Entrepreneurship and the business cycle: the “Schumpeter” effect versus the “refugee” effect—a French appraisal based on regional data

People who are unemployed choose to create their own job.

In an entrepreneurial society, most new jobs are created by new-firm start-ups.

We notice that an increase in the GDP on a horizon from 2 to 3 years is translated by an increase in the number of new-firm start-ups which can be interpreted as being a demonstration of the opportunities of new-firm start-ups resulting from the growth described by Leibenstein (1968). Leibenstein presents the entrepreneur as a «gap-filler»and an «input completer». There opportunities for profit and arbitrages are essential to the entrepreneurial activity. It then benefits from the growth of the economic activity which offers true perspectives for new economic activities (and new-firms) in the market niches it gives rise to.

http://​www.​deloitte.​com/​assets/​Dcom-France/​Local%20​Assets/​Documents/​publications/​Plaquette_​Fast50_​2013.​pdf.

Young and innovative firms.

The “Pull” effect is defined by Storey (1991, p. 171): “New-firm formation takes place when an individual perceives an opportunity to enter a market to make at least a satisfactory level of profit”.

A descriptive analysis of GDP and unemployment shows that unemployment is strongly countercyclical.

A high unemployment rate reduces the opportunity cost of becoming an entrepreneur and stimulates new-firm start-ups. In a period of growth (measured by the consumption of industrial products), the optimism of the economic agents is translated by favorable anticipations and incites them to engage in new-firm start-ups.

National Institute for Statistics and Economic Studies.

“The Global Entrepreneurship Monitor (GEM) project is an annual assessment of the entrepreneurial activity, aspirations, and attitudes of individuals across a wide range of countries. Initiated in 1999 as a partnership between London Business School and Babson College, the first study covered ten countries, since then nearly 100 ‘National Teams’ from every corner of the globe have participated in the project, which continues to grow annually” http://​www.​gemconsortium.​org/​What-is-GEM.

The X12 procedure (2011), an adaptation of the U.S. Bureau of the Census X-12 Seasonal Adjustment program, seasonally adjusts monthly or, in our case, quarterly time series. The mechanics of seasonal adjustment involve breaking down a series into trend-cycle, seasonal, and irregular components. It is difficult to estimate seasonal effects when the underlying level of the series changes over time. For this reason, the program starts by detrending the series with a crude estimate of the trend-cycle. It then derives crude seasonal factors from the detrended series. It uses these to obtain a better trend-cycle and detrended series from which a more refined seasonal component is obtained. This iterative procedure, involving successive improvements, is used because seasonal effects make it difficult to determine the underlying level of the series required for the first step. Many economic time series are related in a multiplicative fashion.

Industrial specialization is not globally favorable to employs. Yet, in some regions, some sub-industrial sectors record a less bad performance than the national average performance of the considered sub-industrial sector. One may then speak of a relative good performance (relatively to a sub-sector that is nationally in negative growth).

“A spurious regression occurs when a pair of independent series, but with strong temporal properties, are found apparently to be related according to standard inference in an OLS regression” (Granger et al. 1998).

Results are not reported here but are available upon request from the authors.

These results are robust to the choice of \(p\) in the estimated individual ADF (p) regressions for p \(=\) 1, 2, 3 and 4.

Cross-section dependence can arise due to spatial or spillover effects or could be due to unobserved (or unobservable) common factors.

Indeed, based on Monte Carlo experiments, Moon and Perron (2004) state that their tests require at least 20 cross-sectional units for precise estimation of the number of factors where \(T > N\) . Once this is accomplished, the tests show good size properties and are quite powerful when no deterministic component is estimated. Pesaran (2007) shows by simulation that CIPS test has satisfactory size and power even for relatively small values of N and T. Gengenbach et al. (2010) underline that these tests are designed for large \(N \) cross-sectional units and \(T \) time series observations due to the estimation of the common factor(s) either by using principal components or by including the cross-sectional mean as proposed by Pesaran (2007). They investigate small sample properties of the tests by Monte Carlo experiments considering three different values for \(N \) and \(T \) each, namely, 20, 50, and 100 and conclude that the presence of serial correlation leads to size distortions for all statistics when \(T \) is small, which can be quite strong in some cases and even persist for T \(=\) 100.

The filter is used to obtain a smoothed-curve representation of a time series. This trend is more sensitive to long term than to short-term fluctuations. The cycle component, which is used in our study, is the difference between the original series and the trend. The adjustment of the sensitivity of the trend to short-term fluctuations is achieved by modifying a multiplier \(\lambda \) For quarterly data, the value 1,600 is recommended (Zarnowitz and Ozyildirim 2006).

Following Fiorito and Kollintzas (1994), we consider that the two cyclical components are strongly correlated, weakly correlated or not correlated for a shift \(j\) when the correlation coefficient is significant at a 1 % level, 5  %, not significant at 10 %, respectively. If the cross-correlation \(\rho \left( j \right) \) is positive, null or negative then new-firm start-ups are respectively procyclical, acyclical, or countercyclical. Furthermore, if \(\left| {\rho \left( j \right) } \right| \) is significant at a 5 % level for a positive, null, or negative value of \(j\) then the cycle of new-firm start-ups is leading the other cycle by \(j\) periods, is synchronous or is lagging the other cycle by \( j\) periods, respectively.

The survival rate after 3 years of the new companies is about 60 %.

«Système d’informations sur les nouvelles entreprises ». It is a survey of new French firms (new-firms start-ups and takeovers) that occurs every 4 years for a new cohort.

Cyclical correlations between unemployment rates and GDP are available upon request.

The long-run relation is estimated by the Ordinary Least Squares. The OLS estimator is then asymptotically super convergent. So for two cointegrated series, it is necessary that the estimated residuals of the long-term relationship should be stationary. The stationary of residuals is tested by means of the Dickey–Fuller test.

In order to save space, results are not presented but they are available upon request.

One may distinguish the south of Franche-Comté region which is a more entrepeneurial territory thanks to the reconversion of the watch industry into microtechniques and precision equipment.

Results concerning cointegration relationships can be obtained from authors upon request.

For applications where the cross-sectional dimension grows reasonably large, usual time series tests such as the Johansen (1995) procedure are likely to become unfeasible.

There is no universally accepted definition of what constitutes a high-growth firm. One of the definitions of the high-growth firm is that of firms, generally recent, that employ at least 20 employees and for which staff expenditure other than that relating to executives increased by at least 15 % in each of the two previous years. In 2004, the estimation of new gazelle companies in the UK is 4,500 firms per year while in France it is only 3,000 firms (Zimmern 2005).

Results are available upon request.

We refer to the identification problem discussed in Johansen and Katarina (1994).

Results support the constancy of parameters, the cointegrating relationships may be considered stable over the period under study. The recursive eigenvalues test developed by Hansen and Johansen (1999) to check parameter constancy has been performed.

Indeed, we have demonstrated that all the variables are pushing or pulling forces of the system performing a bloc exogeneity test within the cointegration space (Juselius 2006), no variable can be considered weakly exogenous.

Public aid was eligible for enterprises doing R&D or able to collaborate with other firms in the sector or directly with the research laboratories. Consequently, this kind of enterprise is more effective in terms of productivity and exports (Aw et al. 2011).