Publications

    Analytical Studies Branch Research Paper Series

    Source-country Female Labour Force Participation and the Wages of Immigrant Women in Canada

    Source-country Female Labour Force Participation and the Wages of Immigrant Women in Canada

    by Kristyn Frank and Feng Hou
    Social Analysis and Modelling Division, Statistics Canada

    Skip to text

    Text begins

    Abstract

    Previous studies have found a strong association between source-country female labour force participation rates and immigrant women’s labour market activity in the host country. This association has been interpreted as the lasting influence of source-country gender role attitudes. These studies, however, do not measure gender-role attitudes directly and examine labour market activities only in terms of labour force participation in the host country. Employing data from the 2006-to-2012 Labour Force Survey and the 1994-2007 World Values Survey, this analysis includes both source-country female labour force participation rates and an explicit measure of source-country gender role attitudes to account for group differences in immigrant women’s labour market outcomes in Canada across several measures. The results indicate that the effect of source-country gender role attitudes on immigrant women’s labour market participation in Canada works through its relationship with source-country female labour force participation rates. Moreover, source-country female labour force participation rates are positively correlated with immigrant women’s wages in Canada independent of source-country gender role attitudes. This is mostly accounted for by occupational and industry differences, as immigrant women from countries with high female labour force participation rates are more concentrated in higher-paying industries and occupations in Canada than immigrant women from countries with low levels of female labour force participation.

    Executive summary

    Previous studies have found a strong association between source-country female labour force participation rates and immigrant women’s labour force participation in the host country. This relationship is interpreted as the enduring influence of source-country gender-role attitudes on immigrant women’s labour market activity. However, the assumption that source-country female labour force participation levels closely capture cultural gender-role attitudes has not been carefully examined. Furthermore, little is known about how source-country characteristics might be correlated with immigrant women’s labour market outcomes after entering the host country’s labour market.

    This paper extends the current literature by addressing three questions: What is the relationship between source-country gender-role attitudes and source-country female labour force participation? Does the relationship between the source-country female labour force participation rates and immigrant women’s labour force participation in the host country persist when source-country gender-role attitudes are taken into account? Are source-country female labour force participation rates and source-country gender-role attitudes associated with immigrant women’s wages in Canada?

    Data from the Labour Force Survey and the World Values Survey are used. The combined May and November files of Statistics Canada’s 2006-to-2012 Labour Force Survey provide information on the labour market outcomes of women who immigrated to Canada as adults and who, at the time of the survey, were aged 25 to 64 and had been living in Canada for at least one year. The data for the source-country gender-role attitudes are derived from the World Values Survey and source-country labour force participation rates are obtained from the World Bank Data Indicators on social development.

    The results indicate that source-country female labour force participation rates and source-country gender-role attitudes are only moderately correlated. Moreover, although both of these source-country variables are correlated with immigrant women’s labour force participation in Canada when entered separately into a regression model, only the source-country female labour force participation rate remains statistically significant when both are included. This suggests that the relationship between source-country gender-role attitudes and immigrant women’s labour force participation in the host country functions through the moderate correlation between source-country gender-role attitudes and source-country female labour force participation rates.

    The wage analysis, however, indicates that source-country gender-role attitudes and source-country female labour force participation influence immigrant women’s labour market performance through different pathways. Women from countries with more egalitarian gender-role attitudes have higher wages in Canada than immigrant women from nations with less egalitarian attitudes. Additionally, a positive relationship is observed between the source-country female labour force participation rate and immigrant women’s wages in the host country. This relationship is largely explained by occupational and industrial allocation: women from countries with higher levels of female labour force participation are more concentrated in higher paying occupations and industries. These results suggest that women from nations with high female labour force participation rates may navigate the host country’s labour market differently than other immigrant women, resulting in higher-paying employment. These findings indicate that the relationship between the source-country female labour force participation rate and immigrant women’s labour market outcomes extends beyond cultural gender-role attitudes.

    1 Introduction

    Immigration research consistently shows that labour market outcomes differ between immigrants from different source countries or source regions (Boyd and Thomas 2002; Picot 2004; Reitz and Sklar 1997; Thompson 2000). Often such studies measure these effects using country- or region-of-origin dummy variables, obscuring the varying influences that a range of national characteristics may have (Preston and Giles 2004). This study moves beyond country-of-origin indicator variables to assess specific measurements of source-country characteristics, providing a more nuanced understanding of the association between national characteristics and immigrant women’s labour market activities in Canada.

