The Current Population Survey (CPS) is a survey of households, conducted by the U.S. Census Bureau for the U.S. Bureau of Labor Statistics (BLS). It produces data on the employment status of residents, both full- and part-time, and includes data on the labor force, unemployment, and self-employment. At the state level, these labor market data are produced in conjunction with BLS using time-series models that incorporate the CPS data.
Graph 1 shows the growth rates in Connecticut annual nonfarm jobs from the establishment survey, along with resident non-agricultural employment from the household survey broken out by wage and salary employment and self-employment. The growth rates shown are for the first 43 months of recovery/expansion beginning from the month of the trough for each of the four recovery/expansion periods since 1975. Household survey data for Connecticut are incomplete for the 1975 cycle as data on the self-employed are not available. However, data on establishment and resident wage and salary employment is available and, as illustrated in Graph 1, the growth in state jobs and resident employment was in sync over the 1975 cycle, as well as over the 1983 cycle. Though there is no data on self-employment for the 1975 cycle, it is apparent that self-employment outpaced even the strong growth in establishment and resident wage and salary employment over the recovery/expansion that began, for Connecticut, in 1983.
Both resident wage and salary employment and self-employment declined over the 1990s recovery/expansion. However, the pattern over the 1990s was not uniform. There were significant differences in performance during the beginning, the middle, and the end of the decade. To see that, Graph 2 tracks the behavior of Connecticut's self-employment over the state's business cycle from 1982 to 2007. A linear trend is superimposed on the 12-month moving average (MMA) of self-employment in Graph 2, with Connecticut recessions shaded in grey.1 As is apparent, the self-employment cycle is not necessarily in sync with the state's business cycle. Further, the self-employment cycle has a much higher frequency. Since 1980, the state's economy has experienced three complete business cycles (measured trough-to-trough), using establishment employment as the reference series. Over the same period, self-employment has gone through eight complete cycles. The average recovery/expansion of the state business cycle has lasted 1.74 times longer than a recession, but an expansion in self-employment has only lasted 1.39 times as long as a contraction. Though the long-run trend is upward for self-employment, it waxes and wanes around that trend at a high frequency.
There are three periods, in particular, that appear to have produced interesting patterns in the behavior of self-employment: the "Great Recession" (February 1989-December 1992), the mid-to-late 1990s, and the current cycle, which began in 2003. During the "Great Recession," self-employment declined, but after two years into that recession, self-employment turned around and began growing again. This is probably related to the massive restructuring that occurred in the state's economy over this period. This is consistent with research that has found that growth of self-employment in recessions, as opposed to expansions, is more likely to be related to serving as a path out of unemployment in an atmosphere of limited employment opportunities.2 Self-employment growth in expansions is more likely to be related to entrepreneurial, risk-taking behavior. It is an economic environment more conducive to "taking a chance."
During the next period of interest, the mid-1990s, there was a steep decline in self-employment between April 1996 and July 1998. Over this 15-month period, self-employment declined from its peak to its lowest level since the early 1980's. What happened? This is about the time that the late-nineties boom in the economy was underway. Connecticut nonfarm employment grew at a rate of 30,500 new jobs per year between 1996 and 1998, based on the annual average. If, in fact, Connecticut's self-employment growth was due to a lack of job opportunities during, and immediately following, the Great Recession, then the boom in nonfarm employment growth over this period could account for the steep, and rapid, decline in self-employment.
Then the trend reversed between July1998 and March 2000. Over this period, nonfarm employment growth decelerated to a rate of 18,850 jobs per year as the expansion ended and Connecticut entered its next recession in July 2000. Over the current cycle, there has been another round of restructuring in the national and state economies that, once again, has changed the behavior of the growth in self-employment.
What is notable is the strength in the growth of self-employment during each of these periods, and that in the beginning stages of the last recession self-employment behaved in a fashion similar to that of the 1981-83 and 1989-92 recessions. In both 1983 and 1992, the decline in self-employment recovered as the recoveries progressed. However, over the 2003 recovery/expansion, the decline in self-employment was steeper and went through two complete cycles (measured peak-to-peak), and during the last cycle recovery in self-employment was much stronger. In fact, self-employment grew from its last trough of 101,920 in December 2004 to 131,214 at its last peak in November 2006. Despite the anemic growth in both establishment and resident wage and salary employment, self-employment has put in a strong performance, though by February 2007 it had declined from its peak to 120,132.
