The Impact of Self-employment on Productivity growth in Canada and the US

by J. Baldwin* and J. Chowhan

Introduction

Over the 1980s and 1990s, demographic trends and the business cycle have led to increasing self-employment.¹ Most studies have focused on the causes of this phenomenon, rather than its impact on the economy. This paper examines the effect of the self-employed on labour productivity growth in the business sector in the 1990s. A comparison is also made of the relative performance of productivity in Canada and the US.

The importance of the self-employed for productivity originates in its rapid growth. From 1990 to 1998, over half (55%) of our total net job growth of 8.1% (1,056,000 jobs) came from self-employment. The self-employed run different types of businesses. Some generate a large amount of GDP, have employees and use capital in the production process. Others generate little in the way of GDP, hire no employees and use no capital except for perhaps a computer and a room in their house. The impact of the growth of self-employment depends on its distribution across these various types of businesses.

The increase in the self-employed in the 1990s was driven by those who have no employees. This group earned substantially less than both the other self-employed and paid workers. The growth in self-employment in this low productivity segment during the early 1990s was an important factor contributing to the stagnation of our overall productivity.

Statistics Canada’s labour productivity program provides estimates of real output per unit of labour input (hours worked) for the aggregate business sector and its component industries. This data is broad in scope, and is often used to examine efficiency across industries. This paper decomposes the aggregate business sector in a different way. It takes total GDP and hours worked, and then removes the income of the self-employed from the former and the hours worked by the self-employed from the latter. We ask what would have happened to labour productivity if self-employment had not grown so rapidly and if its earnings had not fallen behind those of paid workers.

Finally, a comparison is made with the US to focus on why relative productivity growth in Canada declined from 1987 to 1998 (comparable points on the business cycle).²

The Growth of Self-employment

There are two main classes of workers considered in the productivity accounts: paid workers and unincorporated self-employed. Within the latter, there are self-employed with paid workers, self-employed with no paid workers (own-account) and unpaid family help. During the 1990s, several changes occurred in the composition of jobs. First, the growth in self-employment substantially outpaced that of paid workers (Figure 1). Second, the self-employed without employees grew more quickly than the self-employed who hire other workers. These own-account workers grew by 65.4% from 1987 to 1998, and accounted for almost 80% of all the self-employed by 1998. This contrasts with a drop in the self-employed with employees, as well as unpaid family workers.

Figure 1

While the number of self-employed were growing in the 1990s, their earnings were falling behind paid workers. More importantly, the own-account self-employed earned less than most other workers. Their earnings ranged between 68% and 53% of paid workers, with the largest differences appearing in the late 1990s (Figure 2). Further, earnings of the own-account self-employed averaged only about half that of the self-employed who had employees. The growth of own-account self-employed, who did not support other workers and who received a net income that was low and falling relative to other workers, had a negative impact on productivity.

Figure 2

Labour Productivity Growth

Labour productivity is calculated as the ratio of output to labour input. The measure of output used here is nominal GDP at basic prices for the business sector. The business sector excludes all non-business activities as well as the implicit rental value of owner-occupied dwellings (including institutions such as government, education, hospitals and religious organisations). Labour input is captured by hours worked, whether paid or not. The number of hours worked is the annual average of all jobs multiplied by their average hours worked.

In what follows, we examine the impact of the self-employed on productivity by recalculating the numerator and the denominator of output per hour-worked. In the numerator, we remove from GDP the net earnings accruing to the unincorporated business sector. In the denominator, we remove from total hours the hours worked of the unincorporated self-employed.

In doing so, two distinct sectors are created: the Self-employed (whose sub-components are non-farm and farm), and the remainder of the commercial sector, referred to here as the Business-sector Apart from Self-Employed (the BASE sector). The BASE sector consists mainly of the incorporated portion of the economy, where most paid employees work. The opposite is the self-employed sector, which consists mainly of unincorporated entrepreneurs working on their own.

It should be noted that we cannot remove all of the GDP created by the self-employed. Therefore, the incorporated/unincorporated dichotomy is not perfect, since the wages and salaries and hours worked of paid workers in the unincorporated sector are included in the BASE sector in our calculations. But this does not seriously affect the conclusions of this paper: since almost all of the increase in the self-employed came from those who had no employees, their net income probably captures almost all of their contribution to GDP. When corrections are made to bring net income per self-employed worker closer to GDP per worker, the results are unchanged.

From 1987 to 1998, business sector labour productivity (GDP per hour worked) steadily increased from a low of $21.79 in 1987 to $30.48 per hour in 1998 (Table 1). This was lower than in the BASE sector (where productivity was over over 10% higher at $33.81 in 1998). Business sector productivity is lower than it otherwise would have been because of the low hourly earnings of the self-employed in both the non-farm and farm sub-sectors. The downward pressure on GDP per hour worked intensifies as a result of the increasing share of the self-employed in hours worked and a decline in their relative earnings.

Table 1: Canadian Nominal Output per Hour Worked ($)

The cumulative growth in productivity in the self-employed non-farm and farm sectors was below the growth in the BASE sector (Table 2 and Figure 3). Growth over the entire period for the BASE sector was 42.8%, significantly outpacing the non-farm and farm self-employed sectors which grew 13.1% and 13.5%, respectively. The low increases for the self-employed meant that the cumulative 40% productivity growth in the business sector was lower than in the BASE sector.

