Practical use of microdata to
inform policy:
Firm level competition data
Chris Jenkins
Economics Director, CMA
14 October 2016
1
Introduction
● Can we use cross-economy microdata to identify
markets where there might be competition problems?
● Approach = use ONS and FAME microdata to construct
sectoral competition indicators
- Also productivity indicators?
● Outline
- Methodology
- Initial findings
- Productivity indicators
- Where next?
2
Indicators and data sources
3
Area Indicator Years Database
Concentration Number of firms 2009-13 ONS Business Structure
Database (BSD)
HHI 2009-13 BSD
Market share of largest firm 2011-13 BSD
Profitability EBIT margin 2011-13 FAME
Dynamics Churn - (entry+exit)/ turnover 2009-12 BSD
Coefficient of variation of
market leader
2008-13 BSD
Coefficient of variation of the C3 2008-13 BSD
Productivity Labour productivity by sector
compared with productivity
of related sectors
2009-12 ONS Annual Business
Survey (ABS) and ONS
BRES
Change in dispersion of labour
productivity
2008-13 ABS and BRES
Market size Total turnover by sector 2009-13 BSD
Pros and cons of competition indicators
● Pros:
- Comparable indicators
across whole economy - 728
sectors at 4/5 SIC code level
- Results can be updated over
time
- Provides top-down screen to
use alongside other sources
of intelligence
● Cons:
- SIC code often not a good match
for economic product markets
- Data is UK-wide – in practice
markets may be local, or
supranational
- Robustness of data may be
limited at a narrow sectoral level
4
● Overall: competition indicators will never be sufficient in themselves to
identify competition problems, but could potentially provide valuable
information to put alongside other sources of intelligence
5
Illustrative findings (1) – previous
market study sectors
6
Illustrative findings (2) - selected sectors
● Sectors score highly
across a number of
indicators
● However, in some
cases likely to span
many markets – (eg
‘organic basic
chemicals’)
7
Illustrative findings (3) - Financial sectors
Useful results?
● Results generally match our expectations of competition in these
sectors
- Useful source of information alongside other metrics
● Limit to how far we can take the analysis given difficulty of matching
SIC codes with economic markets
● One extension would be to consider imports/exports data as a proxy
for geographic scope
- Significant imports and exports might suggest competitive constraints
wider than UK
● We could also update indicators over time and look at movements in
indicators over a longer period
8
Can productivity be used as an
indicator to identify problem markets?
● In theory productivity could also be a useful measure – target
intervention on low-productivity sectors
- Rationale = competition is a driver of productivity, so low productivity
might indicate competition concerns
- However, low productivity might have nothing to do with lack of
competition – only a filter
● Key challenge is in developing a meaningful indicator which can be
compared across sectors
● Labour productivity varies widely between sectors, primarily because
of differences in capital-labour ratio (ie capital intensity) and quality of
capital
● We have therefore examined the productivity of a sector relative to its
industry average e.g. glues to all chemicals
9
Worst-ranked sectors based on relative
productivity – BUT note significant caveats
set out in following slides
10
Sector Relative
labour
productivity,
average 2008-
12 (£ 000’s)*
Absolute
sector
productivity,
average 2008-
12 (£ 000’s)
Change in
absolute
sector labour
productivity,
2008-2012 (£
000’s)
Strength of
competition
(high ranking
= less
competitive)
Satellite telecommunications activities -186 -64 NA 444
Inland passenger water transport -122 26 +7 384
Radio broadcasting -115 93 NA 471
Manufacture of basic pharmaceutical products -110 64 NA 310
Wholesale of petroleum and petroleum products -90 -39 -962 404
Renting of video tapes and disks -77 17 -20 451
Wired telecommunications activities -64 59 NA 243
Renting and leasing of recreational and sports goods -58 36 +42 291
Other treatment of petroleum products (excluding mineral
oil refining/petrochemicals manufacture)
-56 142 NA 377
Renting and leasing of personal and household goods -48 46 +41 298
* Relative labour productivity is obtained by subtracting productivity of the industry (2 digit SIC) from productivity of
the sector (4/5 digit SIC)
11
Estimated productivity based on GVA is likely to fall in
the short-run as competition increases
● Example competitive market:
- Workers produce 100 units
- Prices are competitive at £1
- Productivity ≈ £100
● In an uncompetitive market:
- Workers produce only 70 units
- The competitive price level is also £1,
so true productivity ≈ £70
- But price are excessive at £2
- So apparent productivity ≈ £140
● Is low productivity just a signal
of low profits i.e. effective
competition?!
Limitations of relative
productivity (1/3)
Robustness of survey evidence
● Worst-ranked sector on basis
of relative productivity is
satellite communications
● But chart suggests significant
data problems
- Negative productivity? (Driven
by negative GVA estimates)
- Very unstable productivities
ranging from £100 to -£250
(economy average ≈ £50)
- Is the industry benchmark really
comparable?
● Similar data problems affect
other sectors
12
Limitations of relative
productivity (2/3)
● Based on empirical
literature, would expect
low relative productivity to
be correlated with low
levels of competition
(upwards sloping line)
● Lack of any relationship
suggests relative
productivity measure is
not informative
● Question is whether we
could come up with any
better measures?
13
We do not find any relationship between our measures of
competition and relative productivity
Limitations of relative
productivity (3/3)
14
Possible next steps
● Longer time series data to look for trends?
● Use imports/exports data as a proxy for
geographic scope?
● TFP rather than labour productivity?
● Cross-country comparisons?
● Use indicators as an evaluation tool?

