3.3 Estructura de la màquina
3.3.1 Calculo y analisis de esfuerzos sobre la estructura
Compared to employed people, unemployed people suffer a double disadvantage in relation to business start-up finance: not only do they tend to have fewer savings, but they also tend to have greater difficulties in accessing commercial loans (Metcalf and Benson, 2000). In this line, self-employment policies based on positive discrimination in favour of unemployed would be justified. However, European self-employment programs for the unemployed seems to be a re- sult of i) having faced high and persistent unemployment rates for two decades, and ii) the belief that higher rates of self-employment contribute to innovation and economic growth. According to this argument, the justification of these programmes is still unclear. In other words, there are still no strong reasons in the literature to believe in general that self-employment is preferable to conventional employment through subsidies74, as an alternative to unemployment.75 An obvious
risk of these measures is that they can distort occupational choice, by encouraging non-skilled individuals to enter self-employment in the short term. However, these individuals may return to unemployment when economic conditions change or even, when incentives disappear.76 Thus,
new studies in the literature defend the idea that some unemployment stimulation programmes may actually be harmful (Van Stel and Storey, 2004).
Consequently, the existent relation between unemployment and self-employment is still a topic of controversy and interest in the literature.77 During this exercise, our interest lies in tran-
sitions from unemployment to self-employment. Hence, the subsample is selected of individuals who are unemployed during a particular period and either continue as unemployed or switch into self-employment during the following period. After filtering, the final sample used for es- timation has 11,393 observations where 666 (5.8 percent) of them refer to transitions.78
73 See Table A9 (Appendix A). France, Luxembourg and Sweden are excluded from this specification.
These results must be cautiously interpreted, taking into account the distribution of observations across countries for our exercises (see Tables A5 and A6 –Appendix A-).
74 Employment subsidies are active labour market programmes which help unemployed workers to be
hired at an initially reduced cost for the employer.
75 As exception, Carling and Gustafson (1999) suggest the increase in the supply of skilled unemployed
workers as one of the justifications for the Swedish case. Thus, they find that the risk of re- unemployment is more than twice as high for the subsidized employment programme compared with the self-employment programme.
76 In line with this argument of distortion of occupational choice, Carrasco (1999) reports that the prob-
abilities of entering self-employment from unemployment exceed that probability for those paid- employees. However, she also observes that those individuals with unemployment experience in the past are more likely to exit self-employment, precisely into unemployment.
77 See Reize (2004), and Audretsch et al. (2005).
78 This refers to the main exercise performed. However, a complementary one is also estimated where we
control for individuals having born abroad and for those receiving inherit, gift or lottery winnings. However, the inclusion of these variables means that Germany and The Netherlands and the UK are ex- cluded from this complementary analysis. Tables A5 and A6 (Appendix A) present the distribution of
124 José María Millán
The first two columns of Tables A3 and A4 (see Appendix A) show the logit estimation re- sults of the probability of transition into self-employment, conditional on being unemployed. In the last four columns, multinomial estimates are reported in order to account for the differences in the transition probabilities into self-employment (both for own-account worker and em- ployer), using the same explanatory variables as in the previous exercise.79 As commented
above, Tables A7 and A8 (see Appendix A) also compare predicted probabilities for individuals with given characteristics of entering into self-employment from paid employment and from unemployment. Finally, predicted probabilities of transitions for individuals living across dif- ferent countries are presented in Table A9 (see Appendix A).
Thus, consistent with that obtained by Carrasco (1999), our results show that the probability of switching from unemployment to self-employment is 13.3%, which is more than 5 times the probability of switching from paid-employment (2.4%).80
One of the most interesting results refers to the variable unemployment benefits. The fact that receiving unemployment benefits decreases the probability of entering employment (self- employment included) is well known.81 This variable is strongly significant which seems to jus-
tify recent efforts aimed at allowing benefits to be used in an entrepreneurial adventure (subsi- dies capitalization). However, receiving benefits seems to be less harmful for those becoming employers –precisely those a priori facing higher needs of capital investment- than those switching to own-account work.82 In this line, and also related with the possible existence of li-
quidity constraints, those unemployed individuals receiving capital and property incomes are more likely to switch to employer while the transitions to own-account work are not signifi- cantly affected by this variable.83 On the other hand, initial assets, measured by the proxies
home ownership and inherit, gift or lottery winnings, seem to increase the transitions to own- account work and not transitions to self-employment with employees.84 In short, despite the dif-
ferent strength of variables measuring wealth, our results are consistent with the existence of li- quidity constraints.
With respect to the effect of the endowments of human capital, the effect of age is consistent with that obtained for paid workers entering into self employment. Thus, the overall effect of age is nonlinear, i.e., there is a concave relationship between age and the probabilities of enter- ing self-employment from unemployment (the turning point is reached when individuals are 36.7 years old). Other interesting result is related with education variables. In particular, there is
observations across countries for these main and complementary exercises. Tables B3 and B4 (Appen- dix B) summarizes the mean values of all self-employment entrants from unemployment.
