This presentation by Chiara Criscuolo (Head of the STI Division, OECD) was delivered during a workshop on “Methodologies to measure market competition” held virtually for competition authorities officials on 23 February 2021. More materials on the topic can be found at http://oe.cd/mmkts.
This presentation was uploaded with the author’s consent.
Methodologies to measure market competition – Chiara Criscuolo – Feb 2021 OECD Workshop
1. “COMPETITION” TRENDS IN
THE DIGITAL ECONOMY:
EVIDENCE FROM OECD
ANALYSIS
OECD Workshop for Competition Authorities
“Methodologies to measure market competition”
Chiara Criscuolo
Head of Division, STI, OECD
2. Multiple measures to try to catch trends in competition:
• Concentration
• Mark-ups
• M&A
• Entry and exit rates
• …
Each of them has big limitations in capturing “product market competition”.
However, they all seem to point in the same direction.
Just a coincidence? Maybe not!
Structure of the talk
3. Measuring concentration
So forth, we adopted 2 alternative measures:
1. Share of industry sales generated by the 10% largest firms
• Each country-2 digit industry is a “market”
• Representative data needed
2. Share of sales due to 8 largest business groups:
• Each world region- 2 digit industry is a “market”
• Definition of group and Allocation of sales across countries not trivial
• Importance of the denominator
Some additional challenges:
• Industry ≠ market
• Concentration of domestic production
• Country vs. world region vs. local
4. Rising industry concentration
Top 10% firms share in industry sales Top 8 firms share in industry sales
Share of sales due to 8 largest groups (rel. to 2000)
5. Measuring mark-ups
De Loecker & Warzynski (2012) methodology:
�
𝜇𝜇𝑖𝑖𝑖𝑖 =
�
𝑂𝑂𝑂𝑂𝑖𝑖𝑖𝑖
�
𝐼𝐼𝐼𝐼𝑖𝑖𝑖𝑖
Intuition: in perfect competition input shares = output elasticities, Solow (1957)
Possible criticisms:
1. Numerous assumptions are needed:
• Each firm is cost-minimising
• One fully flexible input e.g. intermediates
• Specification of the production function (e.g., Cobb-Douglas, translog)
2. Mismeasurement due to overhead costs? Market power ≠ high overhead costs
cost of intermediates as a
share of the firms revenue
estimated elasticity of
output with respect to
intermediates
6. Dynamics pushed by the top
Cobb-Douglas Translog
Log Mark-up growth over time (2001-2014) in different parts of the distribution
Source: Calligaris, Criscuolo, Marcolin (2020).
8. Declining business dynamism
Average within country-sector trends in job reallocation,
entry and exit rates
Source: Calvino, Criscuolo and Verlhac (2020).
9. COVID-19 exacerbates these trends
Aggregate business registration: differences 2020 vs.
2019
Employment effects of a missing generation of
firms
Simulation of the impact of a shock to entry on employment under two
scenarios
Note: Some data are experimental and may differ from official statistics.
Source: OECD (2021) “Business dynamism during the COVID-19 pandemic.
Which policies for an inclusive recovery?”
Source: OECD (2021) “Business dynamism during the COVID-19
pandemic. Which policies for an inclusive recovery?”
10. DO WE SEE DIFFERENT PATTERNS IN
DIGITAL SECTORS?
11. • Key role of the digital transformation and shifts to a knowledge economy
– Digital technologies lower entry costs, ease sharing of ideas, ease market
penetration
– Need complementary investments, increasing importance of tacit knowledge,
intangibles are non-rival and scalable
– Economies of scale and network externalities can reinforce winner-takes-most
dynamics
• This may generate
– Higher barriers to diffusion, lower experimentation and dynamism
– Advantages for larger firms that gain market shares and apply higher mark-ups
The role of the digital transformation
12. Increase in concentration stronger in intangible-
intensive industries
Intangible-intensive
Low-intangible
Share of sales due to 8 largest groups
Source: Bajgar, Criscuolo and Timmis, forthcoming.