    The importance of national level characteristics is central in cross-national comparative studies of women’s paid and unpaid work. Such studies have investigated the potential effects of national attributes, such as the level of economic development, the quality of education, and various cultural factors on women’s housework, paid labour, and marital gender equality across countries (e.g., Batalova and Cohen 2002; Fortin 2005; Fuwa 2004; Kan, Sullivan and Gershuny 2011; Knudsen and Wærness 2008; Nieuwenhuis, Need and Van der Kolk 2012; Yodanis 2005). These studies suggest that cultural factors play a significant role in explaining women’s paid and unpaid labour activities.

    Studies examining the relationship between national characteristics and immigrant women’s labour activities conclude that source-country female labour force participation levels closely capture gender-role attitudes, and that the observed association between source-country female labour force participation and immigrant women’s subsequent labour market activities in the host country primarily reflects the long-lasting effect of gender-role attitudes formed early in life. However, this assumption has not been carefully examined. An important feature of this study is the inclusion of both source-country female labour force participation rates and an explicit measure of source-country gender-role attitudes. The inclusion of both allows us to assess how each of these variables is correlated with immigrant women’s subsequent labour market outcomes in Canada across several measures.

    The results indicate that source-country female labour force participation rates and source-country gender-role attitudes are only moderately correlated. Moreover, although both of these source-country variables are correlated with immigrant women’s labour force participation in Canada when entered separately into a regression model, only the source-country female labour force participation rate remains statistically significant when both are included. This suggests that the relationship between source-country gender-role attitudes and immigrant women’s labour force participation in the host country functions through the moderate correlation between source-country gender-role attitudes and source-country female labour force participation rates.

    The wage analysis, however, indicates that source-country gender-role attitudes and source-country female labour force participation influence immigrant women’s labour market performance through different pathways. Women from countries with more egalitarian gender-role attitudes have higher wages in Canada than immigrant women from nations with less egalitarian attitudes. Additionally, a positive relationship is observed between the source-country female labour force participation rate and immigrant women’s wages in the host country. This relationship is largely explained by occupational and industrial allocation: women from countries with higher levels of female labour force participation are more concentrated in higher paying occupations and industries. These results suggest that women from nations with high female labour force participation rates may navigate the host country’s labour market differently than other immigrant women, resulting in higher-paying employment. These findings indicate that the relationship between the source-country female labour force participation rate and immigrant women’s labour market outcomes extends beyond cultural gender-role attitudes.

    This paper is organized into six sections. Section 2 surveys the research literature that is relevant to assessing the relationship between source-country characteristics and immigrant women’s labour market outcomes. Section 3 addresses the data, measures and methods. Section 4 provides the descriptive and multivariate analysis results for the study. Section 5 discusses the findings and concludes.

    2 Source-country characteristics and immigrant women’s labour market outcomes

    Research examining the integration of immigrant women indicates that gender-related source-country factors are correlated with immigrant women’s labour force participation in the host country. Studies examining this relationship conclude that a "permanent, portable" cultural factor continues to influence women’s decision to participate in paid labour after migration (Antecol 2000, p. 419; Blau, Kahn and Papps 2011; Frank and Hou 2013). This is attributed to the enduring influence of source-country gender role attitudes. However, the assumption that female labour force participation levels closely capture cultural gender role attitudes has not been carefully examined.

    These findings also raise additional questions regarding the impact of source-country gender-role attitudes and female labour force participation rates on other labour market outcomes of immigrant women, such as occupational attainment and wages in the host country. These outcomes have not received attention in the research literature. If source-country female labour force participation not only reflects gender-role attitudes but also contributes to human capital and job search skills, its influence may extend beyond labour force participation and also affect the occupational attainment and wages of immigrant women in Canada.

    Previous research suggests that women who grow up in nations with high levels of female labour force participation often have different expectations of their future involvement in the labour force than women from other nations. Individuals who expect to engage in paid labour throughout their adult lives are more likely to be “career oriented” and to invest more in training and skill development (Blau, Kahn and Papps 2008, p. 2; Polachek 2004). Women with greater labour force involvement prior to migration may also develop greater “non-tangible” human capital that facilitates access to higher paying employment, such as leadership, cooperativeness, flexibility and “re-trainability”, as well as increased familiarity with effective job search methods. (Blau, Kahn and Papps 2008; David and Lopez 2001, p. 2).

