Www.ccp.uea.ac.uk Investigating Market Power: a New Application Catherine Ball CLEEN New Researchers Workshop 2008 ESRC Centre for Competition Policy,

www.ccp.uea.ac.uk

Investigating Market Power: a New Application

Catherine Ball

CLEEN New Researchers Workshop 2008ESRC Centre for Competition Policy, UEA

13/06/08

Introduction - Homebuilding

• Barker Review, ‘Delivering stability: Securing our future housing needs’.

• OFT market study into new housing (‘homebuilding’).

• Need for some empirical research into the relationship between concentration and competition in the industry.

Introduction - Methodology

• Applying Bresnahan and Reiss framework to the homebuilding market.

• Using the relationship between the number of firms in a market and the size of the market to infer how competitive pressure in the market changes as the level of concentration changes.

Introduction - Methodology

Market size required for n

firms to break even

Number of firms1

S

2S

2

Introduction - Methodology

• Applying Bresnahan and Reiss framework to the homebuilding market.

• Using the relationship between the number of firms in a market and the size of the market to infer how competitive pressure in the market changes as the level of concentration changes.

• Advantages: Very low data requirements. In particular, as price-cost margins are inferred indirectly, there is no need for price data.

• Disadvantages: Relies on specific assumptions about functional form, market definition, etc. which may be difficult to relax.

Introduction - Methodology• Methodology allows demand entry thresholds to be calculated:• Assuming a firm’s latent profit function is:

• Then the market size required to support n firms is:

• And the per-firm demand entry threshold ratio is:

ˆ

ˆˆexp

nnS

DEM

n

n

n

S

n

Sss

nnnnnn 1

ˆ

ˆˆexp

1//

111

ni

nii

ni

ini

ni

S DDEM

)ln(

Data + data issues

• Market definition – Local Authority districts

• Number of Firms – EMAP data• Three ‘different’ counts

1. All firms2. Firms split into groups depending upon whether they are capable of

large or small scale builds.3. Firms split into groups depending upon whether they are capable of

building houses or flats

• Definition of market size – population/number of houses/housesales/number of jobs?

• Other factors that might affect firm profitability.

• Functional form/econometric specification?

Data + data issues

0.0

5.1

.15

De

nsity

0 10 20 30 40 50Number of Firms

Data + data issues

1

1.05

1.1

1.15

1.2

1.25

1.3

m

Data + data issues

0.0

5.1

.15

De

nsity

0 10 20 30 40 50Number of Homebuilding Firms in a Local Authority (flats sub-sample)

Data + data issues

0.0

5.1

.15

.2D

ens

ity

0 5 10 15 20 25Number of Homebuilding Firms in a Local Authority (houses sub-sample)

Data + data issues

• Market definition – Local Authority districts

• Number of Firms – EMAP data• Three ‘different’ counts

1. All firms2. Firms split into groups depending upon whether they are capable of

large or small scale builds.3. Firms split into groups depending upon whether they are capable of

building houses or flats

• Definition of market size – population/number of houses/housesales/number of jobs?

• Other factors that might affect firm profitability.

• Functional form/econometric specification?

Baseline Results – Ordered Probit model – Different market size definitions.

Baseline Results – Ordered Probit model – Different market size definitions (cont).

Baseline Results – Ordered Probit model – Different market size definitions (cont).

DETs – Ordered Probit model – Different market size definitions (cont).

Specification

1a(S=pop)

1b(S=housesales

1c(S=households)

1d(S=jobs)

s2/s1 0.937 (0.31) 0.962 (0.10) 0.942 (0.25) 0.999 (0.00)

s3/s2 1.031 (0.16) 1.041 (0.27) 1.035 (0.19) 1.07 (0.60)

s4/s3 1.086 (1.79) 1.091 (2.00) 1.088 (1.88) 1.119 (2.50)

s5/s4 1.143** (5.39) 1.152** (5.87) 1.144** (5.41) 1.181** (6.33)

s6/s5 1.064 (2.12) 1.067 (2.35) 1.064 (2.12) 1.093* (3.32)

s7/s6 1.053 (1.72) 1.054 (1.80) 1.053 (1.74) 1.077* (2.74)

s8/s7 1.049 (1.66) 1.049 (1.67) 1.050 (1.67) 1.701 (2.63)

Where the values in parenthesis are the Wald test statistics for the null hypothesis that the ratio is equal to one.

Flats/Houses Results – Ordered Probit model

Flats/Houses Results – Ordered Probit model

Flats/Houses Results – Ordered Probit model

DETs – Ordered Probit model – Flats/Houses

Specification

1a(S=pop)

3a(S=flats)

3b(S=houses)

s2/s1 0.937 (0.31) 0.914 (1.00) 1.906* (3.18)

s3/s2 1.031 (0.16) 0.989 (0.03) 1.326* (3.05)

s4/s3 1.086 (1.79) 1.030 (0.34) 1.587*** (7.34)

s5/s4 1.143** (5.39) 1.014 (0.35) 1.374*** (6.73)

s6/s5 1.064 (2.12) 1.091 (0.15) 1.174* (3.41)

s7/s6 1.053 (1.72) 1.078* (3.74) 1.220** (4.38)

s8/s7 1.049 (1.66) 1.016* (3.08) 1.112 (1.86)

Where the values in parenthesis are the Wald test statistics for the null hypothesis that the ratio is equal to one.

Conclusions

• Regardless of which measure of market size is used, the construction wages, the affordability of housing relative to income and the burden of providing affordable housing under S106 all have a significant effect on homebuilder profitability.

• All four measures of market size provide similar results, suggesting that a precise definition of market size is relatively unimportant provided the correct demographic controls are included in the model.

• Competition intensifies with the entry of the fifth firm into the market suggesting firms are able to avoid strong competition at high levels of concentration.

Conclusions (cont.)

• There are significant differences in the relationship between market concentration and competition in the submarkets of firms capable of building flats and firms capable of building houses.

• Competition between firms capable of building houses starts at much higher levels of concentration than is the case of firms capable of building flats.

• This is evidence that flat-builders and house-builders should be considered to be operating in different markets.