13. Mark-ups higher in digital intensive sectors
Source: Calligaris, Criscuolo and Marcolin, (2020).
Average percentage differences in mark-ups
14. Increase in M&As
Large increase in number of
deals with digital targets
Source: Bajgar, Criscuolo and Timmis, forthcoming.
15. Declining Business Dynamism –
particularly in digital intensive sectors
Entry rates
Source: Calvino and Criscuolo, 2018.
16. Digital intensive industries have been relatively
sheltered from COVID-19
Higher ICT task content is associated with a lower decline in entry in 2020-Q2
Source: OECD (2021) “Business dynamism during the COVID-19 pandemic.
Which policies for an inclusive recovery?”
18. Share of industry sales generated by the 10% largest
firms: data
• The MultiProd dataset: distributed microdata approach
• Harmonized Stata routine sent to NSOs with access to confidential (administrative) firm-level
longitudinal data.
• They run the code on firm-level longitudinal data, which produces micro-aggregated data.
• Micro-aggregated data are sent back to us for analysis.
Sample
• To measure concentration only countries for which data are fully representative of the population
of firms: AUT, BEL, DEU, DNK, FIN, FRA, HUN, NOR, PRT, SWE.
• Manufacturing and non-financial market services.
• 2001-2021(14)
Concentration: 𝜗𝜗𝑗𝑗
10
≡ ∑
𝑖𝑖∈𝑆𝑆𝑗𝑗
10
𝑆𝑆𝑖𝑖𝑖𝑖
𝑆𝑆𝑗𝑗
where where 𝑆𝑆𝑖𝑖𝑖𝑖 designates the sales of firm i operating in industry j, 𝑆𝑆𝑗𝑗 the total sales in industry j and 𝑆𝑆𝑗𝑗
10
the top decile in the 2-digit
sector j of firms ranked by sales.
19. Share of sales due to 8 largest business groups: data
WORLDSCOPE
Group-Subsidiary Ownership Data
(2.8 million firms)
Sales Data for Subsidiaries, Parent & Group (100 Countries)
Sample
• Europe (BEL, DEU, DNK, FRA, FIN, GBR, GRC, ITA, PRT, ESP, SWE) + United States + Japan
• Manufacturing + construction + non-financial market services
• 2002-2014
Concentration: 𝐶𝐶𝐶𝐶𝑐𝑐𝑐𝑐𝑐𝑐
8
≡ ∑f=1
8 𝑆𝑆𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂
𝑆𝑆𝑐𝑐𝑐𝑐𝑐𝑐
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
21. Some methodological solutions: defining the “market”
(Affeldt, Duso, Gugler and Piechucka, 2021)
Idea:
• Industry ≠ market
• Country/region ≠ market
Relevant market defined on substitution patterns across product and geographical
dimensions.
How? Database that identifies over 20,000 product/geographic antitrust markets affected by
over 2,000 mergers decisions by DG Comp for 1995-2014.
Two selection issues:
• sampled only markets affected by large horizontal mergers
• market shares only sampled when DG Comp reports them
Concentration measure: market-specific post-merger HHI
Main result: concentration is increasing, and much larger than in the extant literature.
22. Rising mark-ups
Average of firm log mark-up: growth 2001-2014
Source: Calligaris, Criscuolo, Marcolin (2020).
23. Good vs. bad concentration?
Market power Superstars
– Increase in mark-ups and profit rates
• E.g. Gutiérrez and Philippon (2017), Barkai
(2019), De Loecker and Eeckhout (2017),
IMF (2019), Calligaris et al. (2018)
– Concentration associated with lower
investment & higher prices; role of
lobbying and regulations
• Gutierrez and Philippon (2019a,b)
– BUT concentration associated with higher
innovation and productivity
• Autor et al. (2019), Bessen (2017); Ganapati
(2018)
– Structural change disproportionately
benefits large/productive firms
• Autor et al. (2019), Liu et al. (2019). Crouzet,
Eberly (2018)