Page content follows

International differences in low-paid work

View the most recent version.

By Sébastien LaRochelle-Côté and Claude Dionne

Full article in PDF format

• International differences in low-paid work
• Low-paid work in Canada and Australia
• Conclusion
• Data sources and definitions

The Canadian economy includes numerous low-paid jobs, and not just for part-timers. According to the Survey of Labour and Income Dynamics (SLID), one in seven full-time employees (1.4 million workers) were paid less than $10 per hour in 2004. Other studies, using varying definitions of low-paid work, also found a large number of low-paid jobs (Morissette and Picot 2005, Morissette and Johnson 2005, and Chung 2004).

However, Canada's proportionately larger number of workers with low pay in comparison with other Organisation for Economic Co-operation and Development (OECD) countries is perhaps less well known. In fact, Canada has one of the highest proportions of low-paid workers among similarly industrialized countries (OECD 1996 and 1998).¹ By and large, Canada's rate of low-paid work is higher than in European countries and similar to the American rate. In contrast, Scandinavian countries typically have the lowest shares of low-paid workers (Nolan and Marx 1999) (see Data sources and definitions).

International differences in low-paid work are commonly attributed to institutional and regulatory factors clustered among groups of countries. For instance, countries with higher rates of low-paid work are assumed to have a lower degree of labour market intervention with a laissez-faire approach to the labour market (referred to as Anglo-American). In contrast, countries with lower rates of low pay are characterized as more interventionist, with a European approach to the labour market (Cantillon, Marx and Van den Bosch 2002). The contrast between these two typologies has helped fuel debate over the advantages and disadvantages of low-paid work. While some argue that a higher rate of low-paid work provides much-needed flexibility for workers (Siebert 1997), others are concerned by potential problems for individual and family well-being (Maxwell 2002). However, such generalizations must be interpreted with caution as they have been supported by little empirical evidence (Freeman 2005).

In addition, international differences in low-paid work can also result from fundamental differences in demography, industrial structure, living standards, tax incentives, labour supply and institutions. Clearly, the complexity of issues relating to international differences in low-paid work makes it difficult to draw clear inferences to inform labour market policy debates. However, if the low-wage share in Canada differs from countries with similar characteristics and a similar approach to the labour market, like Australia, then the study of differences may be more informative.

This paper provides an update on international differences in low-paid work and then explores potential explanations for the large difference between Canada and Australia, two countries that share many similarities in demography, industrial structure, taxation and living standards.

International differences in low-paid work

International comparisons of low-paid work are not straightforward. One approach is an absolute level of low pay—for instance, the proportion of workers earning less than $10 per hour. But establishing something like a ‘living wage' would pose problems for international comparisons: an amount deemed appropriate to measure deprivation in Canada may not be so in other countries, simply because of differences in perceptions and in cultural norms. Even with agreement on a basket of goods and services corresponding to a minimum standard of living, converting the basket into various currencies would be difficult.

Measures of relative deprivation—the extent to which a worker's earnings fall below their country's median—have been developed to avoid these problems (see Low-pay threshold). A measure of relative deprivation can be interpreted as the number of workers who fall significantly below the financial well-being of the median worker. For example, the OECD defines low-paid work as the proportion of full-year, full-time workers who fall below two-thirds of the country's median earnings (OECD 1996 and 1998). This approach is widely used in comparative studies (Nolan and Marx 1999).

Canada and the United States had the highest proportions of low-paid workers among the 12 countries for which data are available, with nearly 1 in 4 workers earning less than two-thirds of median annual earnings in 2000 and in 2004 (Chart A). The United Kingdom (21.3%) and Ireland (18.9% in 2000) also had relatively large contingents of low-paid workers compared with other countries in continental Europe and Australia. Four countries (Germany, Spain, Austria and Belgium) had similar shares of low-paid workers, varying from 13.0% to 15.7% (for 2000, as 2004 figures were unavailable for these countries). Finally, for both 2000 and 2004, the Scandinavian countries in the sample (Sweden, Denmark and Finland), as well as Australia,³ had relatively small shares of low-paid workers. For the year 2004, the share of low-paid workers varied from 7.1% (in Finland and Denmark) to 11.4% (in Australia).

