (0.002)*** (0.002)***
Age between 25 and 34 0.033 0.007 0.036 -0.004 -0.031 0.008 0.008 0.056 -0.011
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Age between 35 and 44 0.044 0.045 0.030 -0.005 0.017 0.028 0.024 0.079 0.011
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Age between 45 and 54 0.042 0.021 0.040 0.026 0.010 0.017 0.029 0.075 0.06
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
High school dropout (excluded group) … … … … … … … … …
… … … … … … … … …
11 to 13 years of schooling/graduate 0.027 0.031 0.031 0.006 0.069 0.027 0.016 0.003 0.028
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
At least some postsecondary diploma 0.007 0.032 0.035 0.015 0.064 0.049 0.012 0.034 0.03
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
University: bachelors or graduate degree -0.059 -0.049 -0.053 -0.067 -0.012 -0.047 -0.082 -0.038 -0.059
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Firm with less than 20 employees (base) … … … … … … … … …
… … … … … … … … …
Firm with 20-99 employees 0.136 0.166 0.160 0.172 0.165 0.189 0.225 0.205 0.207
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
Firm with 100-500 employees 0.236 0.229 0.251 0.249 0.226 0.262 0.307 0.273 0.285
(0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)***