    Further examination of the relationship between source-country gender role attitudes and immigrant women’s labour market outcomes is also needed. Research examining the association between gender-role attitudes and women’s earnings typically focuses on the gender wage gap, attributing differences to sex-typed occupational choice (England 1992; Reskin and Roos 1990). However, some researchers have found that this relationship persists when controlling for occupation, reporting that more ‘traditional’ gender-role attitudes (i.e., male breadwinner, female caregiver) are associated with lower earnings net of occupational differences. They argue that women who demonstrate more traditional gender-role attitudes may be less assertive or less motivated to pursue raises and promotions (Christie-Mizell 2006; Firestone, Harris and Lambert 1999; Judge and Livingston 2008; Stickney and Konrad 2007) and devote less time or effort to their paid work to “conserve energy” for child-rearing and household tasks (Stickney and Konrad 2007, p. 803). This may decrease both their productivity in the workplace and their opportunities to advance to higher-paying positions, resulting in lower earnings. Furthermore, Judge and Livingston (2008) assert that women who place greater value on household responsibilities over their financial role in the family may be more satisfied with lower pay than women with more egalitarian gender attitudes. Thus, earnings differences between women may result despite similar occupational choices.

    3 Data, measures and methods

    3.1 Data

    This study uses the combined May and November, 2006-to-2012 files of the Labour Force Survey (LFS) collected by Statistics Canada. The LFS is Canada’s official source of monthly estimates of total employment and unemployment. It collects information on the labour market activities of the population aged 15 and older, excluding residents of collective dwellings, Aboriginal settlements, and full-time members of the Canadian Forces. It also provides monthly information on the wages of employees since 1997 and immigration status since 2006. The monthly survey contains information on more than 120,000 individuals, or close to one half of 1% of the Canadian population aged 15 and older. In any given month, the LFS sample is divided into six rotation panels that were recruited into the survey one month apart. Each rotation panel remains in the survey for six consecutive months, so up to five-sixths of the sample respondents overlap in two consecutive months. It is only when they are six months apart that the two monthly surveys have no overlap of sample respondents. Therefore, the combined May and November data used in this study contain all unique respondents in a given year.

    The selected sample for the study consists of individuals who immigrated to Canada as adults (aged 20 or over at the year of landing) to ensure a long period of exposure to their source country’s culture and labour market before emigration. Immigrants who arrived before the 1990s were excluded from the sample because most of them had passed their prime working age by the end of the 2000s. Immigrants who had been living in Canada less than one year were also excluded so that immigrant respondents had at least one year to adjust to the Canadian labour market. Additional sample selection criteria were that immigrants were aged 25 to 64 at the time of the survey and were not permanently unable to work. Finally, to merge individual-level data with the source-country gender-role attitudes data derived from the World Values Survey (WVS), immigrants from source countries that did not participate in the WVS over the 1994-to-2007 period were excluded (see Subsection 4.2 for details). This exclusion reduced the LFS sample size of immigrant women by about 4,900, or 14%.Note 1 The final immigrant sample in the study comprises 29,811 women. This sample is used to study immigrant labour force participation. A subsample of 17,089 immigrant women who were paid employees and had positive wages in the LFS reference week is used to model immigrant wages.

    Source-country attributes were appended to the individual-level data for the immigrant sample selected from the LFS through the use of detailed country of birth.Note 2 As a result, immigrants in the LFS were assigned a set of source-country attributes measured at their year of arrival in Canada.

    3.2 Measures

    Two outcome variables are examined in this study: the first is labour force participation. It is coded as 1 if a respondent is either employed or unemployed, or 0 if the respondent is not in the labour force. Previous studies generally use this outcome to examine the association between source-country gender roles and immigrant women’s labour market activities in the host country. While a significant relationship has been consistently reported between these variables, we re-examine this outcome to ensure that it is replicated in our data.Note 3 The second outcome variable is log hourly wages that workers usually received in their main job. Hourly wages are calculated from the information on weekly wages/salary and usual paid work hours per week.

    The focal explanatory variable is the source-country female labour force participation rate. These rates are calculated for individuals aged 15 and older.Note 4 An alternative measure is the ratio of the female labour force participation rate to the male rate. This relative measure can mitigate the impact of possible cross-country differences in the definition and measurement of labour force participation, as such issues affect women’s and men’s labour force participation rates similarly within countries (Antecol 2000; Blau, Kahn and Papps 2011). The relative measure can also alleviate the effects of cross-country differences in socio-demographic and institutional labour market factors that affect the overall labour force participation rates for both women and men. However, model estimates based on these two alternative measures yield the same findings. Since the coefficient of female labour force participation rates can be interpreted more intuitively than the alternative, only the results using the former measure are presented here.

