Moonlighting Trends and Related Policy Issues in Canada and the United States

Canadian Public Policy

Moonlighting Trends and Related Policy Issues in Canada and the United StatesAuthor(s): Jean Kimmel and Lisa M. PowellSource: Canadian Public Policy / Analyse de Politiques, Vol. 25, No. 2 (Jun., 1999), pp. 207-231Published by: University of Toronto Press on behalf of Canadian Public PolicyStable URL: http://www.jstor.org/stable/3551889 .

Accessed: 15/06/2014 21:04

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

University of Toronto Press and Canadian Public Policy are collaborating with JSTOR to digitize, preserveand extend access to Canadian Public Policy / Analyse de Politiques.

http://www.jstor.org

This content downloaded from 185.2.32.90 on Sun, 15 Jun 2014 21:04:36 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=utp

http://www.jstor.org/action/showPublisher?publisherCode=cpp

http://www.jstor.org/stable/3551889?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Moonlighting lrends and Related

Policy Issues in Canada and

the United States

JEAN KIMMEL

WE. Upjohn Institute for Employment Research Kalamazoo, Michigan LISA M. POWELL

Queen's University Kingston, Ontario

L'objet de cet article est de fournir une comparaison d6taillde des tendances du travail au noir au Canada et aux Etats-Unis et d'estimer les causes possibles de ces tendances. Les statistiques d6montrent que les deux pays ont connu de fortes hausses dans les taux de travail au noir chez les femmes, les personnes qui ne se sont jamais marides, les jeunes et les travailleurs qui oeuvrent dans le secteur des services alors que les personnes avec une

6ducation universitaire ont constamment maintenu des taux 61ev6s de travail au noir. Les travailleurs au noir amdricains demeurent plus sujets a combiner un emploi 'a temps plein et un emploi a temps partiel alors que les canadiens deviennent davantage des d6tenteurs de plusieurs emplois a temps partiel. Nous examinons la mesure dans laquelle les changements dans les taux de travail au noir sont caus6s par des changements dans la composition de la population active, des facteurs provenant du c6t6 de l'offre d'emploi et des facteurs provenant du c6t6 de la demande d'emploi. Des recommandations de politiques concernant la transition du bien-&tre social au travail et la protection infantile, les taxes sur la masse salariale et les avantages sociaux concluent.

The purpose of this paper is to provide a detailed Canada-United States cross-country comparison of moonlighting trends and to assess the possible underlying causes of such trends. The statistics show that both countries have experienced strong increases in moonlighting rates for women, never-married individuals, young persons, and service workers, while university-educated persons consistently have maintained high rates. United States' moonlighters remain more likely to combine a full-time job with a part-time job, while Canadians are increasingly becoming holders of multiple part-time jobs. We examine the degree to which changes in moonlighting rates are driven by labour force compositional effects, labour supply-side factors, and labour demand-side factors. Labour market policy recommendations are made with respect to welfare- to-work transitions and child care, payroll taxes, and non-wage benefits.

INTRODUCTION

Changes in moonlighting, or multiple job-holding,

have occurred alongside many other labour market transformations experienced by both Canada

and the United States over the last few decades. The increased prevalence of moonlighting has not gone unnoticed by researchers on either side of the border. In Canada, Cohen (1994) and Webber (1989) have characterized moonlighters and have highlighted

CANADIAN PUBLIC POLICY - ANALYSE DE POLITIQUES, VOL. XXV, NO. 2 1999



208 Jean Kimmel and Lisa M. Powell

increases in the incidence of moonlighting among women, young persons, part-time primary job- holders, and service workers, while Pold (1995) has documented increases in the moonlighting rates

among families. In the US, Stinson (1986, 1990) and Sekscenski (1980) have described the strong increases in moonlighting for women, while Levenson

(1995) has highlighted the increased prevalence of

multiple job-holding among college-educated workers and, in particular, college-educated women.

This paper examines trends in moonlighting over time for both females and males in Canada and the United States. We describe the trends in moonlighting rates across a variety of characteristics and highlight changes in the composition of moonlighters. Our analyses link the trends in moonlighting to other labour market changes and differing institutional structures in order to assess possible underlying causes. We examine the degree to which changes in

moonlighting rates are driven by labour force

compositional effects, labour supply-side factors, and labour demand-side factors.

We also provide an assessment of the relative

plausibility of these alternative explanations for

multiple job-holding. To do so we draw on two sources of information. First, we review econometric evidence that examines hypotheses for multiple job- holding such as hours constraints and job heterogeneity. Second, we tabulate survey data that provide self-reported reasons for taking a second job.

Finally, we draw on our results to make a series of conditional policy recommendations. An examination of the potential underlying reasons for mul-

tiple job-holding can have implications for changes to existing public policies or the introduction of new policies to improve labour market conditions. For example, recent changes to the Canadian unemployment insurance program (now called employment insurance) have extended coverage to include jobs under 15 hours per week (Government of Canada 1995). This public policy amendment eliminates the

pre-existing distortion in the non-wage payroll com-

ponent of part-time versus full-time compensation. And, like their full-time counterparts, it provides workers holding multiple part-time jobs access to insurance against job loss.

MOONLIGHTING TRENDS

Moonlighting rates in Canada and the US have risen over the last few decades in the midst of an increas-

ingly changed work world. In this section, we examine and compare the trends in moonlighting rates across a variety of characteristics and describe the changes in the composition of moonlighters. Be- fore delving into the statistics, we provide a brief overview of general labour market changes that have occurred in Canada and the US over the last two decades, upon which we will expand throughout our

analyses.

Both Canada and the US have experienced increases in the percentage of their working- age population who are employed mainly as a result of increases in labour market activity by women. How- ever, labour force participation among young workers of both sexes in Canada and for very young male workers in the US has fallen throughout the 1990s.

Unemployment rates in the two countries were similar in the 1960s and 1970s but diverged after the 1981-82 recession, as the rates in Canada remained

relatively high and those in the US returned to lower levels. Work patterns have changed dramatically in both countries. Indeed, beyond changes in moon-

lighting, both countries have experienced increased

part-time work (particularly involuntary part-time employment), more temporary work, and increased

self-employment. Also, workers, on average, have suffered from stagnant real wages and increased

earnings polarization. Poverty rates have increased as well, in particular for children. Income levels have also become more polarized, although to a greater degree in the US as the Canadian social safety net cushioned changes in the income levels of many




Moonlighting Trends and Related Policy Issues in Canada and the United States 209

Canadians. The industrial areas in which people work have also changed: both countries have

experienced a shift from the manufacturing sector to the service sector. Finally, the institutional structures within which employers and employees oper- ate have also experienced change. For instance, unionization rates and real minimum wages have fallen in the United States. In both countries, the non-wage component of employee remuneration has increased as payroll taxes have risen and non-wage benefits

play an increasingly important role in compensation.

Let us now turn our attention to moonlighting trends. The Canadian data for our trend analyses are drawn from the Survey of Work Arrangements for the year 1991 and. the Labour Force Survey for the

years 1981, 1985, and 1995. The US data for the

years 1981, 1985, and 1991 are drawn from the May supplement of the Current Population Survey (CPS). Statistics for 1994 come from unpublished tables from the US Bureau of Labor Statistics.' We include in our sample individuals aged 17-64 with positive hours of work (including students) and we omit un-

paid family workers.

Trend Statistics Tables 1-3 provide moonlighting rates over time across a variety of characteristics and by gender for both Canada and the United States.2 The changes in the distribution of moonlighters are found in Tables 4-6. Both Canada and the US have experienced increases in their overall rates of moonlighting since the early 1980s, although to a greater extent in Canada. In Canada, there has been roughly a 50-

percent increase in the incidence of moonlighting, from 3.38 percent in 1981 to 5.04 percent in 1991. In the US, the rate reached 6.01 percent in 1991, up from 4.62 percent ten years earlier. In the last few years, the incidence of moonlighting has continued to climb

steadily in Canada, reaching 5.47 percent in 1995, while it leveled off at just under 6 percent in the US.

The moonlighting rate for women has increased substantially in both countries. In fact, in the US it

is the increasing female rate that has driven the overall increase in the incidence of moonlighting, as the male rate has been more or less flat over the last two decades. In 1991, the female moonlighting rates in Canada and the US were roughly equal at 5.3

percent, up by 2.29 and 1.17 percentage points, respectively, from 1981. In Canada, the male moon-

lighting rate has been growing, albeit more slowly than the rate for women, hence by 1991 the female rate had surpassed the male rate. Indeed, between 1981 and 1995 the female moonlighting rate doubled in Canada. Furthermore, women now make up 50

percent of all moonlighters in Canada due both to their increasing moonlighting rates and increased

employment rates. In the US, during the 1980s, the female share among moonlighters remained roughly constant at about 40 percent, but by 1994 women accounted for 46 percent of all moonlighters. These trends in female moonlighting mirror the overall labour market trend for North American women workers, namely that their labour force behaviour has been becoming more like that of men for many years. This is driven in part by rising education levels and changes in occupational choices.