79 Obviously, variables such as the presence or absence of unemployment benefits, or the unemployment
duration are now included and all those variables referred to the previous employment are excluded.
80 See Table A7 (Appendix A).
81 The effect of unemployment benefits on labour market transitions has been widely explored. Atkinson
and Micklewright (1991) surveyed the literature. There are also several studies for Spain, such as Alba- Ramírez and Freeman (1990), Alba-Ramírez (1999), or Bover et al. (2002). With respect to its effects over the probability of self-employment, Carrasco (1999) studys the transitions from unemployment to employment, by distinguishing self-employment from paid-employment as final states. Thus, she ob- serves how the probabilities to switch decreases when unemployment benefit is being received for both destinations.
82 Transitions to own-account work decreases by 51% when unemployed receive benefits while transi-
tions to employer decreases by 34% in the same situation (see Table A7, Appendix A).
83 Each additional €1,000 in this concept increase the probabilities of becoming employer by 5.8% (see
Table A7, Appendix A).
84 The presence of inherit, gift or lottery winnings within the household, increases the probability of be-
Chapter 5. The Emergence of New Entrepreneurs in Europe 125
evidence that non-educated unemployed individuals have a lower probability of entering self- employment. Furthermore, this result is also significant when the final state is disaggregated in terms of employers and own-account workers.85 Focusing now on the presence of previous
spells of self-employment, the unemployed are more likely to become self-employed when they have been in self-employment in the past, which is a new sign about the absence of the “stigma of failure”.86 Finally, past spells as paid-employees reduce the probabilities of entering own-
account work, while does not alter transitions to employer in a significant way.87
The dummies describing gender differences and personal characteristics indicate that males88,
married individuals and those with lower number of children under fourteen are more likely to switch into self-employment.89 Foreigners are also seen to be as likely to enter self-employment
from unemployment as those not born abroad.90 When we try to capture the effect of intergen-
erational transfers of human capital and entrepreneurial ability, there is further evidence sup- porting that the presence of self-employed relatives increases the probability of transition.91
Another interesting result we observe is that unemployment duration decreases the probabil- ity of becoming self-employed.92 With respect to the effect of the aggregated conditions, no evi-
dence was found regarding the relationship between business cycle and this sort of transitions.
Finally, in regards to country specific effects, and again using Spain as the reference, Portu- gal is seen to be the country where the transition from unemployment to self-employment is more likely while lower probabilities were found across The Netherlands and Italy. However, when the distinction was between own-account workers and employers as final states, Spain is
85 However, the probability of switching to employer increases 128% while the probability of switching to
own-account worker increases 72% when the individual presents university studies (see Table A7, Ap- pendix A).
86 The probability of switching to employer for those who were employers in the past increases 332%
while previous spells of own-account work increases this probability to around 118%. Moreover, those unemployed with previous experience as own-account workers are 78% more likely to switch to self- employment without employees again, while if past experience within self-employment was as em- ployer, the probabilities of entering own-account work increases by 47% (see Table A7, Appendix A).
87 This result might be related with the negative effect of unemployment benefits over the probabilities of
entering self-employment. As we commented before, the key question is: Is it due to a disincentive ef- fect of benefits or to their role as a proxy for the attachment of the individual to the wage labour mar- ket? Our result does not necessarily support the second hypothesis, but does confirm that previous paid- employment experiences, decreases the transitions to self-employment (see Table A7, Appendix A).
88 The probability of switching to self-employment from unemployment decreases by 71.5% for females
(see Table A8, Appendix A).
89 However, the number of children under fourteen only marginally reduces the transitions to self-
employment.
90 The effect of whether or not an individual has been can be observed in Table A4 (Appendix A).
91 The probability of switching to self-employment increases by 69% increase when there are relatives
working as employer, while the same probability increases by 34% with the presence of own-account worker relatives (see Table A7, Appendix A). Furthermore, transitions to own-account work are sig- nificantly affected by the presence of relatives working either as employer or own-account worker. Thus, transitions to own-account work increases by 62.5 and 49% when relatives work as employer or own-account worker respectively (see Table A7, Appendix A). However, just relatives working as em- ployers –and not as own-account workers- significantly affect transitions to employer. The probability of switching to employer increases by 79% when there are relatives working as employer (see Table A7, Appendix A).
92 Transitions to self-employment decreases by almost 41% when unemployment duration is about 11
years, compared with those whose unemployment duration is about 3 years (see Table A7, Appendix A).
126 José María Millán
the country where transitions to employer are less likely, while transitions to own-account work are more common within Denmark, Portugal and Spain.93