Can Unemployment Insurance Spur Entrepreneurial Activity? Evidence from France

• Author: DAVID SRAER,JOHAN HOMBERT,ANTOINETTE SCHOAR,DAVID THESMAR
• Date: 01 June 2020
• Published date: 01 June 2020
• DOI: http://doi.org/10.1111/jofi.12880

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 3 •JUNE 2020

Can Unemployment Insurance Spur

Entrepreneurial Activity? Evidence from France

JOHAN HOMBERT, ANTOINETTE SCHOAR, DAVID SRAER,

and DAVID THESMAR∗

ABSTRACT

We evaluate the effect of downside insurance on self-employment. Weexploit a large-

scale reform of French unemployment benefits that insured unemployed workers

starting businesses. The reform significantly increased firm creation without decreas-

ing the quality of new entrants. Firms started postreform were initially smaller, but

their employment growth, productivity,and survival rates are similar to those prere-

form. New entrepreneurs’ characteristics and expectations are also similar. Finally,

jobs created by new entrants crowd out employment in incumbent firms almost one-

for-one, but have a higher productivity than incumbents. These results highlight the

benefits of encouraging experimentation by lowering barriers to entry.

The problem with the French is that they have no word for entrepreneur.

Attributed to George W. Bush.

OVER THE LAST TWO DECADES, policymakers and academics alike have em-

braced the idea that reducing barriers to self-employment and entrepreneur-

ship is important for promoting job creation.1The primary focus of empirical

research has been on understanding how such barriers affect the level of en-

trepreneurial activity (i.e., the number of new firms created). Many recent stud-

ies find evidence of significant heterogeneity among potential entrepreneurs’

∗Johan Hombert is with HEC Paris. Antoinette Schoar is with MIT-Sloan, NBER, and CEPR.

David Sraer is with UC Berkeley, NBER, and CEPR. David Thesmar is with MIT-Sloan and

CEPR. This article is the substantially revised version of a paper previously circulated under

the title “Should the Government Make it Safer to Start a Business? Evidence From a French

Reform.” We thank participants at many conferences and seminars for comments and suggestions.

In particular, we are indebted to Ashwini Agrawal, Steve Davis, Guy Laroque, David Matsa, Toby

Moskowitz, Marina Niessner, Elena Simintzi, and Amit Seru for their valuable insights. The data

used in this paper are confidential but not the authors’ exclusive access. We have read The Journal

of Finance disclosure policy and have no conflicts of interest to disclose.

Correspondence: David Thesmar, MIT Sloan School of Management, CEPR, 100 Main Street,

Cambridge, MA 02142; e-mail: thesmar@mit.edu

1See the fast-growing literature on the impact of financial market and regulatory reforms on

entrepreneurship, for example, Bertrand, Schoar, and Thesmar (2007), Cole (2009), Djankov et al.

(2002), and Klapper, Laeven, and Rajan (2006).

DOI: 10.1111/jofi.12880

C2020 the American Finance Association

1247

1248 The Journal of Finance R

ability to grow, risk tolerance, ambition, and even optimism.2In light of such

heterogeneity, the welfare implications of reducing barriers to entrepreneur-

ship become unclear, as they depend crucially on how such policies affect the

quality of new entrepreneurs. On the one hand, if individuals have private

information about their entrepreneurial abilities, lower barriers to entry may

decrease the quality of the average entrepreneur. On the other hand, reduc-

ing start-up costs can lead skilled but risk-averse individuals to enter self-

employment (Kihlstrom and Laffont (1979)) by allowing them to experiment

and learn about their ability (as in Jovanovic (1982), Caves (1998), Manso

(2011)).

In this paper, we empirically investigate how large reductions in the cost

of entry affect selection into entrepreneurship and allocative efficiency more

generally. To do so, we exploit a large-scale reform of the French unemploy-

ment insurance (UI) system that provides downside insurance to unemployed

workers starting a business. To derive testable predictions, we first develop an

equilibrium model of occupational choice that features risk-averse individuals

with heterogeneous entrepreneurial skills. Individuals decide between wage

employment or self-employment. The success of the firm is determined by a

combination of skill (known in advance) and chance. In the model, provid-

ing downside insurance to entrepreneurs increases entry, but the number and

quality of new businesses depend on the dispersion of talent. When talent is

relatively heterogeneous, decreasing downside risk has a modest effect on new

firm creation, reduces entrepreneurial quality, and leads to a smaller reduction

in the size of incumbent firms competing for the same resources. We refer to

this explanation as the selection view. In contrast, when the talent distribution

