UNIONS, COLLECTIVE BARGAINING AND LABOUR MARKET OUTCOMES FOR CANADIAN WORKING WOMEN: PAST GAINS AND FUTURE CHALLENGES by Andrew Jackson, Senior Economist, Canadian Labour Congress and Grant Schellenberg, Assistant Director, Canadian Council on Social Development INTRODUCTION: UNIONS, LOW PAY, AND EARNINGS INEQUALITY The major purposes of this paper are, first, to examine the impacts of collective bargaining on labour market outcomes for women workers in Canada, specifically with respect to pay, benefits coverage, the incidence of low pay and the extent of earnings inequality, and, second, to suggest ways in which positive impacts could be extended via the expansion of collective bargaining coverage. This part of the paper briefly reviews the literature on the impacts of collective bargaining on earnings, low pay, and earnings inequality, and Part II provides some background description of the labour market position of Canadian working women. Particular attention is paid to the situation of the majority of women who continue to work in lower paid, often insecure and part-time, clerical, sales, and service jobs. The central conclusion of the empirical analysis in Part, mainly based on data from Statistics Canada's 1995 Survey of Working Arrangements, is that collective bargaining coverage, controlling for other factors, has significant positive impacts in terms of raising pay and access to benefits, and in terms of reducing the incidence of low pay among women workers. However, the level of collective bargaining coverage for women is very low in precisely those sectors of the economy where women in low paid and insecure jobs are most concentrated, namely in private services and in smaller enterprises. Promoting better labour market outcomes for women workers accordingly requires a major extension of collective bargaining.

Part IV of the paper briefly considers ways in which this could be achieved through trade union action and through changes to public policy. The 1996 OECD Employment Outlook comprehensively documented profound differences in the degree of earnings inequality and the incidence of low pay in the advanced industrial countries, noting that these two labour market characteristics are closely related in that 'the incidence of low pay tends to be highest in those countries where earnings inequality is the most pronounced. ' While there is significant variation between countries, a generalized pattern is that continental European countries, particularly in Northern Europe, have a strikingly more equal distribution of earnings and a much lower incidence of low pay among both working women and men than do the U.S., the UK, and Canada. To indicate the extremes, the earnings gap between the top and bottom deciles of women earners in Canada is double that in Sweden (i.e. the ratio between the upper limit of D 9 and the bottom limit of D 2 is 3.7 compared to 1.8), and the incidence of low pay among full-time women workers (defined as earning below two-thirds of the economy wide median wage) is 34.3% in Canada compared to just 8.4% in Sweden. OECD countries such as Canada and Sweden are exposed to broadly comparable forces of 'structural' change, such as exposure to international trade and investment flows and to rapid technological change, but differ significantly in terms of labour market institutions.

This suggests that institutions such as collective bargaining can have significant impacts on the quality of jobs as well as on the level of inequality between wage earners. As the OECD notes, a major explanation for large differences between countries is labour market institutions: 'different institutional settings with regard to wage bargaining, legal minimum wages and the generosity of unemployment and other related benefits appear to account for some of the wide variation across countries in the over all incidence of low pay. ' More specifically, it was established in the OECD analysis that there is a high and negative correlation between collective bargaining coverage and the incidence of low pay. Similarly, a major set of Canada-U.S. comparative studies has showed that labour market institutions, broadly defined, have significantly attenuated income inequality in Canada compared to the U.S. (Card and Freeman, 1993).

Among those studies, Lemieux (1993) showed that the higher level of unionization in Canada compared to the U.S. was one important factor explaining the lower level of wage inequality and somewhat lower incidence of low pay among Canadian compared to U.S. workers in the mid-1980's. He calculates that the pattern and extent of unionization in Canada compared to the U.S. explained 40% of the difference in the wage inequality among men in Canada compared to the U.S. The impact of unionization and collective bargaining on the quality of employment in terms of pay, benefits, security, and other outcomes is a complex and controversial subject. However, there is a substantial consensus that collective bargaining raises pay and benefits for unionized workers compared to otherwise comparable non-union workers, particularly lower skilled workers (the wage premium effect) and promotes greater equality of wages and working conditions within the unionized sector by compressing wage and benefit differentials (the compression effect). Chaykowski (1995) provides evidence on the impact of Canadian unions on pay and access to benefits, and on pay differentials within and between firms. Collective bargaining generally reduces the incidence of low pay and promotes equality between workers by simultaneously lifting the wage floor and by compressing wage differentials in the unionized sector, but collective bargaining can also increase inequality in the labour market as a whole by creating a pay gap between union and non-union workers.

