regional policy counterfactual impact evaluation what it tells us… and what it doesn't daniel...

Regional PolicyRegional Policy

Counterfactual impact evaluation

What it tells us… and what it doesn't

Daniel MouquéDG Regional Policy, European Commission


Reminder:Counterfactual = comparison


In practice, comparison group


This imposes conditions…

• Similar intervention over large "n"

• (law of large numbers)


… which only hold for certain measures

• Interventions which target individuals or enterprises

• Not infrastructures (exception: impact of infrastructures on individuals)

• Perhaps for area based initiatives (provided similar goals/means)


There are also data needs

1. Good data on the intervention (nature, scale, dates)

2. Good data on target indicators (before and after, including for non-beneficiaries)

3. The ability to link 1 and 2


Lessons learned from enterprise support studies

• DG Regional Policy doing & encouraging since 2008

• What are we learning? And what would we like to know? In terms of:1. Investment, capital constraints and other market

failures (and how vary by firm and support size)2. Impact of support on the enterprise (productivity,

innovation, employment)


… about investment, capital constraints and other market failures?

What do we learn…


Impact on investment in Eastern Germany (GEFRA 2010)


E. Germany not an isolated example


Small is beautiful 1 – the firmsScheme Study Finding on large

enterprisesComparison between SME size classes

Investment grants, IT

ASVAPP (2012)

No or negative impact firms >250 employees

Thorough exam: no difference

RSA invest grants, UK

Crisculo et al (2012)

No impact for firms > 150 employees

Impacts slightly higher for firms < 50 employees?

DK Innovation Consortia

CEBR, Denmark (2010)

No impacts for firms > 150 employees

Not examined

Innovation support, DE

Czarnitzki et al (2011)

Small much better, but firm or grant size?

Smaller may do better

Invest support, E. Germany

GEFRA/IAB (2010)

(Did not study large enterprises)

No difference by SME size class


Small is beautiful 2 – the support

• ASVAPP (2012) even controlling for firm size, smaller grants more effective (cpj €79,000 for smallest grants, rising to €489,000 for largest).• ASVAPP (2012) outright grant to SMEs similar effect to soft loan of same size• Czarnitzki et al (2011) presence or absence of a grant was the crucial factor - smallest grants had almost the same innovation impact as the largest • Comparing across studies: schemes of smaller support tended to have better results (eg RSA, UK)


Business advice can be cost effective

• Better survival rates 2-4 years later in North Jutland.

• €7500/net firm €1500/net job

• (Rotger and Gørtz, 2009)


What do we learn?

• Capital rationed for SMEs, but only partially• Grants help – do not substitute private money• This argument applies to small enterprises and

(probably) to medium sized but not large firms• Less support and/or financial instruments would

still work• Capital constraints not the only market failure:

success of advice => information failures more serious, at least for smallest and newest firms?


What would we like to know?

• The mechanism for capital constraints? Knowing this would help for…• Targetting by firm? And what too big for support?• More effective solutions than direct financial support? (E.g.

change capital market)• What is the optimal level and form of support?

• What information failures?• What is good soft support (incl. business advice)?• How to target/tailor by context and firm?

=> Need more CFs and other types of evaluation


… about impacts on the firm? Productivity, innovation & jobs

What do we learn…


Broader more often than deeper

Scheme Study > empl > productivityLaw 488 invest grant, IT

ASVAPP (2012) ++ None

SME support in Piemonte

ASVAPP (2012) ++ For loans, not for grants

Enterprise support NI

Hart & Bonner (2011)

Small but stat. sig.

Small but statistically significant

SME grants, PL in Poland

Trzciński (2011)

++ None

RSA invest grant UK

Criscuolo (2012)

++ Statistically insignificant

DK Innovation Consortia

CEBR (2010) Statistically insignificant

Supported profits grew 12% more over 10 yrs


A closer look at some exceptions

• CEBR (2010) in DK: innovation consortia increased profitability 12% vs controls over a 10 year period (adds up to €260,000 extra profits per firm).

• Czarnitzki (2007): R&D subsidies in Germany had a significant effect on research and innovation where the firm also benefitted from networking

• Czarnitzki (2007): in Finland both financial R&D support and networking effective, and additive


CIS indics, Germany (Czarnitzki, 2011)


But innovation is not a panacea

• GEFRA (2010) investment impact of R&D grants < modernisation grants (leverage 0.9-1.0 vs 1.4-1.5). Innovation benefits worth loss in impact?

• De Blasio, Fantino & Pellegrini (2009) No additional impact from investment scheme: less tangible nature => more possibilities for deadweight


Jobs created, but < monitoring data

Scheme Study Jobs supported

(monitoring)

Jobs created

(from CF)

Investment support, E. Germany

GEFRA/IAB (2010)

107,000 "created",plus 439,000 "safeguarded"

27,000

Law 488 invest support, IT

ASVAPP (2012)

82,000 "gross created"36,000 "net" (beneficiary survey)

12,000

SME invest grants, PL

Trzciński (2011)

25,000 "created" 10,500


Job quality good

• ASVAPP (2012) average firm salary and productivity same or slightly greater

• Trzciński (2011) jobs created in SMEs received similar pay rises to those in the control group – and that jobs were maintained five years after support.


What do we learn?

• Relatively easy to make firms proportionately bigger (e.g. with grants)

• More difficult to make firms more innovative/productive (soft support better?)

• Measures with less tangible targets eg innovation can be abused (maybe we knew this already?)


What is left unanswered? And how would we answer this?

• Is soft support really the key to innovation?• What is the mechanism for productivity and

innovation? What types of innovation influenceable, how to target by firm etc?

• What constitutes a "smart" support package? What soft support, what mix with financial support?

• How to avoid abuse of innovation and networking measures?

=> Need more CFs and other types of evaluation


In conclusion …


In summary

• New lessons from CFs about impacts (partial capital constraints for SMEs, scaling up effect, importance of information failure)

• More to learn about impacts from CFs (e.g. soft support, financial instruments)

• Need other evaluation methods to open "black box" of mechanisms (targetting most effective solutions, best investments)

• Some factors too intangible for quantified approach? (innovation)


• Beware the man whose only tool is a hammer…

• … for every problem comes to resemble a nail

• - Abraham Maslow


For further information

InfoRegio:ec.europa.eu/inforegio

Impact evaluation centre:

http://ec.europa.eu/regional_policy/information/evaluations/guidance_en.cfm#2

regional policy counterfactual impact evaluation what it tells us… and what it doesn't daniel...

Documents

regional policy impact

regional policy small

regional policy reminder

regional policy broader

regional policy lessons

impact of support

innovation support

comparison slide