Differences in rate changes are also observed across age groups. In both countries, moonlighting rates have increased for young persons (age 17-24). By 1991 in Canada, young persons had the highest moonlighting rate (5.42 percent) among all age cat-

egories. This rate increased to 7.3 percent in 1995, driven by the escalation in the moonlighting rate for

young women to 8.63 percent. In the US, while the

percentage increase in moonlighting rates was the

highest for young persons, those aged 25-44 continued in 1991 to have the highest moonlighting rate

among age categories at 6.57 percent. By 1994, young persons in the US moonlighted at rates greater than any other age group. While the incidence of

moonlighting increased for young workers in both countries, the share of young persons among moonlighters actually decreased over the same period due to a smaller share of youth workers and increasing youth unemployment.





TABLE 1 Incidence (in percent) of Moonlighting Over Time in Canada and the United States for All Workers, by Characteristics

Canada United States

Characteristics 1981 1985 1991 1995 19811 1985 1991 1994

All workers 3.38 3.99 5.04 5.47 4.62 5.18 6.01 5.9

Age: 17-24 years 3.04 3.77 5.42 7.30 4.00 5.17 5.43 6.3 25-44 years 3.72 4.14 5.15 5.58 5.30 5.65 6.57 6.2 45-64 years 2.99 3.82 4.61 4.41 3.86 4.26 5.14 5.6

Education: 0-8 years 2.68 2.96 2.61 3.06 4.95 2.31 3.31 na Secondary school 2.95 3.30 4.04 4.25 3.53 4.38 4.65 na Some post sec./diploma 3.77 4.60 6.07 6.11 5.47 6.11 7.22 na University degree 5.22 5.89 6.21 6.94 6.98 6.88 8.04 na

Students: 3.83 4.29 3.34 6.07 na 2.01 5.31 na Full-time students 3.45 3.74 3.13 5.04 na 1.72 4.85 na Part-time students 5.45 6.83 4.48 13.53 na 6.36 8.14 na

Marital Status: Married 3.38 4.12 5.02 5.20 4.57 5.11 5.96 5.6 Never married 3.26 3.80 5.11 6.31 4.36 5.34 5.83 6.4 Other 3.79 3.49 5.06 5.06 5.27 5.23 6.58 6.2

Children aged 0-5 3.48 4.45 5.07 5.52 na na 6.26 na Part-time primary job na 6.83 na 10.69 5.56 6.91 7.37 na Full-time primary job na 3.46 na 4.34 4.48 4.91 5.81 na Occupation:

Managerial 4.27 4.95 5.39 6.16 6.53 5.81 6.54 na Clerical 2.77 3.33 5.54 5.18 3.24 4.64 6.15 na Sales 2.77 4.08 4.91 5.79 4.02 5.32 5.28 na Service 3.76 4.19 5.33 6.22 6.39 6.30 7.35 na Primary 5.76 5.82 8.69 8.55 4.96 5.52 6.30 na Processing 2.37 2.69 4.03 3.32 3.11 4.09 4.98 na Construction, transportation,

material handling 3.03 3.15 3.16 4.13 3.37 4.03 4.44 na Industry:

Agriculture 7.09 7.22 8.96 9.88 6.98 5.15 7.40 na Other primary 2.80 2.59 4.33 3.57 3.52 3.95 3.58 na Manufacturing, non-durable 1.97 2.75 3.98 3.21 2.80 4.08 4.94 na Manufacturing, durable 1.93 2.45 2.33 2.60 3.46 4.26 5.32 na Construction 2.58 3.65 3.64 4.18 3.82 4.42 3.74 na Transportation 4.00 3.98 3.89 4.96 4.48 4.45 6.32 na Wholesale trade 3.30 3.85 4.30 5.16 3.00 5.36 7.14 na Retail trade 2.67 3.65 5.59 5.35 4.42 4.54 4.52 na Finance 2.98 3.09 5.26 3.96 4.66 4.83 4.82 na Community services 4.92 5.71 6.97 7.94 6.42 6.95 7.59 na Personal services 3.01 3.80 4.50 6.18 5.04 4.21 5.00 na Business services 3.82 4.74 5.26 5.86 4.12 4.53 6.00 na Public administration 4.02 3.69 4.83 5.12 6.01 7.85 10.00 na

Number of observations 60,707 56,639 29,875 51,887 14,469 59,720 14,727 na

Note: 'In the US, 1981 data exclude the self-employed.





TABLE 2 Incidence (in percent) of Moonlighting Over Time in Canada and the United States for Males, by Characteristics



All male workers 3.61 4.13 4.82 4.98 4.98 5.72 6.56 5.9 Age:

17-24 years 2.97 3.71 4.34 6.04 3.90 5.25 6.06 5.6 25-44 years 4.04 4.19 5.07 5.17 5.85 6.26 7.10 6.3 45-64 years 3.29 4.28 4.56 4.18 4.20 4.95 5.67 5.7














TABLE 3 Incidence (in percent) of Moonlighting Over Time in Canada and the United States for Females, by Characteristics



All female workers 3.02 3.79 5.31 6.07 4.17 4.49 5.34 5.9 Age:








Agriculture 4.50 7.87 8.53 11.48 3.17 5.43 7.66 na Other primary 4.42 1.00 3.77 4.81 1.12 3.44 0 na Manufacturing, non-durable 1.61 2.18 4.26 3.13 2.19 2.55 5.20 na Manufacturing, durable 0.90 2.29 3.01 1.75 2.51 2.37 3.63 na Construction 5.06 6.08 9.59 7.40 2.36 5.85 4.30 na Transportation 3.63 4.49 5A16 5.54 5.31 4.13 5.60 na Wholesale trade 2.81 4.73 3.93 4.62 1.99 4.97 6.87 na Retail trade 2.44 3.34 4.76 5.90 4.07 4.51 3.78 na Finance 2.53 2.53 6.33 3.10 4.21 4.31 4.20 na Community services 3.73 4.57 6.20 7.76 5.33 5.72 6.55 na Personal services 3.14 3.85 5.45 6.15 3.99 3.38 5.12 na Business services 4.60 5.28 4.52 6.74 5.40 3.98 5.06 na Public administration 2.66 2.41 3.40 5.62 3.22 4.08 5.95 na







TABLE 4 Distribution of Moonlighters Over Time in Canada and the United States for All Workers, by Characteristics



Age: 17-24 years 20.4 18.8 16.6 18.2 19.4 18.5 13.1 16.3 25-44 years 56.7 57.1 58.7 58.4 57.6 59.0 63.3 55.9 45-64 years 22.9 25.1 24.7 23.5 23.0 22.5 23.5 26.6




Children aged 0-5 18.5 17.8 17.9 15.2 na na 20.0 na Part-time primary job na 26.8 na 34.7 15.6 18.0 16.0 22.6 Total hours 45.2 43.8 43.3 43.9 51.2 52.4 51.9 48.2 Occupation:




Number of observations 2,486 2,571 1,606 3,076 737 3,291 941 na






TABLE 5 Distribution of Moonlighters Over Time in Canada and the United States for Males, by Characteristics



Male (% of all workers) 59.7 59.9 51.1 50.2 59.5 62.0 60.0 54.0 Age:









Number of observations 1,676 1,527 877 1,528 442 1,984 536 na






TABLE 6 Distribution of Moonlighters Over Time in Canada and the United States for Females, by Characteristics



Female (% of all workers) 40.3 40.1 48.9 49.8 40.5 38.0 40.0 46.0 Age:








Agriculture 2.0 3.1 2.8 3.3 0.6 1.4 1.5 na Other primary 1.1 0.2 0.4 0.5 0.1 0.3 0 na Manufacturing, non-durable 5.1 4.8 5.1 3.3 4.8 4.3 6.2 na Manufacturing, durable 1.3 2.3 1.8 0.9 5.3 3.7 4.0 na Construction 2.3 2.1 2.4 1.5 0.4 1.5 0.9 na Transportation 5.8 5.4 4.2 4.2 5.9 4.1 5.1 na Wholesale trade 2.9 3.5 2.3 2.1 1.0 2.9 3.3 na Retail trade 12.8 13.7 13.1 13.8 18.0 19.2 12.8 na Finance 7.3 5.3 9.5 4.0 8.9 8.3 6.5 na Community services 32.1 32.6 34.8 38.3 42.6 41.4 45.0 na Personal services 12.1 13.1 12.8 12.4 5.1 4.5 5.6 na Business services 10.0 9.7 6.6 10.4 3.7 4.0 4.4 na Public administration 5.4 4.3 4.3 5.2 3.5 4.4 4.8 na