The impact of collective bargaining on the incidence of low pay and on inequality in the labour market as a whole is thus crucially determined by three major variables: the wage premium effect (the impact of unionization on wages and differentials of union compared to non workers); the compression effect (the impact on earnings distributions in the unionized sector); and by the extent of collective bargaining coverage among the labour force as a whole. A crucial institutional difference between many continental European countries and North America is that collective bargaining coverage is much higher in the former because of generally higher unionization rates, and because many non-union workers are covered by informal or legal extension of collective agreements. Union wage premiums, by contrast, have generally been found to be higher in the U.S. It can be observed that, because of the impacts on the labour costs of union compared to non-union employers operating in competitive markets, relatively low collective bargaining coverage combined with a high union wage premium is likely to considerably increase employer hostility to unions and decrease the prospects for long-term union survival and growth compared to wide coverage and a lower wage premium (Chaykowski and Slotsve, 1996). Recent empirical work in the U.S. has found that the inequality reducing wage compression effects of collective bargaining within the unionized sector outweigh the inequality increasing wage premium effect among male workers. Declining unionization has thus implied increasing wage inequality, and it has been estimated by Card (1992) and Freeman (1993) that about one-fifth of the increase in earnings inequality among U.S. men in the 1980's can be attributed to the sharp fall in unionization over the decade. The impacts of collective bargaining on wage inequality among women have been found to differ because of the historically lower rate of unionization among women (though this is no longer true of Canada), and because unionization among women, unlike men, tends to be more heavily concentrated among highly educated workers.

This is, in turn, a function of the relatively high unionization rates in public or quasi public services such as health and education which employ a large proportion of women with high levels of education, and very low unionization rates in private services which employ a high proportion of women, such as retail trade, personal services and accommodation and food services. Fortin and Lemieux (1996) have found that changes in labour market institutions, broadly defined, account for about one third of the increase in wage inequality among U.S. men and women in the 1980's, but that the fall in collective bargaining coverage specifically had little impact on inequality among women. Lemieux (1993) found, on the basis of Canadian data for 1986, that unionization significantly raised the wages of unionized compared to non-unionized women, that the unionization rate of women increased with skill level, and that unionization compressed wages among unionized women. The 'wage raising' effect of unionization among women was not entirely offset by the wage compression effect in the unionized sector, so unionization very slightly increased wage inequality among women. Unionization did, however, very modestly improve the position of women relative to men in the overall wage distribution.

In both the U.S. and Canada, research has shown that unionization has little impact on the overall gap between the wages of women and men, in significant part because the unionization rate has been much higher for men, thus concentrating the income gains of collective bargaining among male workers (particularly blue collar workers). However, the impact of unionization on inequality of earnings between men and women was found to be small in Canada in 1984 (30 cents per hour), and was concentrated in the private sector where the unionization rate of women is low. Unionization has considerably reduced earnings differences between men and women in public services where the unionization rate of women is high. (Maki and Ng, 1990).

The impact of collective bargaining on earnings, inequality between women and men will increase if the unionization rate of women in private services increases, and / or if more women work in relatively highly unionized blue collar occupations. There is, then, evidence that the pattern and extent of unionization are major influences on the distribution of earnings and the incidence of low pay, and that these patterns are different for women and men. Higher rates of unionization and collective bargaining coverage among women could potentially have a significant impact on the very high incidence of low pay among Canadian women workers, on high levels of inequality of earnings among Canadian women, and on the still significant earnings gap between women and men. These issues are clearly of public concern, but the impact of unionization on women workers in the 1990's has not been extensively studied. A central purpose of this paper is to document the independent impact of collective bargaining coverage on the level of wages, the incidence of low pay, and the distribution of wages, with a specific focus on the impacts upon women. The data are presented so as to facilitate comparisons with the impacts on men.