Practical use of microdata to inform policy: Firm level competition data

  • 1.
    Practical use ofmicrodata to inform policy: Firm level competition data Chris Jenkins Economics Director, CMA 14 October 2016 1
  • 2.
    Introduction ● Can weuse cross-economy microdata to identify markets where there might be competition problems? ● Approach = use ONS and FAME microdata to construct sectoral competition indicators - Also productivity indicators? ● Outline - Methodology - Initial findings - Productivity indicators - Where next? 2
  • 3.
    Indicators and datasources 3 Area Indicator Years Database Concentration Number of firms 2009-13 ONS Business Structure Database (BSD) HHI 2009-13 BSD Market share of largest firm 2011-13 BSD Profitability EBIT margin 2011-13 FAME Dynamics Churn - (entry+exit)/ turnover 2009-12 BSD Coefficient of variation of market leader 2008-13 BSD Coefficient of variation of the C3 2008-13 BSD Productivity Labour productivity by sector compared with productivity of related sectors 2009-12 ONS Annual Business Survey (ABS) and ONS BRES Change in dispersion of labour productivity 2008-13 ABS and BRES Market size Total turnover by sector 2009-13 BSD
  • 4.
    Pros and consof competition indicators ● Pros: - Comparable indicators across whole economy - 728 sectors at 4/5 SIC code level - Results can be updated over time - Provides top-down screen to use alongside other sources of intelligence ● Cons: - SIC code often not a good match for economic product markets - Data is UK-wide – in practice markets may be local, or supranational - Robustness of data may be limited at a narrow sectoral level 4 ● Overall: competition indicators will never be sufficient in themselves to identify competition problems, but could potentially provide valuable information to put alongside other sources of intelligence
  • 5.
    5 Illustrative findings (1)– previous market study sectors
  • 6.
    6 Illustrative findings (2)- selected sectors ● Sectors score highly across a number of indicators ● However, in some cases likely to span many markets – (eg ‘organic basic chemicals’)
  • 7.
    7 Illustrative findings (3)- Financial sectors
  • 8.
    Useful results? ● Resultsgenerally match our expectations of competition in these sectors - Useful source of information alongside other metrics ● Limit to how far we can take the analysis given difficulty of matching SIC codes with economic markets ● One extension would be to consider imports/exports data as a proxy for geographic scope - Significant imports and exports might suggest competitive constraints wider than UK ● We could also update indicators over time and look at movements in indicators over a longer period 8
  • 9.
    Can productivity beused as an indicator to identify problem markets? ● In theory productivity could also be a useful measure – target intervention on low-productivity sectors - Rationale = competition is a driver of productivity, so low productivity might indicate competition concerns - However, low productivity might have nothing to do with lack of competition – only a filter ● Key challenge is in developing a meaningful indicator which can be compared across sectors ● Labour productivity varies widely between sectors, primarily because of differences in capital-labour ratio (ie capital intensity) and quality of capital ● We have therefore examined the productivity of a sector relative to its industry average e.g. glues to all chemicals 9
  • 10.
    Worst-ranked sectors basedon relative productivity – BUT note significant caveats set out in following slides 10 Sector Relative labour productivity, average 2008- 12 (£ 000’s)* Absolute sector productivity, average 2008- 12 (£ 000’s) Change in absolute sector labour productivity, 2008-2012 (£ 000’s) Strength of competition (high ranking = less competitive) Satellite telecommunications activities -186 -64 NA 444 Inland passenger water transport -122 26 +7 384 Radio broadcasting -115 93 NA 471 Manufacture of basic pharmaceutical products -110 64 NA 310 Wholesale of petroleum and petroleum products -90 -39 -962 404 Renting of video tapes and disks -77 17 -20 451 Wired telecommunications activities -64 59 NA 243 Renting and leasing of recreational and sports goods -58 36 +42 291 Other treatment of petroleum products (excluding mineral oil refining/petrochemicals manufacture) -56 142 NA 377 Renting and leasing of personal and household goods -48 46 +41 298 * Relative labour productivity is obtained by subtracting productivity of the industry (2 digit SIC) from productivity of the sector (4/5 digit SIC)
  • 11.
    11 Estimated productivity basedon GVA is likely to fall in the short-run as competition increases ● Example competitive market: - Workers produce 100 units - Prices are competitive at £1 - Productivity ≈ £100 ● In an uncompetitive market: - Workers produce only 70 units - The competitive price level is also £1, so true productivity ≈ £70 - But price are excessive at £2 - So apparent productivity ≈ £140 ● Is low productivity just a signal of low profits i.e. effective competition?! Limitations of relative productivity (1/3)
  • 12.
    Robustness of surveyevidence ● Worst-ranked sector on basis of relative productivity is satellite communications ● But chart suggests significant data problems - Negative productivity? (Driven by negative GVA estimates) - Very unstable productivities ranging from £100 to -£250 (economy average ≈ £50) - Is the industry benchmark really comparable? ● Similar data problems affect other sectors 12 Limitations of relative productivity (2/3)
  • 13.
    ● Based onempirical literature, would expect low relative productivity to be correlated with low levels of competition (upwards sloping line) ● Lack of any relationship suggests relative productivity measure is not informative ● Question is whether we could come up with any better measures? 13 We do not find any relationship between our measures of competition and relative productivity Limitations of relative productivity (3/3)
  • 14.
    14 Possible next steps ●Longer time series data to look for trends? ● Use imports/exports data as a proxy for geographic scope? ● TFP rather than labour productivity? ● Cross-country comparisons? ● Use indicators as an evaluation tool?