7 Discussion and conclusions.

Self-employment rose sharply in the UK during the 1980s, encouraged by focused government intervention and supported by financial liberalization. The period was also characterized by sustained and rapid economic growth. But did greater self-employment cause heightened growth?

Greater numbers of self-employed workers in an economy should, in theory, increase labor market flexibility in response to demand shocks because there is no binding wage contract on the number of hours worked. In turn, Millard (2000) argues that this should lead to greater output and consumption and lower unemployment. There is no empirical evidence to support such a theoretical proposition. As we noted above, self-employment appears to be uncorrelated with unemployment, although transitions between employees and self-employment are negatively correlated with unemployment, while transitions from unemployment to self-employment are positively correlated. In aggregate these two effects roughly cancel each other out. What about any relationship with output?

Evidence from a series of GDP growth equations for 23 countries over the period 1966-1996 presented in Blanchflower (2000) suggested that a higher self-employment rate does not increase the real growth rate of the economy; in fact there was even some evidence to the contrary. (1) We repeat this analysis here for a longer time period and more countries. Table A.19 examines the relationship between the growth in real GDP and changes in the self-employment rate, using time series data on the 30 OECD countries for the period 1967-2005 (the additional seven countries are Czech Republic, Hungary, Korea, Mexico, Poland, Slovakia, and Switzerland). As in Blanchflower (2000), the regressions should be thought of as a Cobb-Douglas production function, where the change in the numbers of employees over the previous period is included to distinguish the labor input. Capital is assumed to grow linearly and as the model is estimated in changes the effect of capital will be in the constant. Also included in the regressions are a set of country dummies plus a lagged dependent variable. The columns in Table A.19 experiment with different measures of self-employment. Columns 1 and 2 define self-employment as the number of self-employed as a percentage of total employment. Columns 3 and 4 define self-employment as the number of non-agricultural self-employed as a percentage of total non-agricultural employment. (2) These results presume a particular direction of causation, from self-employment to growth and not the reverse. Columns 1 and 3 include the change in the self-employment rate and a lagged GDP term. Columns 2 and 4 add the change in the number of employees. In no case is the change in the self-employment rate significant: experimenting with longer lags produced similar results. The results confirm Blanchflower's (2000) earlier findings, but for a richer data set--we find no evidence that changes in self-employment are correlated with changes in real GDP.

Another measure of economic benefit could be greater happiness. Table A.20 presents evidence on life satisfaction using data from the Eurobarometer trend file of 1970-2002 for 16 European countries in column 1 and for the UK in column 2. The final column uses data from the most recently available (14th) sweep of the British Household Panel Study of 2004/2005. In all three columns the self-employed have significantly higher life satisfaction than do employees. So can a higher self-employment rate lead to greater aggregate happiness? The results are not supportive. We substituted mean life satisfaction scores on a four point scale taken from the World Database of Happiness for 1976-2006 for GDP growth and repeated our previous analysis. The question asked was "How satisfied are you with the life you lead?"--very satisfied; fairly satisfied; not very satisfied; not at all satisfied, where very is coded as 4 down through not at all which is coded as 1. We include GDP growth as an explanatory variable, along with inflation and the unemployment rate and a complete set of country and year dummies. (3) Self-employment fails to provide any incremental information in explaining happiness when these other variables are included. We find evidence that both unemployment and inflation lower happiness, although the effect is greater for unemployment than it is for inflation, as found by Di Tella et al. (2001) and Wolfers (2003). The GDP term, which they did not include, enters significantly positive. It remains uncertain why self-employment enters positively into a micro-level happiness equation, but not into a macro-level equation.

These results, of course, do not mean that higher self-employment is a bad thing. A very high proportion of individuals across a number of surveys express the desire to become self-employed. We certainly find no evidence that more self-employment is bad for the economy. The self-employed seem to especially value their independence. Many governments around the world believe that it is appropriate to try and make their economies more entrepreneurial, but that need not necessarily imply a higher self-employment rate. And it also means that a higher self-employment rate cannot be expected to translate directly into greater economic success. A lower self-employment rate could conceivably be better. Can we imagine a society consisting almost exclusively of wage and salary workers? It would seem unlikely that a self-employment rate around zero would be optimal. Unfortunately, we have no way of knowing what number governments should be aiming for. Market forces need to prevail and Blanchflower's (2004) conclusion stands; more may not be better. Let the market rip.

The probability of being self-employed in the UK is higher for men and rises non-linearly with age. Those with craft qualifications, including trade apprenticeships, are more likely to be self-employed, and rates are higher for those working in the construction or retailing industries. Immigrants are more likely to be self-employed, but there is considerable variation by country of birth. Probabilities are also high in the South West and London as well as in Agriculture and Construction. Occupations with high self-employment rates include Health Professionals; Construction Trades; Hairdressers; Artistic and Sports Occupations and Agricultural Occupations.

In the US, the probabilities are higher the more educated a person is, while the opposite is true in Europe. Financial windfalls or housing capital gains are important explanatory variables for self-employment, through their ability to mitigate liquidity constraints. House price increases appear to be associated with increases in the self-employment rate.

Most self-employed individuals in the UK work alone or in a partnership and do not have any employees. They typically work longer hours than their employed counterparts, but generally earn less. There is, however, no evidence that in aggregate increases in self-employment affect growth in GDP, nor happiness, positively. At the very top end the successful entrepreneur earns considerably more than most wage and salary earners. The entrepreneur has a unique skill--he or she has created a job for him or herself and possibly even a job for others. The entrepreneur is an important engine for growth in the economy.

(1) The 23 countries were Australia, Austria, Belgium, Canada, Denmark, Eire, Finland, France, Germany, Greece, Iceland, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Turkey, UK, and the USA.

(2) A split excluding non-agricultural data is not available for Switzerland.

(3) The results are as follows, with the dependent variable being the mean life satisfaction score in year t. [Satisfaction.sub.t-1] 0.4501 (8.45) GDP annual growth rate 0.0060 (2.77) Inflation rate -0.0010 (0.66) Unemployment rate -0.0053 (2.70) Self-employment rate 0.0001 (0.03) Adjusted [R.sup.2] = 0.9623, N = 344

Includes year and country dummies. Countries are Austria; Belgium; Canada; Denmark; Finland; France; Germany; Greece; Ireland; Italy; Japan; Luxembourg; Netherlands; Portugal; Spain; Sweden; UK and the USA. T-statistics in parentheses.

David G. Blanchflower (1) and Chris Shadforth (2)

(1) Bruce V. Rauner Professor of Economics, Dartmouth College, University of Stirling, NBER, IZA and Member of the Monetary Policy Committee, Bank of England, blanchflower@dartmouth.edu; david.blanchflower@bankofengland.co.uk; www.dartmouth.edu/~blnchflr

(2) External Monetary Policy Committee Unit, Bank of England, chris.shadforth@bankofengland.co.uk