is more homogeneous, providing downside insurance has a pronounced effect

on new firm creation, a modest effect on entrepreneurial quality, and a signif-

icant crowding-out effect on incumbents. In the latter case, facilitating entry

increases allocative efficiency by allowing relatively skilled but risk-averse indi-

viduals to become entrepreneurs. Importantly,the model requires that workers

cannot engage in risk-sharing contracts with outside financiers (e.g., through

revolving credit or outside equity), so that the provision of downside insur-

ance by the UI system generates real benefits for unemployed workers. In this

sense, our model, and hence our interpretation of the data, is consistent with

the reform completing a missing insurance market (Rampini and Viswanathan

(2010)). While the provision of downside insurance may also relax new firms’

financial constraints, for example, by allowing entrepreneurs to pledge future

unemployment benefits, we reject this alternative interpretation in the data.

We empirically evaluate these trade-offs by exploiting a reform that facil-

itated the transition of unemployed individuals into self-employment—Plan

d’Aide au Retour `al’Emploi(PARE). Prior to the PARE, unemployed workers

starting a business would lose all access to their UI benefits. After the reform,

2See, among others, Haltiwanger, Jarmin, and Miranda (2013), Nanda (2008), Hurst and Pugs-

ley (2011), Landier and Thesmar (2009), Holtz-Eakin, Joulfaian, and Rosen (1994a), and Schoar

(2010).

Can Unemployment Insurance Spur Entrepreneurial Activity? 1249

the UI agency filled the gap between unemployment benefits and realized en-

trepreneurial income, thereby offering temporary protection against potential

losses from entrepreneurship. We leverage firm- and individual-level admin-

istrative data to evaluate how this large-scale reform affected not only firm

creation, but also the characteristics of newly created firms and industry-wide

employment. Our identification strategy relies on the heterogeneity across in-

dustries in “exposure” to the reform. Unemployed individuals are empirically

more likely to start zero-employee firms and/or register as sole proprietorships.

Accordingly, we define exposure to the reform as the fraction of sole proprietors

among all newly created firms in an industry, where we measure this “treat-

ment intensity” in the years preceding the reform.3Our empirical strategy is

akin to a difference-in-difference analysis that compares changes in the num-

ber and characteristics of newly created firms following the adoption of PARE

across industries with different treatment intensity. Our identifying assump-

tion is that absent the reform, the dynamics of business creation would not

have been systematically related to industry treatment intensity.4

Our first empirical analysis establishes that the reform significantly boosted

new business creation by unemployed workers. Relative to the prereform pe-

riod, the registration of new firms in the postreform period is 14 percentage

points higher in industries belonging to the top quartile of treatment intensity,

relative to industries in the bottom quartile. This estimate is robust to a variety

of robustness checks, in particular to alternative definitions of treatment inten-

sity or event window and to controls for industries exposure to business cycle

shocks. We also find support for our causal interpretation of this estimate by

showing that, in the cross section of industries, the postreform increase in new

firm creation strongly correlates with the entry of unemployed entrepreneurs.

We next document that firms created in response to the reform are not of

(observable) worse quality.We first measure quality using ex post outcomes and

show that, relative to the prereform period, there are no significant changes

in failure rates, hiring rates, or growth rates of firms started after the reform

in the most versus least treated industries. Using administrative survey data,

we also measure quality using ex ante characteristics of entrepreneurs such as

education and self-reported growth expectations. We find no significant effect

of the reform on the composition of entrepreneurs’ educational backgrounds

and we find a small, positive effect on subjective growth expectations. Overall,

the evidence supports the experimentation view, whereby providing downside

insurance induces new firm creation without significantly reducing the average

quality of the new entrepreneurs.

3In Internet Appendix III, we show that our analysis is robust to defining treatment intensity as

the fraction of firms with zero employees among newly created firms in an industry. The Internet

Appendix is available in the online version of the article on The Journal of Finance website.

4We explore the validity of this assumption by conducting numerous robustness checks. In

particular, our relatively high-frequency data allow us to include industry-specific trends in the

regression analysis. Including these trends does not affect our main estimates, which provides

support for the parallel trends assumption.