Number of observations 810 1,044 726 1,548 295 1,307 405 na






Turning to education categories, university graduates in both countries continue to have the highest moonlighting rates, reaching 6.94 percent in 1995 in Canada, and 8.04 percent in 1991 in the United States. By 1991, higher-educated Canadian women

moonlighted at rates greater than comparable men, while in the lower education categories male rates continued to dominate. In the US, men continued to

moonlight at higher rates than their female counter-

parts in all education categories. In both countries, by 1991 the proportion of moonlighters with at least some postsecondary education was roughly 60 percent, and those with university degrees made up one- fifth and one-third of all moonlighters in Canada and the US, respectively. The largest changes in the distribution of education among moonlighters occurred in Canada, where, as a result of increases in

moonlighting rates for more highly educated females combined with an increasing share of such women in the workforce, the proportion of Canadian female

moonlighters with postsecondary education increased from approximately 35 percent in 1981 to 73 percent in 1995. The corresponding trend in this

proportion for US female workers is from 48 percent to 58 percent in 1985. More educated workers are likely to seek full-time work hours, and rela-

tively poor full-time job prospects in Canada (due to persistently high overall unemployment rates) may force these female workers to combine multi-

ple part-time jobs to approach full-time hours.3 Ad-

ditionally, these findings with regard to education contradict the implication in the popular media that

moonlighting is an "affliction" of the most disad-

vantaged workers. On the contrary, because each additional foregone hour of leisure faces a rising mar-

ginal valuation, if the substitution effect dominates then those who are most likely to moonlight, other things held equal, would be those with the relatively greatest wage opportunities on the second job. Also, workers with higher education levels are more likely to be salaried on their primary jobs, and so extra hours worked on the primary job will not increase earnings.

Students also have exhibited rising moonlighting rates. Due to US data constraints, students are

restricted to the ages 17-24, hence the categoriza- tion reflects both age and student status. In Canada, student moonlighting rates rose from 3.83 percent to 6.07 percent from 1981 to 1995, while in the US the rates more than doubled just from 1985 to 1991, rising from 2.01 percent to 5.31 percent. Breaking students into full-time and part-time groups reveals that both subgroups have experienced large increases in moonlighting rates. The most striking increase can be seen for Canadian female part-time students who nearly tripled their moonlighting rates, rising from 5.45 percent to 18.79 percent. However, excluding students from our overall samples for each

year does not significantly affect the overall moon-

lighting rates for each year.

Looking at marital status, in Canada, never- married individuals switched from having the low- est moonlighting rate (3.26 percent) among the marital status groups in 1981 to having the highest rate (5.11 percent), by a small margin, in 1991; by 1995 the rate reached 6.31 percent. The rates for the never- married group were up for both genders but, in particular, for never-married women (7.49 percent in

1995). These results are consistent with the strong increases in moonlighting rates witnessed for young persons. In the US, the never-married and "other" marital groups experienced significant increases over the 1981-91 period, with the "other" group re-

cording the highest rate of all groups by 1991.4

However, by 1994 in the US, similar to Canada, never-married individuals emerged with the highest rates. Again, this result was consistent across

gender. While the proportion of moonlighters who are married has declined in both countries, driven

by declines in the proportion of married male moon-

lighters, married individuals still comprise the ma-

jority of all moonlighters.

The incidence of moonlighting for individuals with children aged 0-5 increased in Canada from 3.48 percent to 5.07 percent between 1981 and 1991 and reached 5.52 percent in 1995. Unfortunately, we do not have data on this characteristic for the US for the years prior to 1991. However, we should note





that the Canadian rate, even in 1995, still remained lower than the 1991 US rate of 6.26 percent. Note that the US rate differs substantially by gender (7.89 percent for males versus 3.90 percent for females). There are also gender differences worth noting for Canada. In 1981 and 1985, men with young children moonlighted at rates greater than those for women with young children, but by 1991 this trend had reversed itself. And, by 1995, women with children under the age of six were moonlighting at a rate of 5.59 percent in Canada. This suggests a re- versal in the relative strength of the income and substitution effects for women in Canada, implying that the desire (or need) for extra income supersedes the desire for extra time with children. In the US in 1991

(from calculations not appearing in the tables), extra information can be garnered by looking at marital status along with the presence of young children. Divorced or never-married fathers of young children

moonlight at relatively high rates. In fact, divorced fathers with young children moonlight at a rate of 12.88 percent, twice as high as the general population. This shows that marital status is important to the moonlighting behaviour of fathers.

Turning now to the changes in the hours moon-

lighters work, we see that the average total number of hours worked by moonlighters decreased over the 1980s by about two hours per week in Canada and

by only half an hour in the US, and remained significantly higher in the US at 51.9 hours per week

compared with 43.3 hours per week in Canada. In the US, this overall change in hours has been driven

by a fall in average hours of work by women (due to women entering the labour force with part-time work) that counteracted the increase in hours by males. In Canada, quite the opposite has occurred, with the overall change being driven by adjustments in hours worked by men: female moonlighters worked on average slightly more hours per week, reaching 39.2 hours per week in 1995, while males reduced their hours to 48.6 by 1995. This downward trend in hours in Canada is driven by fewer moonlighters working in the high tail of the hours distribution and more moonlighters working part-time

hours. As will be discussed later in the paper, this

may be driven in part by persistently high overall

unemployment rates in Canada.

The incidence of moonlighting among those who hold a part-time primary job reached 10.69 percent in Canada in 1995, up from 6.83 percent in 1985. The percentage increase in moonlighting rates

among this group was more than double the increase in the rate for those holding full-time primary jobs, rising from 3.46 percent in 1985 to 4.34 percent in 1995. Further, by 1995 just over one-third of all Canadian moonlighters held a primary job that was

part-time. For females, one-half of moonlighters held part-time primary jobs, but only 20 percent of male moonlighters do so. Also, we can see from the distribution of hours that 25 percent of female moon-

lighters remained part-time workers. This suggests that fully one-quarter of female moonlighters may be holding multiple jobs in order to attain full-time

weekly hours of work. On the other hand, one-quarter of female moonlighters remain part-time workers even after holding a second job.

In the US, the incidence of moonlighting in 1981 was roughly a percentage point higher (at 5.56 percent) for workers with part-time versus full-time

primary jobs. By 1991, workers with a part-time primary job moonlighted at a rate of 7.37 percent, versus 5.81 percent for those holding a full-time primary job. This increased gap, however, is significantly smaller than the gap in rates between these two groups for Canada. However, similar to Canada, significant gender differences continue to exist; in 1991, the moonlighting rate for workers with a part- time primary job was 7.76 percent for women versus 6.44 percent for men. The proportion of moon-

lighters who hold a part-time primary job in the US rose only slightly over the 1980s reaching 16 percent, in 1991, but jumped to 22.6 percent in 1994. Indeed, US moonlighters are much more likely to be "job-packaging" (that is, adding a part-time job to a full-time primary job) than their Canadian counterparts, who are becoming increasingly likely to combine multiple part-time jobs. It is likely that the





causes for this full-time/part-time job difference across countries are related to full-time job avail-

ability and the non-wage benefits associated with full-time jobs. As mentioned earlier in the paper, the US has had a tighter labour market than Canada in recent years, suggesting that relatively more full- time jobs are available. Additionally, in the US, full- time hours with a single employer are usually re-

quired in order to be eligible for employer-provided health insurance coverage. These causes probably also explain why US female moonlighters are more like their male counterparts in packaging a full-time with a part-time job.

The incidence of moonlighting among the self-

employed has risen in both countries, remaining significantly above the overall average moonlighting rates in Canada and staying below the average rates in the US. Yet self-employed individuals in the US

experienced an above-average percentage increase in rates, increasing from 3.66 percent in 1985 to 4.89

percent in 1991. In Canada, the moonlighting rate for the self-employed reached 7.80 percent in 1995, driven by an increase in the rate among females which rose just above 9 percent in the 1990s.5

Now we turn to occupational status. Managerial, service, and primary occupations have consistently had above-average moonlighting rates since 1981.6 Consistent with the high agricultural sector moon-

lighting rates for Canada, Canadian men and women

working in primary occupations have continued to have the highest moonlighting rates across all occupational groups. While in 1981, managerial occupations scored the second highest moonlighting rates for both men and women in Canada (though in absolute terms the rate for men was substantially higher than that for

women), the rate for females in managerial positions increased so strongly that by 1995 it surpassed the rate for their male counterparts in this occupation. Indeed, the incidence of moonlighting among female managers rose from 3.73 percent in 1981 to 6.92 percent in 1995, and by 1995 the proportion of female moonlighters in managerial occupations represented 44 percent, up from a share of 32 percent in 1981.