    The key control variable is an explicit measure of source-country gender-role attitudes derived from the World Values Survey (WVS). The WVS collects measures of people’s beliefs, values, and attitudes from nationally representative samples. The questions related to gender-role attitudes are available for 83 countries that participated in at least one of waves 3 to 5 of the WVS over the 1994-to-2007 period.Note 5 Three questions are used to derive the gender-role attitudes scale:Note 6

    1. Do you agree or disagree with the following statement?
      When jobs are scarce, men should have more right to a job than women.
    2. Do you agree strongly, agree, disagree, or disagree strongly with this statement?
      A university education is more important for a boy than for a girl.
    3. Do you agree strongly, agree, disagree, or disagree strongly with this statement?
      On the whole, men make better political leaders than women do.Note 7

    These statements reflect beliefs about whether women should have equal access to jobs and higher education, and whether they have the ability for leadership roles. The values of the gender-role attitudes scale range from 1.83 to 3.59, with a mean of 2.70. Only one unique value of gender-role attitudes is assigned to each source country.Note 8 This scale has good reliability, with a Cronbach Alpha of 0.65.

    Three additional source-country characteristics are included as controls for factors that may be related to immigrants’ readiness for participating in the host-country labour market. The first is gross domestic product (GDP) per capita (in 2005 U.S. dollars) for immigrants’ year of landing in Canada. This is used as an indicator of the source country’s overall level of economic development.Note 9 Immigrants from more developed countries are more likely to participate in Canada’s labour market because of similarities in economic structure between their source country and Canada. The second is a dummy variable indicating whether either English or French is an official language in the source country.Note 10 The third variable is a dummy variable indicating whether the source country is a Western country. These three variables are likely correlated with source-country female labour force participation rates and with immigrants’ preparedness for work in the host-country labour market.

    At the individual level, the following demographic variables are included: level of education, age at immigration, years since landing and its squared term, geographic region of residence, and survey years. Levels of education are coded as four dummy variables: less than high school, high school graduation, some postsecondary education, and graduate degree, with bachelor’s degree as the reference group. The geographic region variable distinguishes the eight largest metropolitan areas in Canada (Toronto, Montréal, Vancouver, and the next five largest metropolitan areas combined—Ottawa, Calgary, Edmonton, Winnipeg, and Hamilton); the Atlantic region; Quebec, excluding Montréal; Ontario, excluding Toronto, Ottawa and Hamilton; Saskatchewan and Manitoba combined, excluding Winnipeg; Alberta, excluding Calgary and Edmonton; and British Columbia, excluding Vancouver. Toronto is the common reference category. Survey years are coded as a series of dummy variables for the LFS years used in this study. Marital status, spouse’s employment status if married, and number of children younger than age 6 are also included in the model predicting labour force participation. Full-time/part-time status (1 = full time; 0, part time) and job tenure (in years) with the current employer are additional variables in the models for wages.

    In addition to the above variables, industrial and occupational distributions are included as key variables in the wage model to test whether part of the association between source-country female labour force participation levels and immigrant wages works through immigrants’ abilities to navigate job opportunities in a modern economy. Industrial distribution, based on the North American Industry Classification System, is coded into five categories: agriculture; primary industries, utility, construction and manufacturing; trade and transportation; finance, professional and business services; and other services. Occupational distribution is based on the first two digits of the 2006 National Occupational Classification for Statistics (NOC-S) after combining a few groups with very small sample sizes for immigrant women in the data.Note 11 There are 41 categories in the occupational distribution variable used in the model.

    3.3 Methods

    For all working-age immigrant women, three probit models are estimated to predict the likelihood of labour force participation. The models are specified as