Figure 3

These aggregate trends also are present across those industries that account for the largest share of self-employed income. Seven industries contribute almost 85% of total self-employed income: Agriculture, Construction, Transportation, Retail trade, Finance and real estate, Business services, and Health and social services. Most of these industries followed the trend toward relatively lower growth in self-employed output per hour worked through the 1990s.

Table 2: Cumulative Growth of Nominal Output per Hour Worked, 1987-1998

Canada-US Comparison

The growth of the self-employed in Canada and the US diverged in the 1990s. Hours worked by the self-employed in Canada increased from 13.4% to 15.3% of the business sector, while in the US it fell from 12.5 to 10.9%. However, in both countries the share of GDP from the self-employed edged up as a proportion of the business sector, from 7.0 to 7.8% in the US from 1987 to 1998 and from 5.8 to 6.0 % in Canada (Figure 4).

It is noteworthy that by 1998 self-employment in Canada accounted for 4.4 percentage points more of total hours worked than in the US (Figure 4), but their share of GDP was 1.9% lower. Table 3 shows how this affected comparisons of productivity growth. Self-employment in the US had a significant positive effect on its in nominal output per hour worked. Total business sector growth is higher than in the BASE sector, as the self-employed raised cumulative US business sector productivity growth by 4.1%; in Canada, self-employment exerted a 2.9% drag on nominal output per hour worked.

Figure 4

Table 3: Nominal Output per Hour Worked (1987=100)

Recession in the early 1990s stalled hourly earnings of the self-employed in both countries (Figure 5). However, earnings after 1991 took quite a different course in the two countries. In the US, self-employed hourly earnings recovered and resumed its upward trend, whereas in Canada it continued to stagnate (a slight bounce in 1995 and 1996 was reversed in 1997).

Figure 5

The difference in labour productivity growth between Canada and the US is mainly driven by self-employment, especially the non-farm
self-employed. Table 4 presents the differences between the cumulative growth of nominal output per hour worked: Table 5 the growth of constant dollar output per hour worked. A positive value implies that the US had higher growth than Canada.

Table 4: Difference between US and Canada in the Cumulative Growth Of Nominal Output per Hour Worked

The difference between the US and Canada in productivity growth in the non-farm self-employed sector totalled 96% over the period (Table 4). This is the main determinant of the upward pull self-employment had on nominal output per hour worked in the US. The cumulative 16 percentage point difference in aggregate business sector productivity is almost halved to only 9 points when the effect of the self-employed is removed.

The result is even more striking once nominal GDP is converted to constant dollars (Table 5). The 7 percentage point difference between the cumulative growth of deflated output per hour worked falls to less than 2%. In other words, Canada’s real labour productivity growth shortfall relative to the US becomes insignificant once the effects of the self-employed are removed from the business sector.

Table 5: Difference between US and Canada in the Cumulative Growth Of Real Output per Hour Worked

Conclusion

Two characteristics of the self-employed have contributed to the growing gap between Canadian and US labour productivity. First, net income per self-employed worker relative to paid workers was much lower in Canada than in the US and the share of self-employment in total employment was larger and rose in Canada relative to the US. Second, there was much weaker growth in self-employment net income in Canada over the 1990s. Together, these two factors resulted in downward pressure on our productivity growth.

The results indicate that there are fundamental structural differences between the self-employed sectors in the two countries. Net income of the non-farm self-employed boosted nominal output per worker in the US, whereas it lowered it in Canada. When self-employed income was removed from the business sector, the gap in the growth of labour productivity between the US and Canada virtually disappears.

References

Baldwin, John and James Chowhan (2003). “The Impact of Self-employment on Labour Productivity Growth: A Canada and United States Comparison”. Economic Analysis Research Paper Services. Statistics Canada Catalogue (11F0027MIE, no. 016).

Kuhn, Peter and Herb Schuetze (2001). “Self-employment Dynamics and Self-employment Trends: A Study of Canadian Men and Women, 1982-1998.” Canadian Journal of Economics. 34(3):760-784.

Lin, Zhengxi, Janice Yates, and Garnett Picot (1999). “Rising Self-employment in the Midst of High Unemployment: An Empirical Analysis of Recent development in Canada.” Research Paper Series: Analytical Studies Branch. Statistics Canada Catalogue No. 11F0019MPE no.133.

Statistics Canada (1997). Labour Force Update: The Self-employed. Catalogue no. 71-005-XPB, Autumn 1997 Vol. 1 No. 3.

Statistics Canada (2001). Productivity Growth in Canada. Catalogue no. 15-204-XPE.

Recent feature articles

Notes

* Director, Micro-Economic Analysis (613) 951-8588.

¹ Several studies document the extent of transitions into and out of self-employment [Kuhn and Schuetze, 2001]; the differences in the role of self-employment between Canada and the US [Lin, Yates, and Picot, 1999]; and how the characteristics of individual entrepreneurs and previous labour market experience are related to the likelihood of entry and exit into self-employment [Labour Force Update, 1997; Lin, Picot, and Yates, 1999].
² More detailed data are provided in Baldwin and Chowhan (2003).