In the US, managers held the highest moonlighting rates at 6.53 percent in 1981, but by 1991 the rate for service workers had increased to reach 7.35

percent, becoming the highest rate across all occupations. This change was driven by increases in the

already high moonlighting rate for male service workers, to 9.84 percent. While in 1981, females in US managerial positions had particularly high rates of moonlighting, by 1991 the rates for those in primary occupations had increased five full percentage points to reach 8.45 percent which, similar to their Canadian female counterparts, was the highest rate across all occupation groups in that year.

Among industrial groups, agriculture has had the

highest moonlighting rates historically, while the

majority of those who moonlight work in the service sector. In Canada, the agricultural industry continued to have the highest moonlighting rates (for both

genders) though the gap for the service industry has narrowed. In fact, over the 1980s, the rates of moon-

lighting in the service sector rose significantly in both countries. In particular for Canada, strong increases occurred for males over the 1981-91 period in the business service sector, and more recently in the personal service sector. The US also witnessed substantial increases in moonlighting rates in the service sector, in particular in community services. Also of note, US males working in the business service industry experienced increases as moonlighting rates rose from 3.40 percent in 1981 to 6.52 percent in 1991, while public administration moonlighting rates rose from 7.45 to 12.81 percent.

Unlike their US male counterparts, women in the US agricultural industry had actually moonlighted at rates below women in the service sector in 1981, but by 1991 this was reversed with a 4.5 percentage point increase in the incidence of moonlighting in the female agricultural sector. Canadian women also

experienced increases in the agricultural sector, with rates doubling from 1981 to 1991 to reach 8.53 percent, and then rising to 11.48 percent by 1995. The most significant change in the distribution of female moonlighters across industries was a further increase





in the proportion of those working in the already well-represented service sectors.7

EXPLANATIONS FOR THE TRENDS

Having described the overall changes in the rates of

moonlighting in both Canada and the US, we now examine potential underlying causes. There may be several contributing factors. First, the overall moon-

lighting rates may be driven by changes that have occurred in the composition of the workforce. That is, there simply may be more workers of the type who traditionally have had above-average moon-

lighting rates. Second, on the supply-side of the labour market, changes in multiple job-holding may be driven by changes in labour supply preferences. Changes in moonlighting rates may reflect a desire for increased flexibility by workers or the desire to secure non-wage benefits (such as health-care benefits in the US). It may also be a response to increased economic insecurity and hardship. Finally, demand-side factors highlight the degree to which the demand for part-time versus full-time labour, contract/temporary versus permanent workers, and

hourly versus salaried workers underlie the changes in moonlighting rates among North American workers. Labour market demand has been affected as firms increasingly try to minimize their wage bills and by changes resulting from distortions in the relative prices of labour due to the differential application of payroll taxes and benefits to various types of labour. To assess the degree to which these un-

derlying factors account for the changes in multiple job-holding, we review econometric evidence that examines hypotheses such as hours constraints faced

by workers on their primary jobs and job heterogeneity, and we draw on survey data that report reasons for taking a second job.

Compositional Effects The composition of the share of the workforce as a whole has evolved in Canada and the US over the last 20 years. An increase in the workforce for those groups who traditionally have had high rates of

moonlighting or of those who have experienced the

strongest increases in moonlighting will serve to

push the overall moonlighting rate upwards. More

working women, more educated workers, more self-

employed workers, more service jobs, and more

part-time, contract, and temporary workers all fit this

description. Changes in the type of employed workers such as part-time, contract, and temporary will be examined under demand-side factors.

Both Canada and the US have experienced significant increases in labour force participation by women over the last few decades. In addition to

strong increases in the 1950s, 1960s, and 1970s, employment among women has continued to rise since the early 1980s, particularly for women with

young children. In Canada, data from the Labour Force Survey show that the employment rate for all women rose from 48.9 percent in 1981 to 52.8 percent in 1991. This rate has leveled off in the 1990s, equaling 52.1 percent in 1995. Women in their prime working years (aged 25-44) experienced a ten-percentage point increase in the employment rate over the period 1981-91, reaching a rate of 70.7 percent in 1991. In the US, the employment rate of women rose from 47.7 percent in 1980 to 54.4 percent in 1990. By 1994 this rate had increased just slightly to 55.3 percent.

Recent declines in the labour force participation rates among young workers (under age 25) in Canada, however, have dampened the overall effects of the increases in moonlighting rates for this age group. Indeed, recall that in Canada, moonlighting rates increased from 5.42 percent in 1991 to 7.30 percent in 1995 for young workers, while Labour Force Survey data show that between 1991 and 1995 the labour force participation rates among individuals 15 to 24 years of age fell from 67.4 percent to 62.2 percent. In the US, overall participation rates for the young have been more or less flat since 1990, but did fall in the 1980s (US Department of Com- merce, Bureau of the Census 1996).

Both countries have witnessed a strong shift from the manufacturing to service industries. According





to Houseman (1995), the percentage of employment in manufacturing in the US has declined steadily, from 28.7 percent in 1969 to 16.2 percent in 1993. The percentage in services rose from 15.8 to 27.4

percent over the same time period. Canadian time- series data from the Labour Force Survey show that from 1976 to 1996, the shares of employment changed from 20.3 percent to 15.2 percent in manu-

facturing and 27.1 to 37.6 percent in services. In addition, we have witnessed strong growth in knowl-

edge-intensive industries (Canada. HRDC 1995). "With globalization and the knowledge/information [era], the world is in the throes of one of its epic transformations. The information revolution will have impacts on the role of human capital not unlike the impact of the industrial revolution on the role of physical capital" (Courchene 1994, p. 232). Note that not only has there been an industrial shift in North America, but the now-increased service sector has also experienced strong growth in moon-

lighting rates.

Labour Supply-Side Factors Multiple job-holding can be thought of as volun-

tary, that is, reflecting individuals' preferences. Changes in the labour-supply preferences for hold-

ing multiple jobs may reflect the simple fact that individuals enjoy the non-pecuniary aspect of a second job; the desire for increased flexibility by workers, in particular by families; or, workers may work in one job to acquire non-wage benefits (in particular, health care in the US) and take a second job to

satisfy personal preferences. This latter issue con-

cerning employer non-wage benefits is discussed in the next section. For other individuals, the choice to hold multiple jobs may stem from increased economic insecurity and hardship.

The desire by workers for increased flexibility in combining work with other responsibilities increasingly may be being achieved by piecing together multiple jobs. Flexibility is no doubt becoming an increasing concern. "Flexible working-time arrangements often offer workers opportunities to carry out responsibilities in ways that better meet

today's standards of fairness and equity within the

family" (Canada. HRDC 1994, p. 9). In particular, women may be packaging their work hours differ-

ently in order to juggle family responsibilities such as child care.8 This would provide an explanation for the substantial increases in moonlighting rates

by women and, in particular, by those holding two

part-time jobs. However, recall that relative to Canada, US female moonlighters are more like their male counterparts in terms of adding a part-time secondary job to a full-time primary job and hence

job packaging.

The moonlighting rates of women with young children have grown steadily in Canada since 1981. Current research in Canada that examines the rela-

tionship between non-standard jobs and child-care

arrangements shows that the use of multiple child- care arrangements (including the use of a spouse and the use of multiple non-parental child-care ar-

rangements) by mothers who moonlight is substan-

tially higher relative to those working mothers hold-

ing standard jobs (Powell 1997a). While mothers

may choose to hold two separate jobs to allow for increased parental care by the spouse, the above-

average use of multiple non-parental arrangements by mothers holding dual jobs may reflect a lack of

flexibility within the child-care market. We cannot

definitively say anything about the causation of this latter relationship, but a concern that arises here is the degree to which children are affected negatively by the instability inherent in being moved from ar-

rangement to arrangement. A preference for flex-

ibility may also be required increasingly among students who hold multiple part-time jobs to fit their school schedules and who have been facing substantial increases in tuition costs.

Let us now turn to economic hardship and insecurity that increasingly reflect the current landscape in which workers find themselves. "In today's economy the buoyant optimism of the 1950s and 1960s is missing. The result is a sense of insecurity that is especially harsh in the modern workplace whether it is experienced in the corporate board





room, in the managerial office, or on the plant floor"

(Canada. HRDC 1994, p. 8). In the face of increased economic insecurity, workers may well be more

willing to trade off leisure for work. Further, real

earnings in both countries have stagnated, requir- ing increased work efforts to maintain stable standards of living, if not to avoid poverty.