    P( Y ij =  1|x ) = Φ( α g X j g  + β X j c  + γ X i )     (1) MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaaeiuamaabm aabaacbiGaa8xwamaaBaaaleaacaWFPbGaa8NAaiaa=bcaaeqaaOGa eyypa0JaaiiOaiaabccacaqGXaGaaiiFaiaa=HhaaiaawIcacaGLPa aacaqGGaGaeyypa0JaaeiiaiabfA6agnaabmaabaGaeqySde2aaSba aSqaaiaa=DgaaeqaaOGaa8hwamaaDaaaleaadaahaaadbeqaaiaa=P gaaaaaleaacaWFNbaaaOGaaeiiaiabgUcaRiaabccacqaHYoGycaWF ybWaa0baaSqaamaaCaaameqabaGaa8NAaaaaaSqaaiaa=ngaaaGcca qGGaGaey4kaSIaaeiiaiabeo7aNjaa=HfadaWgaaWcbaGaa8xAaaqa baaakiaawIcacaGLPaaacaqGGaaaaa@59D1@ P( Y ij =  1|x ) = Φ( α l X j 1  + β X j c  + γ X i )     (2) MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaaeiuamaabm aabaacbiGaa8xwamaaBaaaleaacaWFPbGaa8NAaiaa=bcaaeqaaOGa eyypa0JaaiiOaiaabccacaqGXaGaaiiFaiaa=HhaaiaawIcacaGLPa aacaqGGaGaeyypa0JaaeiiaiabfA6agnaabmaabaGaeqySde2aaSba aSqaaiaabYgaaeqaaOGaa8hwamaaDaaaleaacaWFQbaabaacbaGaa4 xmaaaakiaabccacqGHRaWkcaqGGaGaeqOSdiMaa8hwamaaDaaaleaa caWFQbaabaGaa83yaaaakiaabccacqGHRaWkcaqGGaGaeq4SdCMaa8 hwamaaBaaaleaacaWFPbaabeaaaOGaayjkaiaawMcaaiaabccacaqG GaGaaeiiaiaabccacaqGGaGaaeikaiaabkdacaqGPaaaaa@5DD0@ P( Y ij =  1|x ) = Φ( α l X j 1  +  α g X j g + β X j c + γ X i )    (3) MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaaeiuamaabm aabaacbiGaa8xwamaaBaaaleaacaWFPbGaa8NAaiaa=bcaaeqaaOGa eyypa0JaaiiOaiaabccacaqGXaGaaiiFaiaa=HhaaiaawIcacaGLPa aacaqGGaGaeyypa0JaaeiiaGGaaiab+z6agnaabmaabaGaeqySde2a aSbaaSqaaiaabYgaaeqaaOGaa8hwamaaDaaaleaacaWGQbaabaGaaG ymaaaakiaabccacqGHRaWkcaqGGaGaeqySde2aaSbaaSqaaiaa=Dga aeqaaOGaa8hwamaaDaaaleaacaWGQbaabaGaam4zaaaakiabgUcaRi aabccacqaHYoGycaWFybWaa0baaSqaaiaadQgaaeaacaWGJbaaaOGa ey4kaSIaaeiiaiabeo7aNjaa=HfadaWgaaWcbaGaa8xAaiaa=bcaae qaaaGccaGLOaGaayzkaaGaaeiiaiaabccacaqGGaGaaeiiaiaabIca caqGZaGaaeykaaaa@6469@

    where P( Y ij =1|x ) MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaaeiuamaabm aabaacbiGaa8xwamaaBaaaleaacaWFPbGaa8NAaiaa=bcaaeqaaOGa eyypa0JaaeymaiaacYhacaWF4baacaGLOaGaayzkaaaaaa@3F97@  is the probability of immigrant i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaamyAaaaa@36E5@  from source country j MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaamOAaaaa@36E6@  participating in the labour force given a set of characteristics ( X ) MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaWaaeWaaeaaie GacaWFybaacaGLOaGaayzkaaaaaa@3863@ ; Φ MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaeuOPdyeaaa@3771@  is the standard normal cumulative density function. X j g MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaacbiGaa8hwam aaDaaaleaadaahaaadbeqaaiaa=PgaaaaaleaacaWFNbaaaaaa@3913@  represents source-country gender-role attitudes, X j 1 MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaacbiGaa8hwam aaDaaaleaadaahaaadbeqaaiaa=PgaaaaaleaaieaacaGFXaaaaaaa @38E5@  represents source-country female labour force participation rates, and X j c MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaacbiGaa8hwam aaDaaaleaacaWGQbaabaGaam4yaaaaaaa@38DE@  represents the other three source-country control variables.Note 12 X i MathType@MTEF@5@5@+= feaagKart1ev2aqatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaacbiGaa8hwam aaBaaaleaacaWFPbaabeaaaaa@37F1@  are individual-level socio-demographic variables and survey year fixed effects, as discussed in the previous section. Model 1 examines the effect of source-country gender-role attitudes in the absence of the source-country female labour force participation rate; Model 2 examines the effect of the source-country female labour force participation rate in the absence of source-country gender-role attitudes; Model 3 examines the effects of the two factors in the presence of each other.

    Similar models are estimated for immigrant female paid employees, using log hourly wages as the dependent variable. An additional model is examined for this group, in which industrial and occupational distribution variables are included with source-country female labour force participation rates, gender-role attitudes and other control variables. The purpose of this additional model is to test whether the observed effect of source-country female labour force participation decreases substantially after taking into account differences in industrial and occupational distributions. A large decrease in the source-country female labour force participation coefficient would imply that any earnings advantages among immigrants from source countries with high levels of female labour force participation are at least partly attributable to their favourable allocation in the industrial and occupational structures.