How does the state's experience compare to the U.S.? Graph 3 presents the answer to that question. Since the U.S. recovery/expansions have historically started before Connecticut's, there is a difference in the "comparable" periods over the two sets of expansions. To put the comparison on a standardized basis, Graph 3 shows the compounded, annual growth rates of U.S. and Connecticut self-employment over their respective recovery/expansions over the post-1975 era. Connecticut outperformed the U.S. in self-employment growth over the recovery/expansion that began in 1982 for the U.S., and in 1983 for Connecticut. Again, Connecticut's self-employment contracted in the 1990s, but when compared to the U.S., the growth in self-employment over Connecticut's current recovery/expansion is three times that of the U.S. on a compounded, annual basis.
Most of the current growth in Connecticut's self-employed has been over the last two years of this recovery/expansion. After contracting by nearly 10%, or 11,000, in 2004, the self-employed grew by 11,000, or 11%, in 2005. In 2006, the growth in the number of self-employed jumped by 18,000, or 16%. This is highlighted in Graph 4, which tracks an index of self-employment growth over the first three years of recovery for the 1983, 1992, and current recovery/expansion (again, no data is available for 1975). The surge in the growth in self-employment can be seen for Year 2 (2005) and Year 3 (2006), where Year 0 is 2003, the year Connecticut's (and the U.S.) recovery in employment began.
However, many of these newly self-employed do not seem to be growing to the point where they are adding employees; that is, the number of businesses, as measured by registrations with the unemployment insurance (UI) system, has had much lower growth rates over the same period. During the boom/bubble years of the late 1990s, Connecticut's net growth in establishments (measured fourth quarter-to-fourth quarter) was averaging 1,100 per year, or 1.09%, compounded annually. Over the three years after the bust (2000 to 2003), annual net growth in new establishments dropped to 221 per year, 0.39%, compounded annually. This improved somewhat between 2003 and 2005, the first two years of Connecticut's employment recovery. Net new establishment formation averaged 795 per year, or 0.75% compounded annual growth, and net new establishment growth increased to 1,023, or nearly 1% in 2005.
However, this falls far short of the growth of 11,000 self-employed in 2005, an 11% increase. That would require the average newly UI registered establishment to create, on average, 11 new jobs. That would seem rather high for establishments with zero employment for the previous year given that seasonal employment has been accounted for by measuring growth from the same quarter, year-to-year. A more likely scenario is that many of these newly self-employed are those who were unemployed due to their jobs being downsized or outsourced. Consulting would be a way out of unemployment, and is consistent with research findings on job-loss and self-employment.3 Anecdotal evidence suggests that, in many instances, those providing technical, professional, and managerial services may be consultants to their former employers. Another explanation lies in the smaller, one-person operation in the construction industry, especially in the skilled trades, and particularly on the residential side where "firms" tend to be smaller and quickly assembled and disassembled at the beginning and end of projects. In fact, industry sectors with large numbers of self-employed, like construction, played a significant role in the 2007 benchmark revisions of Connecticut nonfarm employment. Understandably, the larger numbers of self-employed in some industry sectors, and the varying cycles in which self-employment grows and wanes in relation to the business cycle, presents an added challenge to estimating total payroll employment in those sectors.
In summary, there are four points that come out of this discussion:
(1) Over the long-term, self-employment is growing slowly in Connecticut, but as a percentage of resident wage and salary employment, it has remained fairly constant, except during the early 1990s when, in level and as a percent, it increased, probably due to its serving as a path out of unemployment;
(2) When establishment employment was strong in the late 1990s, self-employment declined steeply, again reinforcing the "path-out-of-unemployment" hypothesis;
(3) The self-employment cycle has a much higher frequency than the state's overall business cycle; and
(4) Connecticut's growth in self-employment over the last two years has been strong, and may be related to the strong growth in industry sectors that typically have large numbers of self-employed, such as administrative and support services, professional and technical services, construction, and health care and social assistance services.
1 State business cycles are defined in terms of the behavior of the nonfarm, or establishment survey, employment series. It is the reference series most frequently used to define sub-national business cycles.
2 See Audretsch, David B., Max C. Keilbach, and Erik E. Lehmann, ENTREPRENEURSHIP AND ECONOMIC GROWTH (2006) Oxford University Press: New York.
3 Again, see Audretsch, David B., Max C. Keilbach, and Erik E. Lehmann.
Return to Top