Unemployment rates in Canada have remained

relatively high since the 1981-82 recession. While the US unemployment rates have diverged down- wards away from the Canadian rates, employment ratios in the two countries have remained fairly similar (Card and Riddell 1993). The duration of unem-

ployment spells in both countries has increased over the last few decades, exacerbating the fear of un-

employment (Canada. HRDC 1995; Layard, Nickell and Jackman 1991; Horrigan 1987).

Growing economic insecurity in the US has no doubt been fueled by increased polarization of income and increased poverty rates over the last dec- ade (Danziger and Gottschalk 1995). Beach and Slotsve (1996) show that family income polarization in Canada did not increase over the last two decades. This is due to the pooling effects of dual earners, relatively small increases in individual earn-

ings inequality, and Canada's strong social safety net. Indeed, for the lower tail of the income distribution, Blank and Hanratty conclude that "poverty among both single-parent and two-parent families would decline dramatically if the United States

adopted the average Canadian antipoverty transfer

program" (1993, p. 219).

Economic hardship in North America has also increased among workers who have been able to secure jobs. Real earnings have stagnated in Canada for young women, while for young men they have fallen. Older workers have enjoyed slow but positive real wage growth (Canada. HRDC 1995). Changes in earnings polarization in Canada have also differed by gender; earnings polarization has increased for males, while it has decreased for females mainly due to an increase in full-time female

workers (Beach and Slotsve 1996). In the US, real

wages have also been stagnant throughout the 1980s, and earnings polarization has increased to a much

greater extent compared to Canada (Danziger and Gottschalk 1995; Mishel and Bernstein 1994). In- deed, decreasing (if not stagnant) real wages for

young persons provide a direct impetus for the

strong increase in moonlighting rates that have occurred among this group in North America.

Canada and the US have also experienced increases in the polarization of hours of work for all workers, although mean hours per week have remained fairly stable (Sunter and Morissette 1994). Indeed, not only are the increases in moonlighting rates contributing to the polarization of hours at the

upper tail of the distribution, but non-moonlighters are also increasingly working longer hours, no doubt in an attempt to recoup earnings.

Institutional factors such as unionization and minimum wages have also played a part in these labour market changes. The unionization rate in Canada has remained stable. In the US, decreased unionization and the fall in real minimum wages have contributed to the increase in earnings polarization (Riddell 1993). Lemieux (1993) shows that the presence of unions in both countries reduces overall wage variation for men. He also estimates that the difference in unionization between the two countries explains 40 percent of the Canada/US difference in wage inequality. Further, Kuhn notes that "these policies [relatively high unionization rates and the maintenance of minimum wage standards] may both have mitigated the amount of wage polarization that would otherwise have occurred in Canada, and may be responsible for polarization taking the form of increased long-term unemployment and greater age, or generation gap in Canada rather than the increase in the working poor population seen in the United States" (1996, p. 306).

With respect to higher educated earners, Freeman and Needels (1993) reported substantially higher increases in college/high school wage differentials





for the US versus Canada, attributable primarily to the faster relative growth in the number of univer-

sity graduates in Canada. However, although US

college graduates did experience relative wage gains and real wage growth throughout the 1980s, "older, white-collar workers were considerably more at risk of displacement [in 1991 compared to 1981] ... job displacement rates rose for more educated workers [and] these changes in the incidence of job displacement may be a reason for the reports of heightened anxiety regarding job loss" (Council of Economic Advisors 1996, p. 9).

Demand-Side Factors As firms attempt to minimize their wage bills, they have changed their hiring practices in the competi- tive market place. In addition, firms' demands for labour also have been distorted increasingly due to the differential application of payroll taxes and benefit packages to different types of employees, such as part-time versus full-time, contract versus paid, and temporary versus permanent.

The Economic Council of Canada (1990) reported that 44 percent of all new jobs in Canada created between 1981 and 1989 were short-term or part-time jobs. More recent evidence from Canada (Canada. HRDC 1994, 1995; Pold 1994; Krahn 1995; Nakamura 1995) has confirmed an increasing trend of non-standard work in the form of part-time jobs, temporary employment, and self-employment.

Over the period 1975-93 in Canada, the number of part-time jobs increased at an annual rate of 4.5

percent, while full-time jobs increased by only 1.2

percent annually (Pold 1994). Evidence from the US shows that while part-time employment has increased somewhat since 1980, the rate of involuntary part-time employment has risen by 45 percent over the period 1979-93 (Houseman 1995). House- man notes that these data suggest that the changes in part-time jobs are demand-driven. Indeed, rising numbers of moonlighters with part-time primary jobs, coupled with the increases described for such jobs, clearly contributes to the overall increase in

moonlighting rates for Canada. This is true to a lesser extent in the United States.

The 1994 (Canadian) Report of the Advisory Group on Working Time and the Distribution of Work highlights the move by firms to hire "just-in- time workers" to meet "just-in-time" production. Based on research from Statistics Canada, Human Resources Development Canada (1995) reports that the number of jobs lasting fewer than six months has increased significantly over the last 15 years. Our data available for Canada for 1991 (not detailed in the tables) show that temporary workers moonlight at above average rates. In the US, evidence from the Consumer Expenditures Survey shows that in 1982, 0.47 percent of the non-farm payroll was in temporary help and by 1995 this figure had risen to 1.83 percent (unpublished table from the US Bu- reau of Labor Statistics 1996).

Kuhn (1996), in his analysis of earnings polarization, refers to these types of changes in firms' behaviour as changes in "firms' human relations practices," highlighting that they consist of exten- sive contracting-out and the use of temporary and

part-time workers, while keeping only a small core set of full-time employees. Noting the lack of empirical evidence, he poses the question of the de-

gree to which such changes contribute to increased

earnings polarization. If, as suggested here, such changes are contributing to increased multiple job- holding, then we may well expect that, in the ag- gregate, potential effects on earnings polarization would be even greater if such individuals were not

taking on second jobs.

While firms try to minimize their wage bills by demanding non-standard jobs such as part-time, contract, and temporary work, changes in the relative

non-wage component of labour costs for different bundles of labour have also affected the demand- side of the labour market. Referring to part-time jobs, Gallaway succinctly notes "any additional factor which alters the relative cost of part-time vis- A-vis full-time employees ... [and] ... while these





reasons may be less important in determining the level of part-time hiring, they could be much more

significant in explaining changes in the level of such

hiring" (1995, p. 309).

These non-wage labour costs have been on the rise in both Canada and the United States. Payroll taxes (including UI, CPP, Worker's Compensation, and other health and education taxes) have increased

substantially in Canada over the last two decades

(Abbott and Beach 1997). In the US, "in 1973, fringe benefits accounted for 12.82 percent of all employee compensation, and ... for 1993 ... they account for 17.81 percent of compensation, an almost forty percent increase in the relative importance of fringe benefits" (Gallaway 1995, p. 309).

In Canada, employer payroll taxes such as un-

employment insurance have been payable by firms

only on hours of work above 15 hours per week, providing an incentive to offer part-time versus full- time jobs, all else being constant. In addition, this tax is not applicable for self-employed workers and hence those hired on contract. Only recently has the "under 15 hour" distortion been amended in the Cana- dian unemployment insurance program (Government of Canada 1995). Occupational pensions and other benefit packages also drive a wedge between the relative cost of part-time versus full-time employees and

permanent versus contract workers. Indeed, recent evidence by Lipsett and Reesor (1997) shows that job- related employee benefits differ drastically based on

job type, For instance, in 1995, the percentage of full- time versus part-time and permanent versus temporary employees entitled to occupational pension plans were 58.4 percent versus 18.7 percent and 55.5 percent versus 19.9 percent, respectively. Corresponding figures for entitlement were also reported for other benefits: health plans (other than provincial health care), 68.1 percent versus 17.8 percent and 64.4 percent versus 19.3 percent; dental plans, 63.4 percent versus 15.9 percent and 60 percent versus 16.5 percent; and, paid sick leave, 65.7 percent versus 17.8 percent and 62.2 percent versus 19.3 percent.

In the US, differential costs associated with hir-

ing full-time versus part-time employees relate

mainly to health insurance benefits and pensions. In 1993 "[a] much higher proportion of full-time

employees received employment-based health insurance coverage directly (61.2 percent) ... from an

employer than did part-time employees (16.4 percent)" (Snider 1995, p. 243). With respect to pensions, "[f]ifty-eight percent of full-time employees reported that their employer or union had a pension or other retirement plan (other than Social Security) for employees, [while] thirty percent of part-time employees reported such coverage" (ibid., p. 245).