    Since source-country attributes are measured at the group level, regression models with cluster-robust standard errors are estimated to correct for within-cluster correlation and heteroscedasticity (Angrist and Pischke 2009; Wooldridge 2003). This differs from the Hierarchical Linear Model (HLM) approach, which assumes homoscedastic errors both at the individual and cluster levels. Cluster-robust standard errors assume no particular forms of within-cluster correlation and heteroscedasticity. The cluster in this study is based on the combination of source country (83 countries) and year at landing (a total of 22 years). There are 1,459 clusters in the data, as some year–country combinations offer no observations.

    A potential methodological issue in estimating the models for wages is related to the selectivity of being employed among immigrant women. A large proportion of immigrant women do not participate in the labour market (about 28% in the selected sample). If the decision to take part in the labour market is correlated with the determinants of wages, then regression estimates in the wage model would be biased. This is because these estimates would be based on a non-random sample and therefore could not be generalized to the broad population (Greene 2012). To check this possibility, the Heckman selection correction procedure is used. In the selection equation, the demographic variables used in the wages equation are included, along with marital status, the spouse’s employment status if married and the number of children aged 0 to 5.Note 13 As seen in Table A.1, the error terms in the wage model and in the selection model are not significantly correlated, so the regression estimates from the wage model are likely unbiased.Note 14 The coefficients in the wage models with the selection correction are almost identical to those from the ordinary least squares estimates. For simplicity, only the results without the Heckman correction are discussed in the text.

    4 Results

    4.1 Source-country female labour force participation and gender-role attitudes

    Among the 83 source countries with data on gender-role attitudes, there is a moderate correlation between country-level female labour force participation and gender-role attitudes (Pearson r =0.42, p<0.001). Chart 1 plots the association between source-country female labour force participation and gender-role attitudes. This chart illustrates why the two measures are only moderately correlated. On the one hand, most of the countries with low scores in gender-role attitudes also have very low levels of female labour force participation. These countries include, in ascending order of gender-role attitudes score, Iraq, Pakistan, Saudi Arabia, Jordan, Egypt, and Morocco. On the other hand, countries with very high female labour force participation rates only rank in the middle range in the distribution of gender-role attitudes. The five countries with the highest female labour force participation rates are located in sub-Saharan Africa: Ghana, Ethiopia, Uganda, Rwanda, and Tanzania. In these countries, women traditionally provide the majority of the labour in agriculture production and processing; men are primarily involved in labour activities that generate cash income. The high female labour force participation in these countries is not closely related to gender equality in broad socio-economic aspects; rather, the gender division of non-domestic labour is shaped by patriarchal norms (Doss 1999). Furthermore, countries with the highest gender-role attitudes scores (such as Norway, Sweden, France and the Netherlands) only have moderate levels of female labour force participation.Note 15 In short, very low scores of gender-role attitudes and very low levels of female labour force participation go in tandem, but high gender-role equalities in terms of the broad socio-economic characteristics of a society do not necessarily mean that women participate in the labour market to the same extent as men. Indeed, in countries where women participate in the labour force at similar levels as men, it may often be out of necessity rather than choice (Lim 2002).

    Description for chart 1

    In addition to being moderately correlated with source-country gender-role attitudes, source-country female labour force participation rates are also correlated with other characteristics that may be relevant to immigrant women’s labour market outcomes. Table 1 presents descriptive statistics for the main variables used in the subsequent multivariate models by source-country female labour force participation level. For the purpose of this table, source countries are grouped according to whether they have ‘high’ or ‘low’ female labour force participation rates. High female labour force participation rates are defined as 0.50 (or 50%) or more while low rates are less than 0.50. Note that in the multivariate models, the value of this variable remains unique to each source country in a given year of immigrant arrival, with values ranging from 0.11 to 0.87, with a mean of 0.49.

    As shown in Table 1, about 55% of immigrant women in the study sample, or 16,263 out of 29,811, come from source countries with a female labour force participation rate that is less than 50% (with an average of 38.2%). Source countries with low levels of female labour force participation also tend to have lower levels of economic development, as indicated by GDP per capita, are more likely to use English or French as the official language, and are less likely to be a Western country.

    Immigrant women from countries with low female labour force participation tend to have lower labour force participation rates and lower hourly wages in Canada compared with those from countries with high levels of female labour force participation. However, the two groups of immigrant women are quite similar in their demographic characteristics—educational level, age at immigration, years since landing, full-time status, and job tenure.