These results differ little from previous figures released by the OECD (1996 and 1998), which were based on figures provided by the national statistical agencies. Taking the 1996 study as an example, the United States and Canada had the highest share of low-wage workers, with 25.0% and 23.7% respectively in 1994. By and large, this suggests that international differences in low-paid work seen in the 1990s remained largely unchanged in the mid-2000s (see Evolution of low-paid work in Canada).

These countries differ from each other in many aspects. As mentioned, international differences in low-paid work may relate to varying policy approaches to the labour market. Furthermore, differences in low-paid work also reflect other basic differences in such characteristics as demography, economic structure, labour supply, tax incentives, living standards, and country-specific institutions.

However, Australia has a low-pay incidence more in the European mould, even though it is usually included in the ‘Anglo-American, non-interventionist' group of countries (Esping-Andersen 1990). In view of this, a deeper examination of the difference in low-paid work between Canada and Australia follows.

Top of Page

Low-paid work in Canada and Australia

Australia and Canada share many economic, social and political characteristics, often making them the subject of comparative studies. They share a British parliamentary tradition and a federal system of government. Both have small open economies with a relatively modest population (22 million in Australia and 33 million in Canada) and similar immigration rates (Richardson and Lester 2004). Their industrial structures are characterized by abundant natural resources, large exports of raw materials, and large imports of machinery, equipment and production technology (Harchaoui, Jean and Tarkhani 2003). Each has a liberal economy with a social security system characterized by means-testing and private insurance schemes (Esping-Andersen 1990), and a progressive income tax system with similar tax and social security contribution rates (OECD 2009b). Their standards of living are relatively close, with a gross national income per capita of $35,760 for Australia versus $39,650 for Canada, in 2007 US dollars (The World Bank 2009). Economic and productivity growth over the past two decades were similar, as Canada's prosperity grew at an average rate of 1.9% from 1983 to 2000, compared with 2.4% for Australia (Harchaoui, Jean and Tarkhani 2003). Employment rates are close and have increased in tandem (69.3% for Australia and 70.9% for Canada for persons age 15 to 64 in 2000, 70.3% and 72.5% in 2004, and 72.9 and 73.6% in 2007, according to the Online OECD Employment Database). Furthermore, their low-pay thresholds are similar when expressed in common currency figures (see Low-pay threshold).

Despite these similarities, some observable differences may account for the large gap between the two in low-paid work. These factors include personal characteristics of full-year, full-time workers in the two countries (i.e. specific differences in age-sex distribution and education level) as some demographic groups are more likely than others to be low paid.

Younger workers, women and workers with a lower education level were more likely to have low earnings in both countries (Table 2). The situation of younger workers appears especially striking as 29% of young workers in Australia and as much as 65% in Canada were low paid, compared with national rates of 11% and 24% respectively. Also, the differential between men and women was much smaller in Australia, which is consistent with other research finding that Australia has a smaller male-female earnings gap than Canada (Kidd and Shannon 1996).

Furthermore, even if the two countries are characterized by a strong primary sector, other differences in industrial structure and occupational characteristics could also play a role in explaining differences. Low-paid work is proportionately more prevalent in the wholesale and retail sector, and in personal services (Table 3). Conversely, workers in public administration were least likely to be low paid in both countries. Managers and legislators also tended to exhibit lower rates of low-paid work than others.⁵

Hence, if Canada has proportionately more full-year, full-time workers in lower-paid demographic, industry and occupation groups than Australia, then at least part of the differential in low-paid work could be explained by these. One way to test this hypothesis is to use the Oaxaca decomposition method.⁶ This method works on simple counterfactuals: for example, “What would be the proportion of low-wage workers in Canada if it had the same distribution of workers as Australia across various demographic or industry groups?”

However, results indicate that the difference in low-paid work would persist if Canadian workers had the same demographic, industry, and management characteristics as Australian workers.⁷ This is not entirely unexpected, since inter-country differences in rates of low-paid work were also quite large across nearly all of the above characteristics, suggesting the need to look elsewhere to explain the difference between Canada and Australia.

Canada may also differ from Australia in terms of country-specific labour market institutions. The effect of labour-market institutions on pay rates, inequality, employment and low pay has generated much discussion, but is very difficult to assess empirically (Freeman 2005). However, the literature is clear on one thing: labour-market institutions (pay-setting mechanisms, unionization, and the proportion of workers covered by collective agreements) do affect the dispersion of wages, and, by extension, relative rates of low pay (Wallerstein 1999, and Rueda and Pontusson 2000).