Empirical evidence concerning the degree to which firms change the way in which they demand bundles of labour due to relative price distortions such as those noted above is sparse. Gallaway (1995) provides empirical evidence for the US on the im-

pact of fringe benefits on the demand for part-time employees and the volume of involuntary part-time work. His results show that "[f]ringe benefits have a powerful significant effect on part-time employment ... [a] one percentage point increase in fringes as a percent of employee compensation increases the percent involuntary part-time employed by 0.57

percentage points" (Gallaway 1995, p. 312). Abbott and Beach (1997) have provided recent evidence on the impact of payroll taxes in Canada on employment. Their results show that higher payroll taxes have disemployment effects on paid workers. Their particular study does not distinguish between part- time and full-time employment; however, their results do have implications regarding the differential behaviour by firms with respect to other aspects of labour demand. Their results are consistent with the hypothesis that increasing payroll taxes lead firms to shift their labour demand from paid em-

ployees to contracting-out.

Ranking the Explanations From the above analysis, it is clear that both Canada and the United States have experienced particular labour market and more general economic changes that in a variety of ways support each of the three





broad reasons - labour force compositional effects, supply-side factors, and demand-side factors -

posited as possible driving forces for the changes we have witnessed in multiple job-holding across the two countries. While it is difficult to compart- mentalize neatly the changes in moonlighting into a specific set of causal relationships and explicitly identify a relative ranking of such reasons, we can see particular trends in the motives for holding a second job. We examine existing econometric work that provides empirical evidence on various under-

lying hypotheses for multiple job-holding such as hours constraints on the primary job and job het-

erogeneity. We also tabulate data from the Canadian

Survey of Work Arrangements for 1991 and 1995 and the US Current Population Survey for 1991 and 1997 that explicitly ask questions regarding reasons for holding a second job.9 An examination of these

data allows us to provide evidence on the changes in

reported reasons for moonlighting over that time frame.

The econometric evidence concerning the moti- vation for moonlighting points to primary job constraints as a major reason for taking a second job (see Conway and Kimmel [1998] for a summary of the existing evidence). These constraints typically can be interpreted as the inability to increase earn-

ings on the primary job by increasing hours. For salaried workers, this earnings rigidity is well- known, but hourly workers are often unable to increase their working hours at will as a means for

(temporarily or long-term) increasing earnings. For

descriptive evidence of the importance of primary job constraints, Table 7 shows that in Canada in 1995, and in the US in 1997, roughly 65 percent and 59 percent, respectively, of workers reported

TABLE 7 Main Reasons for Undertaking Moonlighting in Canada and the United States

Canada

Both Sexes Males Females Both Sexes Males Females

% % % % % %

Meet regular household expenses 34.2 34.0 34.3 40.1 35.5 45.6 Pay off debts 11.4 12.4 10.7 5.2 4.0 6.2 Buy something special 4.7 4.5 4.8 1.2 0.0 2.3 Save for the future 12.6 13.1 12.2 18.2 22.2 14.7 Gain experience/build business 10.9 9.8 11.7 9.7 7.4 11.7 Enjoys the work of second job 15.2 14.9 15.4 17.3 22.0 13.2 Other 11.0 11.2 10.9 7.5 8.8 6.4

United States

Both Sexes Males Females Both Sexes Males Females

% % % % % %

Meet regular household expenses 31.2 28.0 36.0 30.9 29.4 32.7 Pay off debts 11.0 11.2 10.7 10.9 10.4 10.7 Buy something special 7.2 6.7 7.9 7.9 7.4 8.5 Save for the future 10.7 11.1 10.1 8.7 10.1 6.9 Gain experience/build business 7.8 7.1 8.7 7.7 8.4 7.0 Enjoys the work of second job 13.9 15.2 11.9 14.5 15.9 12.8 Other 18.4 20.8 14.7 19.8 18.5 21.3





moonlighting to meet regular expenses, pay off debts, buy something special or save for the future. These reasons can be considered to reflect each of the factors that we discussed as important to moon-

lighting behaviour, compositional effects (increased labour force participation by those individuals more

likely to face constraints), labour supply (an increase in the number of individuals who need/wish to increase earnings), and labour demand-side (a change in the way labour is demanded resulting in increased hours constraints).

The more recent econometric studies of moon-

lighting (Lilja 1991; Averett 1996; Powell 1997b; Conway and Kimmel 1998) reveal that constraints can only partially explain moonlighting. Differen- tial non-pecuniary costs and benefits on the two jobs also play a significant role in moonlighting choices. As shown in Table 7, a significant percentage of workers moonlight in order to accumulate valuable human capital, or just because they enjoy the work on their second jobs. Not detailed in this table is the

importance of non-wage benefits, particularly in the United States. Workers might continue with a less than satisfying full-time primary job in order to re- tain health benefits and take a second job for personal fulfillment. The notion that heterogenous jobs motivate moonlighting behaviour reflects both supply- side (non-pecuniary individual benefits) and labour demand (non-wage employer benefits). As noted above, the reported reasons for moonlighting from our data sources do not allow us to quantify the labour- demand aspect of the job heterogeneity motive.'1

Drawing on our survey data, we can examine the changes over time in the reported reasons for moonlighting. Our time frame is 1991 to 1995 for Canada and 1991 to 1997 for the United States. We classify the reported main reasons for holding a second job in terms of implied hours constraints based on the desire to increase earnings (meet regular household expenses, pay off debt, buy something special, and save for the future) and job heterogeneity (gain experience/ build business and enjoys the work of the second job). We also think it is useful to exam-

ine a subset (meet regular expenses and pay off debt) of the former reasons in order to highlight changes in moonlighting due specifically to economic hardship.

From Table 7, we see that the desire to increase earnings and hence the implicit constraint on hours in the primary job is the most widely (over 50 percent) reported reason for taking a second job in both countries. Canadian workers reported an increase in this reason from 62.9 percent to 64.7 percent. This stems from a reported increase for women from 61.8 percent to 68.8 percent, as the percentage of men who reported this reason fell over this time frame. Indeed, just over half of Canadian women took a second job due to economic hardship in 1995 (45.6 percent to meet regular household expenses). In the United States, however, opposite to Canada the percentage of moonlighters reporting financial reasons fell slightly from 60.1 percent to 58.7 percent; up by just less than a percentage point for men and down from 64.7 percent to 58.8 percent for women.

With respect to reported reasons for moonlighting that might reflect job heterogeneity, while there was a very small overall reported increase in both countries, differences exist by gender. In both countries more men and fewer women reported that they moonlighted to gain experience, build a business or because they enjoyed the work of the second job. This may reflect the higher opportunity costs for women associated with work outside the home.

Finally, it is interesting to note the changes over time for women who report the category of "other" as a reason for moonlighting. For Canadian women this category dropped slightly from 10.9 percent to 9.4 percent, while it was up from 14.7 percent to 21.3 percent for their US counterparts. This suggests that women in the US increasingly may be taking a second job for other reasons such as securing non- wage benefits like health care or to juggle family responsibilities by using non-parental child care for one job and working a second job during hours that other family members can care for their children."





Overall, our survey data suggest that in Canada and the US, the main reason for moonlighting con- tinues to be financial for both sexes. For the majority of these workers, the underlying motive is economic hardship.

CONCLUSIONS AND POLICY ISSUES

What do we learn from this moonlighting trend com-

parison? Both countries have experienced strong increases in moonlighting rates for women, never- married individuals, young persons, and service workers, while university-educated persons consist-

ently have maintained high rates. US moonlighters remain much more likely to be "job-packaging" (combining a full-time job with a part-time job), while their Canadian counterparts, particularly female Canadians, increasingly are becoming multi-

ple part-time job-holders. This is highlighted by the fact that US moonlighters continue to work, on

average, longer hours relative to Canadian moon-

lighters, while the incidence of moonlighting for workers with a part-time primary job has risen significantly for the latter group. Indeed, it is important to note that as individuals increasingly work in two part-time jobs to acquire full-time hours of work, we must be wary of reporting full-time/part- time hours statistics rather than full-time/part-time job statistics. By reporting the former we may be

masking the work situation of an increasing proportion of workers who, while working full-time hours from a combination of jobs, do not receive the benefits usually associated with full-time work.

Our analysis showed that evidence exists to support each of the three broad reasons - labour force

compositional effects, supply-side factors, and demand-side factors - posited as possible driving forces for the changes we have witnessed in multiple job-holding across the two countries. Based on reported reasons for taking a second job from survey data, we were able to partially assess these factors. The data showed that in both countries, the majority of workers continue to hold a second job

for financial reasons. The changes in the relative

importance of the underlying motives for multiple job-holding were found to differ by gender and coun-

try. Fewer men in both countries are reporting that

they take on a second job to increase earnings (implying that fewer males are facing hours constraints that can stem from each of the three factors above), while a greater proportion are supplying labour to a second job due to job heterogeneity. While the

moonlighting rate for women has increased in both countries, unlike their US counterparts, a greater number of Canadian women are reporting financial reasons, in particular, economic hardship, as their

primary motive for holding a second job. Increas-

ingly, American women are moonlighting for "other" reasons, reporting a decrease in financial and

job heterogeneity motives.