    4.2 The association between source-country female labour force participation and immigrant women’s labour force participation in Canada

    Chart 2 plots the average labour force participation rates of immigrant women by female labour force participation rates in their respective source countries. The chart shows that there is a large variation in female labour force participation rates across immigrant groups in Canada. While the female labour force participation rates for most immigrant groups are in the 60% to 80% range, some groups have rates lower than 50% (e.g., those from Pakistan, Iraq, Jordan, Saudi Arabia and Taiwan) and some groups have rates higher than 85% (e.g., those from Philippines, Belarus, Romania, Albania and Zimbabwe). The chart also shows that the variation in immigrant women’s labour force participation rates in Canada is positively associated with the source-country female labour force participation rate.

    The bivariate associations presented in Chart 2 do not account for group differences in source-country gender-role attitudes, other source-country attributes, or variations in immigrants’ socio-demographic composition. This is addressed with the probit models presented in Table 2. Table 2 presents the average marginal effects of the explanatory variables from the probit models. For dummy variables, the marginal effect is the difference in the predicted average probability of participating in the labour force between the given category and the reference category. For instance, in Model 1, the marginal effect associated with having less than a high-school education implies that immigrant women’s predicted probability of participating in the labour force if they have less than a high school education is -0.243, or 24.3 percentage points lower than immigrant women with a university degree. For continuous variables, the marginal effect reflects changes in the predicted probability associated with a one unit change in the explanatory variable.

    Description for chart 2

    The results for Model 1 and Model 2 in Table 2 show that, after controlling for other source-country attributes and individual-level characteristics, source-country gender-role attitudes and source-country female labour force participation are each positively associated with immigrant women’s labour force participation in Canada. However, once both are in the same model (Model 3), the source-country gender-role attitudes coefficient ceases to be significant, while the effect of the source-country female labour force participation rate becomes slightly stronger. Thus, the association between source-country gender-role attitudes and immigrant women’s labour force participation is accounted for by the source-country female labour force participation level. Put differently, the effect of source-country gender-role attitudes works through its moderate correlation with the source-country female labour force participation rate. The coefficient associated with the level of female labour force participation in the source country (Model 3, Table 2) indicates that a 0.5-point increase in the source-county female labour force participation rate, which is close to the difference between Pakistan (17%) and Lebanon (20%) at the low end, and China (70%) and Vietnam (71%) at the high-middle end, is associated with a 0.177-point increase (or 17.7 percentage points) in the immigrant women’s labour force participation rate (i.e., one-half of the coefficient).

    Among the other three source-country attributes, immigrant women from a source country that has English or French as the official language have a higher level of labour force participation (Model 3, Table 2). Coming from a Western country is associated with a higher level of labour force participation among immigrant women in Canada; however, the level of source-country economic development is not significantly associated with their labour force participation.

    Among individual-level characteristics, education and years since landing in Canada are strongly and positively associated with immigrant women’s labour force participation. Older age at immigration and number of young children are negatively associated with immigrant women’s labour force participation. Married immigrant women with a spouse who is not working tend to have a lower level of labour force participation than those who are married with a working spouse, likely a result of assortative mating by labour market behaviours (Ultee, Dessens and Jansen 1988).

    4.3 The association between source-country female labour force participation and immigrant women’s wages

    Chart 3 plots the average log hourly wages of immigrant women by female labour force participation rates in their respective source countries. A large variation is clearly apparent in log hourly wages across immigrant groups. While most have hourly wages between $15 (log hourly wage 2.7) and $22 (log hourly wage 3.1), nine groups of immigrant women are above this range and five groups are below.Note 16 The variation in immigrant women’s log hourly wages is positively associated with the source-country female labour force participation rate.

    The regression models for immigrant women’s log hourly wages are presented in Table 3. The results for Model 1 and Model 2 show that, after controlling for other source-country attributes and individual-level characteristics, source-country gender-role attitudes and source-country female labour force participation rates, in the absence of the each other, are strongly associated with immigrant women’s hourly wages. When both source-country attributes are included in the same model (Model 3), their coefficients decrease in size but remain significant. This result suggests that source-country gender-role attitudes and source-country female labour force participation levels are independently associated with immigrant women’s wages, even after their overlapping effects are taken into account. The source-country female labour force participation coefficient indicates that a 0.5-point increase in the source-county female labour force participation rate corresponds to an 8.5% increase in wages (i.e., one-half of the coefficient).

    Among the other source-country attributes, GDP per capita is positively associated with the hourly wages of immigrant women (Model 3, Table 3). Coming from a Western country is also strongly associated with increased wages. This variable likely captures the wage difference between Whites and visible minority groups, as source-country economic development and individual-level demographic variables are controlled for in the model. Unfortunately, this cannot be tested here; the LFS does not collect information on visible minority status.