How does Canada differ from Australia in this regard? Union coverage is one place to start, since unionized jobs tend to be better paid and have a lower dispersion. However, Canada actually has a higher rate of unionization (29.4% in 2007) than Australia (18.5%) (OECD 2009a), so unionization itself cannot account for the lower incidence of low-paid jobs in Australia.

Differences in the pay-determination process are more fundamental. In Canada, the union sector is characterized by a highly decentralized system of collective bargaining, which means that bargaining between unions and employers occurs mostly at the plant level. For those that are not part of a union officially recognized as a legal bargaining unit, and therefore not covered by collective agreements, the basic employment conditions (including minimum wage) are generally defined by provincial labour codes.⁸ Canada therefore has what could be termed a two-tier, more flexible approach with respect to labour regulations, which has been a defining feature of the labour market for some time (Fudge and Vosko 2001).

By contrast, the Australian labour market is characterized by a system of ‘awards' (compulsory arbitration) dating back to 1907. In this system, government institutions prescribe employment conditions and determine minimum wages for a very large proportion of employees (Kidd and Shannon 1996). Furthermore, the awards system typically covers a large number of employers within a given industry or occupation, including non-unionized workers. The end result is a centralized process of wage determination that provides relatively high minimum standards of pay, the equivalent of which does not exist in Canada.⁹

Because more centralization of the wage-determination process leads to greater wage compression (Wallerstein 1999, and Kidd and Shannon 1996), the Australian awards system—by providing higher minimum-wage standards—probably explains a good deal of the difference in low-paid work between Canada and Australia. It would also help explain the smaller gap in low-paid work between men and women in Australia, as the system also includes provisions to promote greater equity in the workplace (Garton and McCallum 1996, and Kidd and Shannon 1996). However, the awards system has become increasingly criticized in recent decades as it provides very little flexibility for unions and employers to determine wages at the plant level (Norris 1993). Furthermore, many believe that the system is an obstacle to job creation and prevents the economy from reaching its full potential (Lewis 2006). Others also argue that it restricts the competitiveness of Australian businesses (Wailes and Lansbury 2000).¹⁰

To address some of these concerns, successive Australian governments have introduced several reforms since the mid-1980s. This has led to progressive decentralization of the pay-determination process, from the government and industry to the enterprise level, in order to allow more flexibility in bargaining between employers and employees. Furthermore, a number of changes were designed to make pay rates better reflect the performance of industries and individual firms. Nevertheless, the Australian government (through centralized labour market institutions like the Australian Fair Pay Commission) continues to play an important role in establishing minimum-wage conditions and ensuring that equity and fairness conditions are retained in pay-determination procedures (Wailes and Lansbury 2000, and Fenwick 2006), which remains very different from Canadian practice.

Studying earnings distribution is one empirical strategy used to see if Australia's system of awards is associated with lower rates of low pay. Since the Australian awards system provides minimum employment standards to individuals at the bottom of the distribution, differences would likely be lower between individuals at the bottom and those in the middle. Furthermore, in the absence of other major differences in labour-market intervention, the difference between individuals at the top of the earnings distribution and those in the middle should be similar in the two countries.

This can be verified by computing a number of earnings dispersion measures (Chart C). In addition to the widely used P90/P10, which compares the earnings at the 90^th percentile with those at the 10^th percentile, the P50/P10 can be used to compare earnings of the median worker with those at the bottom of the earnings distribution, and the P90/P50 to compare earnings of the median worker with those at the top.

Individuals at the 90^th percentile earned 4.8 times as much as individuals at the 10^th percentile in Canada. In comparison, the figure was 2.8 in Australia, suggesting that overall dispersion was much larger in Canada than Australia. The ratio of the median and at the 10^th percentile was also much larger in Canada, as median workers earned 2.4 times more than individuals at the 10^th percentile, compared with 1.5 times in Australia. By contrast, the dispersion at the top of the earnings distribution was similar. This suggests that most of the difference in the overall dispersion between Canada and Australia is due to differences at the bottom of the distribution.¹¹ This also supports the view that the awards system might explain a great deal of the differences in low-paid work between Canada and Australia.¹²

While differences in the pay-determination process explain some of the difference in low-pay rates between Canada and Australia, they likely do not explain all of it. Other, more subtle, differences could play a role as well. For example, even though Canada and Australia have similar rates of immigration and both select immigrants through a points system, the composition of immigrants is different simply because the two countries are not drawing from the same pool (Richardson and Lester 2004). The implication is that centralized policies aiming to increase minimum-pay standards may not have the same impact on the distribution, or the extent of low pay, in both countries. Furthermore, the impact of such policies on other aspects of the economy (competitiveness, trade and productivity) could also be very different.