In order to reduce the demand-side factors that contribute to the likelihood that workers face constraints on their primary jobs, policymakers should re-examine the existence of any legislation that in- troduces distortions in the way in which firms demand bundles of labour. For instance, in Canada, the recent changes to the unemployment insurance

program, which have extended coverage to include

jobs under 15 hours per week, has eliminated the distortion in the non-wage payroll component of

part-time versus full-time jobs. Unemployment insurance (UI) in the US, however, remains far less

moonlighter-friendly. Unemployment insurance

eligibility criteria vary by state, but overall, only about one-third of all unemployed workers in the US receive UI. All but the most highly paid long- term part-time workers are excluded from UI due to the eligibility rules based on earnings and hours per relevant time period. A recent report issued by the US Advisory Council on Unemployment Compen- sation (1996) advised that states' extend their eligibility to permit greater access for low-wage and part- time workers, but such changes are not anticipated in the near future.

The prevalence of multiple job-holding in itself is not disconcerting if such labour market activity





is driven by voluntary choices on both sides of the labour market. Concern does arise, however, when such a group is comprised mainly of involuntary workers and/or those who are marginalized through the exclusion to benefits and training. As discussed in this paper, part-time versus full-time and contract versus permanent workers are much less likely to be entitled to occupational pension plans, health care

(US), supplementary health benefits (Canada), dental benefits, and paid leave. Hum and Simpson (1996) highlight the fact that part-time workers receive less training than their full-time counterparts and explain that this may have adverse implications for productivity growth. Unfortunately, we cannot

expect new government policy measures to reverse this trend given that government resources devoted to active labour market measures have been decreas-

ing. Both Canada and the United States currently have a low international ranking among OECD countries (15th and 21st out of 23, respectively) for labour market program public expenditure as a percent of GDP (Sharpe and Haddow 1997). Sharpe and Haddow (1997) note that the recent development of Labour Force Development Boards in Canada were

expected to provide a forum to develop, among other

things, recommendations for eligibility criteria for income support for training programs. However, the

potential to address the part-time/full-time dilemma noted above effectively through this medium is un-

likely given the general lack of success and closure of such boards in several provinces.

The exclusion of the contingent workforce from

job-related benefits also has serious implications for the welfare-to-work transition. Many low-skill, low- income individuals may only be able to make such transitions in the form of non-standard work, say, by piecing together multiple jobs to generate suffi- cient earnings to overcome the welfare tax rate hur- dle. In the US, low-skill single parents may face trad-

ing off welfare payments plus access to Medicaid to join the labour market, while Canadian social assistance recipients trade off cash assistance plus a host of social services and benefits.

In Canada, the Self-Sufficiency Project (SSP) and the Earnings Supplement Project (ESP) are experi- ments that are designed to aid in the welfare-to-work transition and to help displaced workers on employment insurance (EI) re-enter the workforce. While the projects have similar underlying goals, they have

quite different eligibility requirements regarding hours of work. The SSP, operating in two provinces, is intended to encourage the welfare-to-work transition by offering wage subsidies and a relatively low marginal tax rate. Supplement payments are made to single parents who have been on welfare for at least one year and who work full-time. This

program is moonlighter-friendly in that it allows individuals to fulfill the full-time hours requirement through one or more jobs. The ESP program is aimed at speeding up the re-employment of long-term and

repeat EI users. Operating in seven provinces, the ESP offers temporary earnings supplements for individuals who move off El and take on full-time jobs. Unlike the SSP, however, individuals cannot piece together multiple part-time jobs to qualify for the ESP. (Bloom et al. 1997; Lin et al. 1998).

Child-care policy is also of crucial importance in the welfare-to-work transition. In the United States, recent federal welfare reform transferred much of the policy making to the states. Most states are pushing immediate employment to reduce welfare rolls and child-care subsidies are most gener- ous for those working full-time hours. However, even those remaining on the rolls are encouraged to work, and these recipients tend to be most deficient in work-related skills, relegating them to unstable

part-time work, often with multiple part-time em-

ployers. In Canada, current changes to the federal National Child Benefit program will eliminate the working-income supplement for low-income workers and extend the maximum child benefit level to families with zero income. The provinces, in turn, will reduce social assistance payments to non-working families and redirect these funds to deliver earnings supplements and to extend in-kind benefits to the working poor (Government of Canada 1997). In this regard, provincial governments are expected to





increase subsidies to child care. As workers, in particular women, strive to juggle not only family and work but also to juggle multiple jobs (or acquire other forms of non-standard work), it is important that any new child-care policy address the issue of

flexibility as well as cost.

Finally, we make the case for pro-rated benefits for workers with benefits levels calculated as a func- tion of work hours. Alternatively, we suggest the

possibility of moving toward a system that accom- modates person-related benefits versus job-related benefits. In essence, this would involve pooling risk

among a particular association of individuals as

opposed to those working under specific conditions in a given enterprise. For example, with regard to

supplementary medical and dental benefits, one so- lution would be to move toward a voucher system. This would involve the issuance of pro-rated vouchers by firms to their employees which, in turn, could be redeemed by employees with insurance brokers/

companies. The system could be tailored so as to allow individuals to combine multiple pro-rated vouchers to secure full coverage. Such a system would also allow individuals to purchase insurance

plans which are better suited to their own preferences (i.e., choice with respect to the amount of deductible, percentage of coverage, etc.) where the individual pays the differential in cost for more com-

prehensive plans. In addition, when individuals

change jobs, they could keep the same plan and sim-

ply substitute employer vouchers.

NOTES

The authors thank the participants at the "Changes in Work- ing Time in Canada and the United States" Conference in Ottawa, June 1996, where an earlier version of this paper was presented, and two anonymous referees for their com- ments and suggestions. The authors would like to thank Rebecca Jacobs, Wei-Jang Huang and Derrick Hynes for superb research assistance, and Claire Vogelsong and Sandra Schippers for secretarial support. We are grateful to the Household Surveys Division at Statistics Canada for kindly providing us with Labour Force Survey data.

Financial support from the Queen's University Advisory Research Council is gratefully acknowledged.

'In the United States, the May CPS supplement was modified substantially after 1991, eliminating much of the detail on multiple-jobs. While brief undetailed questions were added to the main monthly CPS survey to off- set this modification, detailed information such as that provided in our tables is unavailable after 1991.

2A discussion concerning the statistical significance of these moonlighting trend comparisons is given in the Appendix.

3While unemployment rates were fairly comparable for the two countries in 1980 (7.5 percent in Canada versus 7.1 percent in the US), in September 1998 the rates stood at 8.3 percent in Canada and 4.6 percent in the US (1997 Statistical Abstract of the United States, US Bu- reau of Labor Statistics Labor Force Statistics from the Current Population Survey, and The Economics Editor, Brown Economic Assessments Inc.)

4The "other" marital status group includes those who are divorced, separated, or widowed.

5These numbers do not appear in the tables because the definition for self-employed differs across the two countries, so the numbers are not completely comparable.

6Primary occupations include forestry, fishing, farm- ing, and mining.

7For both countries, there are substantial differences in moonlighting rates across regions, in both static and trend comparisons. A separate paper is warranted for a detailed discussion of these differences.

8For example, a mother may work one part-time job while her children are in school, and work a second part- time job in the evenings or weekend, during times that an- other family member is available to care for the children.

9The data sources for these reported reasons are the same as for the sources for the previous tables except for the 1997 US numbers. These data were pulled from an unpublished US Bureau of Labor Statistics table made available by John Stinson.

1One could use the duration of the moonlighting epi- sode to provide information concerning the motive for moonlighting, but we have no information on duration.





I I In 1994, nearly 50 percent of preschool children with

working mothers were cared for by relatives (Casper 1997).

REFERENCES

Abbott, M.G. and C.M. Beach (1997), "The Impact of

Employer Payroll Taxes on Employment and Wages: Evidence for Canada, 1970-1993," in Transition and Structural Change in the North American Labour Market, ed. M.G. Abbott, C.M. Beach and R.P.

