Product and labour market regulation and patenting

1
Product and labour market regulationand patenting

• Rachel Griffith
• IFS and UCL

2
Motivation

• European growth lags behind US
• Lower investment in innovation in Europe
• Lisbon Agenda and Barcelona target
• increase RD to 3 of GDP
• What (if any) sort of institutional reforms can
  have this sort of impact?
• Product market reforms?
• Employment protection legislation?

3
Motivation

• "In Europe, fundamental changes in employment
  practices, labour markets, and management
  incentives are necessary to encourage rapid
  adoption of new technology that can raise
  productivity ..." (Feldstein, 2001)

4
Motivation

• Product market reforms - conflicting effects
• Reforms to product markets have opened markets,
  increased competition
• Reduction in firms ability to mark-up prices
  above costs
• discourages entry
• (although reforms that reduce entry costs may
  encourage entry)
• may encourage or discourage innovation by
  incumbents

5
Motivation

• Employment protection legislation - conflicting
  effects
• increases adjustment costs, reduces investment
  Cabellero et al (2004)
• increases unemployment, but improves match,
  Marimon and Zilibotti (1999)
• Other labour market institutions
• that affect workers bargaining power may have
  affect investment decisions
• Impact of increased competition may be muted by
  employment protection legislation

6
The impact of product market reforms

• Firms compare profits they would make if they did
  not innovate (so produced at higher cost) with
  those they would make if they did innovate (so
  produced at lower cost)
• There is a cost to innovating, so firms invest in
  innovation up to the point where costs equal
  expected gains

7
The impact of product market reforms

• Assumptions technology
• There is an exogenously determined world
  technological frontier that progresses at a
  constant rate
• If a firm does not innovate then it falls one
  step behind its current technology (i.e.
  technology depreciates over time)
• Potential entrants start from (at least) one step
  behind the frontier
• Incumbent firms (which have already innovated)
  start at the current technology frontier

8
The impact of product market reforms

• Assumptions product market reforms
• There is a competitive fringe of firms that can
  produce using inferior technology (at higher
  cost)
• Product market reforms increase the extent of
  inferiority of the fringe

9
The impact of product market reforms

• Entrants
• Start behind the technology frontier
• If they innovate they can achieve a higher level
  of technology, but they can not advance so far
  that the competitive fringe does not represent a
  constraint
• Therefore, increased competition means the
  returns to innovation are lower
• Schumpeterian effect

10
The impact of product market reforms

• Incumbents
• Start at the technology frontier
• If they do not innovate they face competition
  from the fringe
• If they innovate they can keep up with the
  technological frontier, and thus the fringe does
  not represent a constraint
• Increased competition means that the loss from
  not innovating is greater (the fringe is more
  competitive), returns to innovating (and escaping
  the fringe) are higher
• Escape competition effect

11
The impact of employment protection legislation

• Negative impact on innovation
• In order to innovate firms have to fire workers
  with obsolete skills and hire workers with new
  skills
• EPL increases the cost of innovating
• theses costs may be higher for firms that are
  closer to the technological frontier
• Positive impact on innovation
• In a labor market with search frictions, the
  existence of unemployment benefits tends to
  reduce job mismatches
• Incentives for firms and workers to invest in
  match specific

12
Empirics
i country (10) j technology class (115) t year
(1980-1999)
13
Data

• Number of patents per 1000 population
• all patents
• patents by new patenters (entrants)
• patents by existing patenters (incumbents)

14
Data

• Count of patents
• granted by European Patent Office (EPO)
• 10 countries
• 20 years (1980-1999)
• 115 technology classes
• allocate patent to countries by inventor location
• sum patents taken out in each country
• by all
• by new patenters (entrants),
• by existing patenters (incumbents)
• Ongoing work
• matching patents to firm data and ownership
  structure

15
Patents per 1000 population
16
Data
Output - labour costs - capital costs labour
costs capital costs

• Ex post rents captured by firm
• (Nickell (1996), Boone (2000), Roeger (1995))

17
Data

• Mark-up endogenous to patents
• Instrument using product market reforms
• EU single market programme
• Government bureaucracy burden on business
• Procedures to start new business

18
Data

• Employment protection legislation index
• procedural inconveniences, direct cost of
  dismissal, notice and trial periods, rules
  applying to temporary workers
• OECD data

19
Data

• Labour market institutions
• Employment protection legislation index from OECD
• procedural inconveniences, direct cost of
  dismissal, notice and trial periods, rules
  applying to temporary workers
• Tax wedge on labour income
• Benefits replacement rate
• first year unemployment benefits as share of
  average earnings
• Coordination index
• centralization of bargaining, employers
  federations etc.
• Union coverage and density

20
Results
Table 4 dependent variable patents per 1000
population
Notes 13781 observations on 10 countries,115
technology classes, 1980-1999, dependent variable
is number of patents per 1000 population, numbers
in () standard errors that are robust to the
presence of arbitrary heteroskedasticity,
Validity of instruments is Hansen J statistic
over-identification test
21
Results
Table 4 dependent variable patents per 1000
population
Notes 200 observations on 10 countries,1980-1999,
dependent variable is number of patents per 1000
population, numbers in () standard errors that
are robust to the presence of arbitrary
heteroskedasticity, Validity of instruments is
Hansen J statistic over-identification test
22
Results
Table 4 dependent variable patents per 1000
population
Notes 200 observations on 10 countries,1980-1999,
dependent variable is number of patents per 1000
population, numbers in () standard errors that
are robust to the presence of arbitrary
heteroskedasticity, Validity of instruments is
Hansen J statistic over-identification test
23
Results
Table 5 dependent variable patents per 1000
population
24
Results
Table 5 dependent variable patents per 1000
population
25
Results
Table 5 dependent variable patents per 1000
population
26
Economic significance

• SMP led to around 6 percentage point reduction in
  average markup in UK (from 1980 to 1999)
• reduction in markup of 0.06 led to
• reduction of around 45 patents per person by
  entrants around 14 of mean
• increase of around 82 patents per person
  incumbents around 13 of mean
• EPL fell by 1 in Belgium
• increase of around 121 patents per person by
  entrants around 60 of mean
• increase of around 442 patents per person by
  incumbents around 42 of mean

27
Results
Table 6 dependent variable patents per 1000
population
28
Results
Table 6 dependent variable patents per 1000
population
29
Power of instruments

• Instruments are
• SMP
• administrative burden
• ease of entry
• EPLSMP

30
Economic Significance

• Impact of increasing competition
• at low EPL
• Australia, Canada, UK, US
• effect on incumbents - positive
• effect on entrants - negative
• at intermediate EPL
• Denmark, Netherlands
• effect on incumbents - small or zero
• effect on entrants - small or zero
• at high EPL
• Belgium, France, Italy, Norway
• effect on incumbents - negative
• effect on entrants - zero

31
Summary and further work

• Showed that increased competition encourages
  incumbents to patent more, not entrants
• Employment protection dampens incentives to
  patent
• High levels of employment protection mitigate the
  impact of competition
• Further work
• Numerous measurement issues
• Link patent ownership using firm level data
• Include other control variables
• Confirm results using other outcome variables