It would be argued by some that the gains of unionization come at the expense of employment, particularly for the lower skilled. However, the OECD analysis of low paid work and earnings inequality found that 'there is little evidence to suggest that countries where low-paid work is less prevalent have achieved this at the cost of higher unemployment rates and lower employment rates for the more vulnerable groups in the labour market such as youth and women. ' Freeman has similarly concluded on the basis of numerous recent studies that union wage effects (higher wages, particularly for the low paid, and greater compression in the distribution of wages) do not come at the expense of productivity and are not responsible for the aggregate level of unemployment, largely because collectively bargained wages flexibly respond to shifting market realities (Freeman, 1992, 1994). According to the 1995 World Employment Report of the International Labour Organization (ILO), it is wrong to view labour market regulation as the fundamental cause of unemployment, and it is important to recognize that such regulation has important positive impacts for society in terms of greater equality and less poverty. The ILO report takes particularly strong issue with the widespread view that labour market 'rigidities' have been an important source of the unemployment problem in Europe, and emphasizes the many negative features of 'flexible' U.S. style labour markets from the point of view of workers. Collective bargaining procedures which secure steady employment at decent wages in a relatively non-polarized labour market are seen as an important source of social well-being, rather than as a barrier to job creation.

In recent years, the ILO has also drawn attention to the positive role of labour market regulation and collective bargaining in addressing the specific labour market problems faced by women. In Canada, the labour movement has made extension of collective bargaining among women, particularly lower paid women, a major priority. The widely distributed 1997 Canadian Labour Congress (CLC) document Women's Work was based on extended discussions with women workers and union leaders and activists across the country, and was intended to closely examine the impact of recent economic restructuring on women's work and women's lives as a basis for critical reflection on current trade union practices and on public policies. The report documented the increased precariousness of women's jobs that has come with the shift to temporary, contract and involuntary part-time work, particularly among women clerical, sales and service workers in the private sector, and the loss of reasonably secure and well paid jobs for women, particularly in public and social services and in manufacturing, that has been another result of economic restructuring. While noting the continued progress of women in the labour market in some key respects, such as declining earnings differentials between full-time, full year working men and women and increased entry of women into managerial and professional occupations, central themes of the report were the increase in insecurity and the continued high prevalence of both insecurity and low pay for many women, and the continued large inequalities which exist between working women and men. The report noted with some reference to broad aggregate data that unionization had raised wages for women workers, particularly women working in part-time and other 'non-standard' jobs, and also documented important equality gains for women which have been made through the collective bargaining process.

However, as the report also highlighted, rates of unionization are very low for women (and men) in the private service sector of the economy, and collective bargaining coverage is particularly low for workers in small firms and workers in 'precarious' jobs, such as contract and casual workers. These are precisely the areas of employment for women which are expanding most rapidly. In short, the labour movement has tempered its recognition of some success in terms of bargaining positive outcomes for working women with recognition of relative failure in terms of extending collective bargaining to the most insecure and vulnerable women workers, including most visible minority and Aboriginal women. The fact that unions are most absent where they are most needed has led the CLC to argue that organizing women, particularly women in private services, must be made a central priority and the report made a number of recommendations for trade union and government action which are noted in Part IV of the paper. Similar recommendations were also made by Alexandra D agg in her capacity as the labour representative on the Advisory Group on the Changing Workplace appointed by the federal Minister of Labour in 1996 (Advisory Committee on the Changing Workplace, 1997). II WOMEN IN THE CANADIAN LABOUR MARKET: AN OVERVIEW The labour market experience of Canadian women is very different from that of men, and profound (though declining) inequalities between men and women continue to exist with respect to pay and benefits, and with respect to hours worked on both a weekly and annual basis.

Inequality of earnings between men and women is driven in large part by differences in hourly pay. Average hourly earnings of women are, on average, 82% those of men ($13.63 vs. $16.58) and the incidence of low wages is much higher among women than men. If low wage is defined as earning less than two-thirds the economy wide median hourly wage, the incidence of low pay among all women in 1995 was 33.6%, or more than half again as high as the 21.5% rate among men. This wage threshold is also approximately the same as that required to place a full-time earner above the Statistics Canada Low Income Cut Off for a single person. The effect of differences in hourly pay on wage and salary income are compounded by the effects of hours worked.

Just 52% of women workers work on a full-time, full year basis, compared to 66% of men. As a result, the difference between the annual earnings of men and women is greater than the difference of hourly earnings. Women's median annual earnings in 1996 were 61% those of men - $17,408 vs. $28,510, while women working full-time, full year earned 73.4% as much as men ($30,717 vs. $41,848). The major difference between annual hours worked by women and men flows from the far higher incidence of part-time work among women: one in four (24%) adult women work part-time, compared to just one in twenty (6%) of adult men. While the higher incidence of part-time work is partly driven by choice and by the gendered distribution of responsibilities for child care and domestic labour, it also represents a lack of access to desired hours of work. In recent years, at least one in three women part-time workers have been underemployed in the sense that they wanted, but could not find, full-time jobs.

Hours of work in part-time jobs also tend to be more variable than those of full-time workers - one in three part-timers have no regular work schedule but work the weekly hours requested by the employer. It can also be noted that the incidence of paid overtime on a weekly basis is significantly higher among men than women (18.4% vs. 9.4%) and that weekly overtime hours average significantly higher for men (5.5% vs. 3.2%). However, the jobs of women tend to be more stable than those of men in the sense that incidence of layoffs is much lower. Many of these differences in pay and hours worked result from the very different occupational distribution of women and men, as well as differences in the jobs held by women and men within occupational categories. As indicated in Table 1, a majority of working women (53%) work in generally low paid, often part-time, clerical, sales and service occupations, double the proportion of men (26%). While clerical work has been in decline, it is notable that more than one quarter of women (28%) still work in clerical jobs, which are overwhelmingly held by women.

Sales and services jobs are much more evenly divided between women and men, and their share of total employment has been rising. Fully one half (48.6%) of women in clerical, sales, and services occupations were low paid workers in 1995 (again defined as earning less than two-thirds the economy wide median hourly wage) and annual earnings for women clerical, sales, and service workers averaged just $17,184.31. 6% of these women were under-employed in the sense that they reported a desire to work more hours than currently worked, and 20.4% worked an irregular or on-call schedule. 38% of women work in generally better paid managerial / administrative and professional occupations (13% and 25% respectively), and this proportion has been steadily rising, driving much of the decreased earnings inequality between women and men. The increased entry of women into these occupations was historically closely associated with the growth of public and social services, notably health care and education. About two in three women in these higher paid occupations work in non market services, very broadly defined to include health, education and social services, in part because the incidence of professional occupations non market services is much higher than in private services, and in part because two out of three workers in non market services are women.

Clearly, the lack of job growth in public and social services in recent years - indeed actual reductions in employment in public administration and in institutional health care - has negatively impacted upon a major source of good jobs for working women. By the same token, future prospects for women in the labour market will very much depend upon the quality of jobs in growing private services. The pay gap between full-time, full year women and men has been declining in the 1980's and 1990's, in part due to the falling real earnings of men. There remain, however, very significant differences between the distribution of annual earnings among men and women.

In 1996, just 6.3% of women earned more than $50,000 compared to 20.4% of men, and 45.3% of women earned less than $15,000 per year compared to 28.9% of men. Table 2 shows that the proportion of women falls at each step in the decile distribution of annual earnings - with women making up 62% of the lowest income decile, and just 17% of the top decile. However, as indicated in the Table, the proportion of women in the upper part of the earnings distribution has increased significantly since 1984. Table 1. Composition of Female and Male Employment and Incidence of Unionization, by Selected Characteristics, Canada 1995 Women Men Composition of Employment % Unionized Composition of Employment % UnionizedAllAged 15 to 24 Aged 25 to 44 Aged 45 to 69 Less than high school High school / Some post-sec. Certificate / DiplomaUniversity Degree Full-timePart-time Management / AdministrationProfessionalClericalSalesServicesBlue Collar Firm size less than 20 Firm size 20 to 99 Firm size 100 to 500 Firm size more than 500 Public Sector Private Sector Agriculture / PrimaryManufacturingConstructionDistributive services Retail Trade FIRE / Business services Community services Personal and other services Public Administration 100%2055251834311770301325281015924152041198121117141429166 3613384426284049392724623211243462750507924 na 29 na 37131264976 100%21532624312916871313146713462118184316845248161199117 40164251403644404318244845153747102847598032334737411916691276 Distributive services includes transportation, communications (including postal services), utilities and wholesale trade.

FIRE includes financial services, insurance and real estate. Community services includes health, education and religious organizations. Personal services and other services includes recreation; food, and accommodation; private households and miscellaneous services. na - Sample size too small to provide reliable estimate. Table 2.460.656.355.753.740. 61.659. 458.359. 255.752.

347.934. 028.817. 246.6 1.13. 0-2.32.

90.0-1.47. 94.09. 65.62. 2 Source: Scott, K. and Loc head, C. 'Are Women Catching Up in the Earnings Race?', Canadian Council on Social Development, 1996. Picot (1996) has recently documented trends in the earnings distribution of men and women in the 1980's and 1990's. He finds that, unlike for men, there has been little or no increase in earnings inequality among women since 1981, and that there has been a modest increase in the real median annual earnings of women.

The real annual earnings of lower paid women have also increased, but it is emphasized that this has been largely driven by increases in the number of weeks worked over a year rather than by increases in hourly pay. There has also been some shifting of women into higher hourly paid jobs. In summary, there has been a decline in the earnings inequality of women compared to men, but women remain concentrated in relatively low hourly wage occupations, and in jobs which offer fewer hours of work than those occupied by men. These differences are also reflected in access to non wage benefits. 47.8% of women belong to employer sponsored pension plans, compared to 54.3% of men; women average 11.6 days of paid vacation per year compared to 13.4 for men; 49.8% of women belong to a dental plan compared to 59.4% of men; 53.5% of women have a health plan compared to 63.9% of men; and 55.8% of women have access to paid sick leave compared to 58.1% of men. It should be noted that visible minority, disabled and Aboriginal women - who comprise respectively 8%, 6%, and 3% of all women workers - have still lower earnings than other women.

On average, visible minority women workers earned 9% less per hour in 1993 than did all women workers, workers with disabilities earned 8% less, and Aboriginal workers earned 12% less. The incidence of unemployment is also significantly higher for these equity seeking groups. Unfortunately, little detailed analysis has been undertaken of the specific barriers encountered by minority women in the labour market, and the Survey of Work Arrangements does not allow for such analysis. The collective bargaining coverage rate for women workers is now quite comparable to that of men - 36% compared to 40% (see Table 1). (Collective bargaining coverage is somewhat higher than the unionization rate since it includes workers who are not union members, but who are covered by the terms of a collective agreement.) However, this is in very large part a reflection of the fact that there is a very high unionization rate among women working in public administration (76%) and in community services (64%).

More than one in three women work in these predominantly non market sectors, double the proportion of men. The incidence of collective bargaining coverage for women in professional occupations is both high and is higher than for men - 62% vs. 48% - reflecting the disproportionate concentration of women professionals in the non market sector. By contrast, the collective bargaining coverage of women in sales and service occupations is very low at 11% and 24% respectively. Coverage is both low and somewhat lower than for men in retail trade (13% vs. 19%), financial and business services (12% vs. 16%), and personal services (9% vs. 12%). Coverage is 34% in blue-collar jobs, but just 9% of women work in such jobs. As will be documented in Part, collective bargaining has particularly significant positive impacts upon labour market outcomes for women, holding other factors constant.

However, these gains are relatively narrowly concentrated on a minority of women because of low bargaining coverage. In the private services sector, the gains of collective bargaining apply to only a small minority of workers, and to an even smaller minority of women. IMPACTS OF COLLECTIVE BARGAINING COVERAGE ON CANADIAN WOMEN WORKERS (A) Union Wage Premium There is a consensus within the literature that unionization is associated with a substantial and consistent wage premium in Canada. As Chaykowski and Slotsve (1996) show, such a premium has been consistently reported in studies over the last two decades. Evidence from the 1995 Survey of Work Arrangements (SWA) confirms the presence of the union wage premium. The average wage of women in unionized jobs was more than $5.00 (or 31 per cent) higher than the average wage of women in non-unionized jobs, while for men, the wage difference was about $4.50 - or 24 per cent.

The wage premium associated with unionization is shown for selected subgroups of women and men in Table 3. It is notable that the apparent union wage premium tends to be higher for less educated workers, though this is more clearly the case for men than for women. This is consistent with the fact that managerial and professional occupations in the private sector have very low rates of unionization. Table 3.

Average Hourly Wages of Women and Men, by Unionization and Selected Characteristics, Canada 1995 WOMEN MEN Union Non-Union 'Union Premium' Union Non-Union 'Union Premium " Allege 15 to 24 Age 25 to 44 Age 45 to 69 Less than high school High school grad. Certificate / DiplomaUniversity degree Full-timePart-time Managerial / Admin. ProfessionalClericalSalesServicesBlue Collar Firm size less than 20 Firm size 20 to 99 Firm size 100 to 500 Firm size + 500 16.6811. 2316.9217.

3712.1614. 6016.5621. 3816.9015. 9518.5919. 4914.47.