Chaykowski (Kingston: IRC Press, Queen's University). Advisory Council on Unemployment Compensation

(1996), "Defining Federal and State Roles in Unem-

ployment Insurance," (Washington, DC: ACUC). Averett, S. (1996), "Moonlighting: Multiple Motives and

Gender Differences," Mimeo. Beach, C.M. and G.A. Slotsve (1996), Are We Becoming

Two Societies? (Toronto: C.D. Howe Institute). Blank, R.M. and M.J. Hanratty (1993), "Responding to

Need: A Comparison of Social Safety Nets in Canada and the United States," in Small Differences That Mat- ter, ed. D. Card and R.B. Freeman.

Bloom, H., B. Fink, S. Lui-Gurr, W. Bancroft and D. Tattrie (1997), "Implementing the Earnings Supplement Project: A Test of a Re-employment Incentive," (Ottawa: So- cial Research and Demonstration Corporation).

Brown Economic Assessments Inc. (1998), The Econom- ics Editor, Issue Number Thirty-Seven, Selective Ac- cess [Online]. Available: http://www.brownecon.com/ ee_oct98.html# consumer._price_index [October].

Canada. Department of Finance (1997), Budget 97: Work-

ing Together Towards a National Child Benefit Sys- tem (Ottawa: Government Services Canada).

Canada. Human Resources Development Canada (HRDC) (1994), Report of the Advisory Group on Working Time and the Distribution of Work (Ottawa: Supply and Services Canada).

(1995), "The Future of Work: Trends in the Chang- ing Nature of Employment," (Ottawa: Applied Re- search Branch Strategic Policy Group, HRDC).

Card, D. and R.B. Freeman (1993), Small Differences That Matter: Labor Markets and Income Maintenance in Canada and the United States (Chicago: University of Chicago Press).

Card, D. and W.C. Riddell (1993), "A Comparative Analysis of Unemployment in Canada and the United States" in Small Differences That Matter, ed. D. Card and R.B. Freeman.

Casper, L.M. (1997), "Who's Minding Our Preschoolers?" Fall 1994 (Update), U.S. Bureau of the Census, Cur- rent Population Reports, Household Economic Stud- ies, P70-62 (November).

Cohen, G.L. (1994), "Ever More Moonlighters," Perspec- tives on Labour and Income (Autumn): 31-38.

Conway, K.S. and J. Kimmel (1998), "Male Labor Sup- ply Estimates and the Decision to Moonlight," Labour Economics 5(2):135-66.

Council of Economic Advisors (1996), Job Creation and

Employment Opportunities: The United States Labor Market, 1993-1996, A Report by the Council of Eco- nomic Advisers with the U.S. Department of Labor, Office of the Chief Economist (Washington, DC: US

Department of Labor). Courchene, T.J. (1994), Social Canada in the Millennium

(Toronto: C.D. Howe Institute). Danziger, S. and P. Gottschalk (1995), America Unequal

(Cambridge, MA: Harvard University Press). Economic Council of Canada (1990), Good Jobs, Bad

Jobs: Employment in the Service Economy (Ottawa: Supply and Services Canada).

Freeman, R.B. and K. Needels (1993), "Skill Differen- tials in Canada in an Era of Rising Labor Market In-

equality," in Small Differences that Matter, ed. D. Card and R.B. Freeman.

Gallaway, L. (1995), "Public Policy and Part-Time Em-

ployment," Journal of Labor Research 16 (3):306-14. Government of Canada (1995), A 21s' Century Employ-

ment System for Canada (Ottawa: Supply and Serv- ices Canada).

Horrigan, M.W. (1987), "Time Spent Unemployed: A New Look at Data from the CPS," Monthly Labor Review 110(7):3-15.

Houseman, S. (1995), "Job Growth and the Quality of Jobs in the U.S. Economy," Labour (Special Issue):S93-124.

Hum, D. and W. Simpson (1996), Maintaining a Com- petitive Workforce: Employee-Based Training in the Canadian Economy (Montreal: The Institute for Re- search on Public Policy).

Krahn, H. (1995), "Non-Standard Work on the Rise," Per-

spectives on Labour and Income (Winter):35-42. Kuhn, P. (1996), "Labour Market Polarization: Canada

in International Perspective," in Policy Frameworks for a Knowledge Economy, ed. T.J. Courchene, Bell Canada Papers Vol. 4 (Kingston: John Deutsch Insti- tute for the Study of Economic Policy, Queen's Uni- versity).





Layard, R., S. Nickell and R. Jackman (1991), Unemploy- ment: Macroeconomic Performance and the Labour Market (Oxford: Oxford University Press).

Lemieux, T. (1993), "Unions and Wage Inequality in Canada and the United States," in Small Differences That Matter, ed. D. Card and R.B. Freeman.

Levenson, A. (1995), "Where Have All the Part-Timers Gone? Recent Trends and New Evidence on Moon-

lighting," Working Paper 951 (Santa Monica, CA: Milken Institute for Job and Capital Formation).

Lilja, R. (1991), "The Problematic and Unproblematic Second Job," Discussion Paper No. 107 (Helsinki: Labour Institute for Economic Research).

Lin, W., P.K. Robins, D. Card, K. Harknett, and S. Lui- Gurr (1998), When Financial Incentives Encourage Work: Complete 18-Month Findings from the Self-Suf-

ficiency Project (Ottawa: Social Research and Dem- onstration Corporation, sponsored by Human Re- sources Development Canada).

Lipsett, B. and M. Reesor (1997), "Flexible Work Ar-

rangements," Research Paper (Ottawa: Human Re- sources Development Canada).

Mishel, L. and J. Bernstein (1994), The State of Working America, 1994-95 (Armonk, NY: M.E. Sharpe and Economic Policy Institute).

Nakamura, A. (1995), "Nonstandard Work: Definitions, Trends, and Policy Relevance," report prepared for the Edmonton Social Planning Council.

Pold, H. (1994), "Jobs! Jobs! Jobs!" Perspectives on La- bour and Income (Autumn): 14-17.

(1995), "Families and Moonlighting," Perspectives on Labour and Income (Summer): 7-8.

Powell, L.M. (1997a), "Family Behaviour and Child Care Costs: Policy Implications," Policy Options 18(1):11-15.

(1997b), "The Determinants of Multiple Job Hold-

ing: Evidence from Canada," paper presented at the Canadian Economic Association Meetings, St. John's, Newfoundland.

Riddell, C.W. (1993), "Unionization in Canada and the United States: A Tale of Two Countries," in Small Dif- ferences That Matter, ed. D. Card and R.B. Freeman.

Sekscenski, E.S. (1980), "Women's Share of Moonlight- ing Nearly Doubles During 1969-79," Monthly Labour Review 103(5):36-39.

Sharpe, A. and R. Haddow (1997), "Labour Market Cor-

poratism and Labour Market Performance," in Social

Partnerships for Training, ed. A. Sharpe and R. Haddow (Kingston: School of Policy Studies, Queen's University).

Snider, S. (1995), "Characteristics of the Part-Time Work Force and Part-Time Employee Participation in Health and Pension Benefits," Journal of Labor Research 16

(3):239-48. Stinson, J.F. (1986), "Moonlighting by Women Jumped

to Record Highs," Monthly Labour Review 109(11 ):22- 25.

_ (1990), "Multiple Jobholding Up Sharply in the

Eighties," Monthly Labour Review 113(7): 3-10. Sunter, D. and R. Morissette (1994), "The Hours People

Work," Perspectives on Labour and Income (Au- tumn):8-13.

United States Bureau of Labor Statistics (1996), unpublished data made available by John F. Stinson, Jr.

__ (1998), Labor Force Statistics from the Current

Population Survey, Selective Access [Online]. Avail- able: http://stats.bls.gov:80/sahome. html [18 Novem-

ber]. United States Department of Commerce, Bureau of the

Census (1996), 1996 Statistical Abstract of the United States, 116th ed. (Washington, DC: Government Print-

ing Office). _._ (1997), 1997 Statistical Abstract of the United

States, 117th ed. (Washington, DC: Government Print-

ing Office). Webber, M. (1989), "Moonlighters," Perspectives on La-

bour and Income (Winter):21-30.





APPENDIX

We conducted statistical significance tests (t-tests) to determine the significance of all changes in moon-

lighting rates noted in Tables 1, 2, and 3. For the United States, all but the following individuals exhibited

statistically significant changes (at the 5-percent level) between 1981 and 1991:

Whole Sample: Industry: Finance

Male Sample: Industry: Finance

For Canada, all changes between 1981 and 1995 also were statistically significant at the 5-percent level

except for:

Whole Sample: Education: 0-8 years Industry: other primary

Male Sample: Education: 0-8 years; university Students: full-time; part-time Industry: other primary; public administration

Female Sample: Education: 0-8 years Industry: other primary; construction; finance




Documents

Moonlighting Trends and Related Policy Issues in Canada and the United States