Top of Page

Conclusion

This study used the Luxembourg Income Study (LIS) to examine differences in a number of OECD countries in low-paid work, defined as the proportion of full-year, full-time workers earning less than two-thirds of a country's median. The study of low-paid work is motivated by competing views of efficiency and equity in the economy. On the one hand, low-paid work can be advantageous by providing needed work experience for youth and ensuring that the economy has maximum flexibility. On the other hand, a large contingent of low-paid workers presents equity challenges if, for example, many are the sole earners in a family.

Given the debate, international differences in low-paid work are sometimes used to provide information on the relative position of Canada vis-à-vis the rest of the world. Such comparisons yield several groupings of countries with similar economic and social systems: Canada, the United States, the United Kingdom and Ireland have higher rates of low-paid workers than other OECD countries; Western European countries occupy the middle rung; and Scandinavian countries tend to have the lowest proportions of low-paid workers.

Australia is often grouped with Canada, the U.S. and the U.K.—Anglo-American economies that are presumed to have less interventionist policies than European governments. Yet it has a rate of low-paid workers that puts it near the low end of the Western European countries. The detailed examination of low-paid work in Australia and Canada shows that differences in low-paid work are not due to a higher concentration of groups more likely to be low-paid, such as young men, workers without a university degree, or workers in personal services and retail trade. Rather, differences in pay-setting processes likely explain much of the discrepancy between Canada and Australia in terms of low-paid work. Minimum-wage conditions are regulated for the vast majority of Australian workers through an awards system that forms the basis of the minimum compensation policy in the country. The system has more than a 100-year history in Australia, which implies it may not be a readily transferable model.

Top of Page

Data sources and definitions

International comparisons are based on the most recent data from the Luxembourg Income Study (LIS). The LIS is a singular source of comparable labour and income microdata for a wide variety of OECD countries. The analysis is supplemented by the Survey of Consumer Finances (SCF) and the Survey of Labour and Income Dynamics (SLID) to generate historical trends of low-paid work in Canada. The SCF was a cross-sectional survey that used a sub-sample of the Labour Force Survey and was conducted every year from 1976 to 1997. The Survey of Labour and Income Dynamics is an annual longitudinal survey that has been conducted every year since 1993. For the overlapping years, a combined sample of the two surveys was used, as their trends were very similar.

Low-paid workers are defined as employees earning less than two-thirds of the median in each country. As a result, the absolute value used to define low pay varies by country and over time. The choice of the cut-off is a compromise between a lower value of, say, 50% (which would be too close to the minimum wage in some countries) and 75% (which would include too many workers in other countries). This method is not a direct measure of deprivation, but is more related to the ideas of inequality and social exclusion. Furthermore, it has been used in many previous studies. Following the OECD approach, the focus is on annual earnings (before taxes) of paid employees who worked full year, full time (in order to avoid cross-country differences in part-time work).²

Evolution of low-paid work in Canada

Canada's high level of low-paid work relative to other countries since the mid-1990s raises the question: Did Canada always have a high share of low-wage workers? Data from the Survey of Consumer Finances and the Survey of Labour and Income Dynamics indicate that the share of full-year, full-time workers in low-paid jobs rose from approximately 21% in 1976 to 25% in the mid-1980s, and has remained relatively steady since then, suggesting that the numbers for the years 2000 and 2004 reflect an enduring feature of the Canadian economy.⁴

However, the face of low-paid workers changed over the 30-year period, especially between 1976 and the mid-1990s. The share of women with low-paid jobs decreased significantly, from approximately 45% in 1976 to less than 35% at the beginning of the 1990s. This is consistent with the large gains in educational attainment by women during the 1980s and 1990s and suggests that better education credentials led to better jobs for many of them. In contrast, men became increasingly more likely to work in low-paid jobs. From 1976 to 1993, the proportion of men earning less than two-thirds of the median rose from 11% to 18% (Chart B).