    Description for chart 3

    When industrial and occupational distributions are included, the coefficient for source-country female labour force participation decreases by three-quarters and ceases to be significant (Model 4, Table 3). This result implies that the higher wages of immigrant women from source countries with higher female labour force participation are primarily attributable to their greater concentration in higher-paying industries and occupations. That is, differential allocation of industries and occupations account for the association between source-country female labour force participation and wages of immigrant women. In comparison, the coefficient of source-country gender-role attitudes decreases by about one-quarter, and remains statistically significant when industry and occupation are controlled for (Model 4, Table 3). The relationship between source-country gender-role attitudes on immigrant women’s wages does not seem to act primarily through industrial/occupational allocation, but rather through career advancement within the same industry/occupation.

    Comparing the top 10 occupations (based on the first two digits of the 2006 NOC-S codes) for immigrant women from countries with low and high levels of female labour force participation yields a more detailed account of differences in their occupational distribution, and of how these differences are related to their wages (Table 4). Variations in the types of occupations held are apparent. Clerical and Elemental sales and service occupations are the top two groups for women from both sets of source countries; however, a greater share of women from countries with low female labour force participation hold these jobs than the share of women from countries with high female labour force participation. Furthermore, several occupational categories differ between the two groups. Most notably, although 7% of women from countries with high female labour force participation rates work in professional occupations in the natural and applied sciences—highly paid occupations—this group is not among the top 10 occupations of women from countries with low female participation rates. Moreover, two low-paying groups appear only in the top occupations of immigrant women from low female labour force participation countries: Childcare and home support workers and Cashiers. Within the same occupation, women from countries with high female labour force participation tend to have higher wages than women from low female labour force participation countries; however, the gap is generally small. For instance, the wage gap in the top two occupations is about 4% to 6%. Overall, these results suggest that the lower average wages of women from countries with low female labour force participation is owing to their relative concentration in low-paying occupations.

    The allocation of industries and occupations is also an important mechanism through which several other variables affect immigrant women’s wages. This is evident from the large decreases in the coefficients of these variables (Model 4, Table 3). In particular, the inclusion of industrial and occupational distributions reduces the size of the coefficients associated with GDP per capita, coming from a Western country, education, age at immigration, and years since landing. These results suggest that immigrant women who emigrated from developed economies and Western countries, have higher educational levels, arrived at a younger age, and have resided in Canada longer are more likely to work in higher-paying industries and occupations.

    5 Conclusion

    This study contributes to the literature in two ways. First, the association between source-country female labour force participation—assumed to represent a country’s gender attitudes—and an explicit measure of source-country gender-role attitudes is examined in detail. Second, the relationships between source-country female labour force participation and the labour market participation and wages of immigrant women in Canada are investigated. In particular, this study examines whether the relationship between source-country female labour force participation rates and immigrant women’s labour market outcomes is limited only to the influence that cultural gender-role attitudes have on women’s participation in the host country’s labour market—or if this characteristic is also associated with their wages.

    This paper finds that high levels of attitudes favouring gender equality do not necessarily translate into high levels of female labour force participation. Only a moderate correlation is found between the source-country female labour force participation rate and source-country gender-role attitudes. And when both of these variables are included in a model of labour force participation in Canada, only the source-country female labour force participation rate is statistically significant. This suggests that the effect of source-country gender-role attitudes works through its relationship with the source-country female labour force participation rate. These findings are consistent with those of previous studies that observe a positive relationship between source-country female labour force participation rates and immigrant women’s labour supply in the host country (e.g., Antecol 2000; Blau, Kahn and Papps 2011; Frank and Hou 2013).

    The analysis of immigrant women’s wages shows that immigrant women from nations with high levels of female labour force participation receive higher wages than women from other nations. This is largely attributable to their concentration in higher-paying industries and occupations. One interpretation is that high female labour force participation in the source country may contribute to the acquisition of specialized skills, labour market knowledge or other unobserved human capital that is useful in obtaining higher paying employment in the host country (e.g., more work experience, greater familiarity with high-paying firms or industries, better understanding of qualifications needed to enter high-paying occupations). These factors may be particularly beneficial for immigrants from countries with a labour market similar to that of the host country.

    Unlike the results for the source-country female labour force participation variable, the statistically significant and positive relationship between source-country gender-role attitudes and immigrant women’s wages persists when occupation and industry are taken into account. This finding indicates that gender-role attitudes affect immigrant women’s wages through a different mechanism and supports previous research that finds a direct relationship between gender-role attitudes and women’s wages. Whether this is due to higher salary expectations, greater assertiveness or greater motivation to pursue promotions and raises among women with more egalitarian gender-role attitudes cannot be determined from the data used for this study.

    6 Appendix

    Notes

    Date modified: