wirtschaft. politik. wissenschaft
Post on 15-Jan-2017
220 Views
Preview:
TRANSCRIPT
Competition Policy and Macroeconomic Outcomes
Tomaso Duso DIW Berlin, BCCP, & DICE Düsseldorf
LOOKING BEYOND THE DIRECT EFFECTS OF THE WORK OF COMPETITION AUTHORITIES: DETERRENCE AND MACROECONOMIC IMPACT
Brussels, September 17 2015
Introduction
Large consensus on the welfare-enhancing properties of competition
Achievement of allocative, productive, and dynamic efficiency it increases
productivity and growth (e.g., Aghion et al., REStat 2009)
Several studies provide measurement of other competition-enhancing policies
and their effectiveness
Privatization (Megginson, JEL 2001), trade liberalization (Pavcnik, RES 2002)
(de)regulations (Nicoletti and Scarpetta, EP 2003; Bourles et al, REStat 2013)
Market specific & focused instruments (e.g. entry regulations, price caps..)
Little evidence on whether competition policy is socially beneficial
Broad policy with many different tools affecting all markets simultaneously
Some authors conclude that competition policy is ineffective (e.g. Crandall and
Winston, JEP 2003)
Others claim that it is difficult to measure it because competition policy is a
deterrence system (e.g. Baker, JEP 2003 and Werden, JEP 2003)
Two Examples: Both based on research projects we undertook for the European
Commission (DG EC Fin, DG COMP)
1
Tomaso Duso 2 Competition Policy and Macroeconomic Outcomes
The Impact of Competition Policy on TFP Growth Study commissioned by DG Economic & Financial Affairs
Published in Buccirossi et al. (JCLE 2012) and Buccirossi et al. (REStat 2013)
Tomaso Duso 3 Competition Policy and Macroeconomic Outcomes
2
Competition Policy– Measurement Issues
How to define competition policy?
Competition legislation: set of prohibitions and obligations
Its enforcement: array of tools for policing and punishing violations
Institutions play an important role
Competition policy institutions –the procedures and rules– might affect enforcement (Duso
et al. JLE 2007, EER 2011, EJ 2013)
Competition policy & institutions are embedded in and interact with a system of other
institutions, in particular legal institutions (e.g., Buccirosssi et al. REStat 2013)
How to measure competition policy?
Competition policy has to be mostly seen as a deterrence system
The optimal level of deterrence is determined by 1) the size of the sanctions 2) the
(perceived) probability of detection and conviction, and 3) the (perceived) probability
of errors (Becker, JPE 1968; Polinsky and Shavell, 2000)
2
Tomaso Duso 4 Competition Policy and Macroeconomic Outcomes
Measuring Competition Policy: Optimal Deterrence
The following policy variables affect these three factors:
Formal independence of the CA with respect to political or economic interests
Degree of separation between the adjudicator and the prosecutor
Quality of the law on the books
Level of loss (sanctions) that firms (and their employees) can expect to suffer as a
consequence of a conviction
Type of investigative powers held by the CA
Amount and quality of the CA’s financial and human resources (the budget and skills
of the CA’s staff)
2
Tomaso Duso 5 Competition Policy and Macroeconomic Outcomes
We submitted a set of tailored questionnaires to the CAs in 13 jurisdictions and
integrated them with information from the OECD country studies and from the
CAs’ own websites
We obtained various information on six policy variables identified as
determinants of deterrence, separately for each type of possible competition
law infringement (hard-core cartels, abuses, other infringements) and for
mergers over the years from 1995 to 2005
Each piece of information at each step of the aggregation process was assigned
a score/weight on a scale of 0-1 against a benchmark of generally agreed best
practice
We tested the sensitivity of this weighting scheme to alternative ones using 1)
equal weights, 2) 1,000 sets of random weights, and 3) factor analysis
2
Tomaso Duso 6
Measuring Competition Policy: Optimal Deterrence
Competition Policy and Macroeconomic Outcomes
Impact of Competition Policy on Growth: The Framework
To assess the effectiveness of competition policy we build on a model of
endogenous growth (e.g., Aghion et al., 2009)
Laggard industries try to catch up with the technological frontier by innovating
Leader industries, try to escape competition by innovating and pushing forward the
technological frontier
We analyze the following causal links:
Competition Policy [Competition] Efficiency
As a measure of efficiency we choose TFP (and LP) growth
We control for all major drivers of TFP growth and estimate the following model:
3
Tomaso Duso 7
1 1 1
ijt
ijt it Ljt ijt ijt it ij t ijt
Ljt
TFPTFP CPI TFP X Z u
TFP
Competition Policy and Macroeconomic Outcomes
Impact of Competition Policy on TFP Growth: The Data
We selected 13 jurisdictions (Canada, Czech Republic, France, Germany,
Hungary, Italy, Japan, Netherlands, Spain, Sweden, the UK, EU, and US) over the
years from 1995 to 2005
For each country, our sample includes 22 industries based on the definitions of
the International Standard Industrial Classification (ISIC)
Data on TFP growth is drawn from the KLEMS consortium and from the
Groningen Growth and Development Center
Other data come from the OECD Structural Analysis (STAN) database, the OECD
Main Economic Indicators (MEI) database, the OECD PMR database, the OECD
Analytical Business Enterprise Research and Development (ANBERD) database,
and the World Bank Worldwide Governance Indicators (WGI) database
3
Tomaso Duso 8 Competition Policy and Macroeconomic Outcomes
Impact of Competition Policy on TFP Growth: Results
Competition policy has a positive impact on TFP and LP growth, which is
statistically significant at the 1% level
The impact is economically significant: e.g. the actual improvement of the CPI in the
UK in 2001-2002, is responsible for 22.1% of the increase of TFP growth in that year
(in “food products” 0.7% out of 5.2%)
The institutional dimension of the policy –more than its enforcement— and
the antitrust policy –more than the merger control— seem to have a stronger
impact
We provide evidence that the measured effect is causal and it is not driven by
omitted factors
3
Tomaso Duso 9 Competition Policy and Macroeconomic Outcomes
Impact of Competition Policy on TFP Growth: Results
We analyze dimensions of heterogeneity of this causal effect
1. We explore the interactions between legal institutions and competition policy.
CPI has a significantly larger impact in countries with better legal institutions
With low cost for the enforcement of contracts (Doing Business),
With high rule of law (Fraser)
2. CPI has a much stronger impact in the manufacturing sectors
Sector specific regulations in services industries (e.g. electricity, gas, water,
communication, financial intermediation) may make them to a lesser extent
subject to the ex-post antitrust scrutiny
3
Tomaso Duso 10 Competition Policy and Macroeconomic Outcomes
The Impact of Competition Policy Enforcement on the Functioning of Energy Markets
Study commissioned by DG Competition
Tomaso Duso 11 Competition Policy and Macroeconomic Outcomes
4
Competition Policy in energy Market
To study the long-term market-wide effects of overall policy enforcement – both
national and EU-wide – in energy sectors
Case studies focus on short-term effects of specific interventions
We postulate a relationship between policy enforcement – i.e., competition
policy and regulatory interventions— and market outcomes
4
Tomaso Duso 12 Competition Policy and Macroeconomic Outcomes
Competition Policy Enforcement: Data
EU competition policy enforcement: detailed dataset on EU competition policy
enforcement between 2005 and 2013 in energy markets (NACE D.35)
200 merger decisions, 17 antitrust decisions (16 abuses and 1 cartel), and 203 state
aid decisions
As energy markets still largely function at the national level, the EU data are
transformed to match the national level of analysis proposed in this study, i.e. the
member state/year unit of observation
Measurement of ‘Enforcement’
Mergers: remedies (in phase 1 or phase 2) and merger withdrawals during phase,
relative to total number of notifications
State aid: the decision to initiate a formal investigation, relative to total number of
notifications
Abuses and cartels: the number of cases (all led to remedies or fines)
Sent a template/questionnaire to all national competition authorities
We construct similar variables as for EU competition policy
Average numbers are quite low, but substantial variation across countries
4
Tomaso Duso 13 Competition Policy and Macroeconomic Outcomes
Competition Policy Enforcement: Data
OECD regulatory energy markets indices
Most used and most complete source of regulatory data in network industries
Used as an aggregate explanatory variable
To divide sectors into high-regulation and low-regulation industries
Assign sectors and firms to high-regulation (low-regulation) sub-sample if regulation index in
that country and year is higher (lower) than the median over all countries and time periods
Use sub-samples based on this definition rather than interactions between continuous indexes
to better exploit the limited variation in the data
Several additional sources to construct control variables
Institutional and governance data sourced from the World Bank
Information on the energy mix and energy imports
Macro factors, e.g., GDP/capita and population growth, from the OECD databases
4
Tomaso Duso 14 Competition Policy and Macroeconomic Outcomes
Outcomes
All measures based on Amadeus data (Nace D35.1 and D35.2 – very large firms)
Investment (Grajek and Röller, JLE 2009)
Monetary measure of investment: change in fixed assets owned by the firms
𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡𝑖 𝑡 = 𝑓𝑖𝑥𝑒𝑑𝑎𝑠𝑠𝑖𝑡 − 𝑓𝑖𝑥𝑒𝑑𝑎𝑠𝑠𝑖𝑡−1
Total factor productivity
The firm-level residual between its value added (VA) and an industry-specific, trans-log
production function f , which depends on capital (K), labour (L) and Material expenses (M)
𝑇𝐹𝑃𝑖𝑡 = 𝑉𝐴𝑖𝑡 − 𝑓(𝐾𝑖𝑡 , 𝐿𝑖𝑡 ,𝑀𝑖𝑡)
Productivity dispersion (Syverson, JEL 2011)
Standard deviation of TFP at sector/country/year level of observation
Elasticity of profits wrt. costs (Boone, RJE 2007)
Increased competition should reallocate profits towards more efficient firms
We estimate this relationship between firm-level (i) relative profits (RPD) and relative costs
(RCE) at sector (s)/country (c)/year (t) level of observation
𝑙𝑛𝑅𝑃𝐷𝑖𝑠𝑐𝑡 = 𝛼 + 𝜷𝒔𝒄𝒕𝑙𝑛𝑅𝐶𝐸𝑖𝑠𝑐𝑡 + 𝜇𝑠𝑐 + 𝜇𝑡 + 휀𝑖𝑠𝑐𝑡
4
Tomaso Duso 15 Competition Policy and Macroeconomic Outcomes
Emprical Framework
Market-level competition
𝑌𝑠𝑐𝑡 = 𝛼 + 𝛾𝑗𝐸𝑈𝐸𝑛𝑓_𝑡𝑜𝑜𝑙𝑐𝑡−1
𝑛
𝑗=1
+ 𝛿𝑗𝑁𝑎𝑡𝐸𝑛𝑓_𝑡𝑜𝑜𝑙𝑐𝑡−1
𝑚
𝑗=1
+𝛽𝑅𝑒𝑔𝑐𝑡−1 + 𝑍𝑐𝑡−2 + 𝜔𝑠 + 𝜔𝑡 + 휀𝑠𝑐𝑡,
𝑌𝑠𝑐𝑡 outcome variable of sector s in country c at year t
𝐸𝑈𝐸𝑛𝑓_𝑡𝑜𝑜𝑙𝑐𝑡 EU competition policy enforcement affecting country c in year t
𝑁𝑎𝑡𝐸𝑛𝑓_𝑡𝑜𝑜𝑙𝑐𝑡 national competition policy enforcement in country c in year t
𝑅𝑒𝑔𝑐𝑡 intensity of regulation in a given national market c in year t
𝑍𝑐𝑡 time-varying country-specific factors
𝜔𝑠 and 𝜔𝑡 sector and time fixed-effects
휀𝑖𝑐𝑡 error, correlated among observations within the same sector-country
Same equation for firm-level outcomes (but with 𝑌𝑖𝑐𝑡 𝑎𝑛𝑑 𝜔𝑖)
4
Tomaso Duso 16 Competition Policy and Macroeconomic Outcomes
Wrap-up of the Main Results
Clear overall picture: EU merger policy enforcement is consistently and
significantly related to better outcomes, especially in low-regulated sectors
Lowers both Boone’s beta and productivity dispersion in these sectors, indicating that
national energy sectors became more competitive after these interventions
Leads to higher investment and higher total factor productivity in these sectors
Interpretation: EU merger policy actions—through the channel of competition—
induce energy firms to invest more, which ultimately leads to a higher
productivity
Competition policy is mostly effective there where the competitive process is not
affected by a high regulation
4
Tomaso Duso 17 Competition Policy and Macroeconomic Outcomes
Wrap-up of the Main Results & Caveats
Other interventions
Other policy enforcement measures show a more patchy impact, i.e. show more
limited correlation pattern with market outcomes
But this cannot conclusively be seen as a lack of their effectiveness
Possible that their low frequency of happening did not allow us to empirically identify
consistent relationships
Careful with causal inferences
Tried to deal with reverse causality bias (by lagging policy variables)
Tried to deal with omitted variable bias (by using fixed effects and other controls)
But our identification strategy is not based on a clear source of exogenous variation
Virtually impossible to find valid exogenous variation in such a broad and
heterogeneous framework
5
Tomaso Duso 18 Competition Policy and Macroeconomic Outcomes
2
Tomaso Duso 19
Source: Buccirossi, Ciari, Duso, Spagnolo, & Vitale (JCLE 2012)
Source: Buccirossi, Ciari, Duso, Spagnolo, & Vitale, Journal of Competition Law and Economics, 2012)
Measuring the Determinants of Deterrence
Competition Policy and Macroeconomic Outcomes
The Competition Policy Indexes 2
Tomaso Duso 20
.3.4
.5.6
.7.3
.4.5
.6.7
.3.4
.5.6
.7
1995 2000 2005 1995 2000 2005 1995 2000 2005 1995 2000 2005
Can Cze Fra Ger
Hun Ita Jap Net
Spa Swe UK USA
CPI Aggregate CPI Aggregate with EU component
year
Graphs by country
Source: Buccirossi, Ciari, Duso, Spagnolo, & Vitale, Journal of Competition Law and Economics, 2012)
Competition Policy and Macroeconomic Outcomes
top related