However, not all men were equally affected by the increase in low-paid work. In fact, young men (age 15 to 24) were particularly affected as their incidence of low-paid work increased from approximately 30% in 1976 to more than 60% in the mid-1990s. The share of low-paid work among men age 25 to 34 also rose significantly, from 8% in 1976 to approximately 20% in more recent years. Older men were less affected, but middle-aged men also saw their share of low-paid work increase over the period. Conversely, low-paid work declined among middle-aged and older women over the period, remained the same among women age 25 to 34, but rose among women age 15 to 24.

Low-pay threshold

With a relative measure of low pay (two-thirds of a country's median), the threshold is not the same across countries (Table 1). For comparison purposes, values are expressed in 2002 and 2004 Canadian dollars—based on purchasing power parities (PPP), which allow earnings to be expressed in common currency units. PPPs also take differences in price levels between countries into account. The closer a country is to the Canadian threshold, the more similar its definition of low pay in terms of living standards.

While thresholds differed across countries, some were close to the Canadian one. In 2004, for instance, the low-pay threshold was $25,700 for Australia (in 2004 Canadian dollars), compared with $26,700 for Canada (a difference of less than 4%).

Notes

1. The OECD also provides statistical information about rates of low pay across countries in its online employment database (OECD 2009a).
2. The definition of a full-time worker may vary across countries (from 27 to 35 hours per week). Furthermore, information on full-time workers could be retrieved only for the survey reference week in some countries.
3. The number of weeks worked was unavailable in the LIS for Australia in 2004, and for only a fraction of the sample in 2000. Results for Australia are therefore based on full-time workers who earned at least the federal minimum wage over 52 weeks. Results obtained are similar to those provided by the Australian government (Australian Government 2008) and are reasonably close to estimates from the smaller 2000 sample with information on weeks worked. Furthermore, taking only paid employees who worked full time during the survey reference week would yield a rate of 17.3% in 2004, still significantly lower than the Canadian rate for full-year, full-time employees.
4. Median earnings remained relatively constant over the same period, varying between $40,000 and $44,000 (in 2006 dollars) over the last three decades.
5. Similar results were obtained with the SLID master file.
6. The Oaxaca decomposition was obtained as follows. First, two regressions were run, one for Canada and one for Australia, modeling the probability of earning less than two-thirds of the country's median. Variables included age, sex, a dummy for university education, industry, a dummy for managerial occupations, women–age interactions, and women–university education interactions. An alternative rate of low-paid work for Canada was then estimated by multiplying average Australian values for variables included in the regressions by the coefficients obtained in the Canadian regression.
7. In 2004, the real difference between Canada and Australia in low-paid work was 13.0 percentage points and would have been 12.5 points if Canadian workers had been distributed as in Australia across demographic, industry and occupation groups, for which information is available in the LIS.
8. In the case of federally regulated industries, which include banking, telecommunications and interprovincial transportation, employment conditions are prescribed by the Canada Labour Code.
9. The federal minimum wage in Australia is AU$14.31 (approximately CAN$13.00) as of October 2008 (Australian Government 2008) and is much higher than the Canadian average, which currently varies between CAN$7.75 and $10.00 across Canadian provinces.
10. Originally, Australia introduced the awards system to provide basic standards of living for workers in combination with high tariff barriers to protect Australian businesses from foreign competition. That arrangement was increasingly called into question as terms of trade for primary products declined and trade liberalization increased.
11. These results were tested by developing another measure of income dispersion, largely inspired by the Foster, Greer and Thorbecke (FGT) index. This is simply a weighted average of income gaps for individuals located below the country's median, expressed as:
    , where n is the number in the sample, q is the number below the median, z is the country's median, and y_i is the income of individual i. One interesting property of the FGT index is that more weight is given to workers away from the earnings threshold (z). The FGT index was 0.032 for Australia and 0.102 for Canada, suggesting that the earnings of Australian workers below the median were much less dispersed than those of Canadian workers.
12. As Frenette, Green and Picot 2006 showed, individuals at the bottom of the distribution may not be covered identically by different data sources. While there is no obvious solution to this problem, it may have an impact on distributional differences between Canada and Australia at the bottom of the income distribution.

References

Authors

Sébastien LaRochelle-Côté is with the Labour and Household Surveys Analysis Division. He can be reached at 613-951-0803. Claude Dionne is with the Income Statistics Division. He can be reached at 613-951-5043. They can also be reached at perspectives@statcan.gc.ca.