Transcript Measuring market power

Measuring market power
Lecture 32
Economics of Food Markets
Alan Matthews
Lecture objectives
• What methods can economists use to
measure market power in the food
industry?
• What is the evidence on the use/abuse of
market power from empirical studies?
Readings
• Processor power
– US Senate and Tweeten response
• Retail buyer power
– Dobson, UK Competition Commission
• Empirical studies
– Bunte, London Economics
A priori stylised facts
• Growing concentration upstream and
downstream of farmers
• Farmers’ share of consumer spending is
falling
• Farm-retail price spreads are widening,
suggesting farmers are being squeezed
• When farm prices fall, retail prices rarely
follow, suggesting profit-taking by
oligopolistic firms
Farmers’ share of consumers’
spending
• Can be measured in various ways:
– The marketing bill:
• the difference between consumer expenditure on
food (excluding imports, beverages and seafood)
and the farm value.
• Reflects changes in price, product quantity,
product mix and the quantity of marketing services.
– The market basket approach:
• measures the changing cost of a fixed basket of
groceries
• Cost changes solely the result of changes in prices
Farmers’ share of the consumers’
euro
– Farm-retail price spread
• Used for measuring farm share of individual products
• Must be measured in equivalent units
• Example
–
–
–
–
For steers, 2.5 kg of live weight yield 1 kg of retail beef cuts
2000 retail beef price = €8.40/kg average all cuts
2000 steer price = €1.20/kg live weight
2000 farm-retail price spread = €5.40/kg retail cut
(= 8.40 – 2.5*1.20)
• However measured, evidence that farmers’
share is falling over time
Evolution of Danish farm retail price
spreads (Source: Baker 2003)
Interpreting changes in the farmers’
share
• Is a large farm-retail price spread necessarily
bad?
– Shift in consumption patterns towards food with
higher value added and more food eaten-away-fromhome (marketing bill)
– Factor productivity increases more rapidly in
agriculture than in manufacturing, let alone services
– Could be due to growing market power
– Latter suspicion fuelled when reductions in farm
prices are not passed through in lower retail prices
Real price of food is falling….
Source: Agri-Aware website
…but Irish food prices remain high
compared to other EU countries
2003 comparative price level indices for the main food sub-groups, EU25=100
Source: Eurostat
Buyer power at retail level
• Refers to the ability of leading retail firms
to obtain from suppliers more favourable
terms than those available to other buyers,
or which would otherwise be expected
under normal competitive conditions.
– Extract discounts and obtain low prices
– Onerous contractual obligations
– Ability to shift financial risk to suppliers
Sources of retail buyer power
• Differences in relative economic
dependency, determined by relative sizes
of contracts and relative switching costs
• Multi-faceted roles of retailers as they
appear to suppliers (Dobson 2005)
– Customers, competitors and suppliers
Examples of retail buyer power
• Depends on whether firms operate in a
market or bargaining framework
• Demand withholding
• Adverse investment and innovation effects
• The ‘waterbed’ effect
• Supply chain practices
Supermarket practices related to
relations with suppliers
Category of practices
Number of practices
against the public
interest
Payments for access to shelf space
4
Imposing conditions on suppliers’ trade with other
retailers
0
Applying different standard to different suppliers
1
Imposing an unfair imbalance of risk
10
Imposing retrospective changes to contractual terms
6
Imposing charges and transferring costs to suppliers
5
Requiring suppliers to use third party suppliers
nominated by the retailer
1
Source: UK Competition Commission (2000)
Measuring welfare loss of market
power
•Role of demand elasticity
•What about the Posner rectangle A?
Approaches to the empirical
analysis of farm-retail price spreads
• Structure-conduct-performance paradigm
• New empirical industrial organisation
• Time series price transmission studies
Structure-conduct-performance
studies
• Examine the cross-industry or geographic (cross
section) or time (time series) variation in prices,
controlling for local market cost variations
– Short review in London Economics Annex 5
• Market concentration measures – Cn measures,
HHI index
• Indicators of market power – price-cost markup
(Lerner index), prices, margins, profitability
• Many studies have found significant relationship
between measures of market concentration and
indicators of market power
Empirical SCP studies
• Market power generates consumer welfare
losses…
• ..but concentration may also have positive
effects on firm efficiency
• Vulnerable to critique that market concentration
is correlated with profits because efficient (i.e.
low cost) firms acquire greater market shares
over time, not because of pricing power
• Rely on reduced form models. Firm behaviour is
not explicitly modelled and no statistical tests are
performed
New empirical industrial
organisation models
• Focus of this literature is on the conduct of firms
within a particular industry
• Based on structural oligopoly models by
specifying the first-order conditions for the profit
maximising behaviour of a single oligopolist
• The key behavioural parameter in this approach
is the conjectural variation of the oligopolist
(usually designated as θ), which measures the
degree to which a firm takes into account its
rival’s reactions to its own output choice
Supply of an oligopolistic firm
Profits for firm j = total revenue less
variable costs
Demand function
Total supply
Conjectural variations elasticity =
Percent change in total output for
a 1% change in firm output
First order condition. When θj = 0,
reduces to the competitive
outcome
Θj/η = Lerner index for oligopoly
power, represents the degree to
which a firm can set output price
above marginal cost
Source: Appelbaum, 1982
New empirical industrial
organisation models
• The expectation of a limited market response to
a change in firm output (low θ) suggests market
is competitive.
• Expectation of an extensive market response
suggests presence of market power (high θ)
• Estimating the structural model allows the
empirical data to provide information on the
value of θ
• Given the value of θ, the size of the market
power mark up can be calculated.
New empirical industrial
organisation models
• Weaknesses
– The estimate of market power relies crucially
on accurate estimates of the underlying
market and cost conditions
– Theoretically highly appealing, but complex to
apply empirically and often requires
simplifying assumptions
Price transmission studies
• Concern is that reductions in farm prices are not
passed on to consumers in form of lower retail
prices
• Three issues
– Prices changes are not fully transmitted
– There is a time lag between the price adjustments at
the respective stages
– There is an asymmetry in reaction between positive
and negative price shocks.
• Imperfections in price transmission may be due
to market power but can also be due to
adjustment costs (relabelling prices, advertising,
goodwill)
Price transmission studies
• Traditional regression methods (regressing
retail price on farm price) ignore time
series properties and give biased results
• Co-integration methods are now preferred,
which also allow for speed of adjustment
and direction of causality to be inferred, as
well as testing for asymmetric price
responses
London Economics case study
• Uses a variety of approaches
– Simple correlation analysis between changes
in farm-retail price spreads and concentration
ratios
– Reduced form SCP model linking spreads to
concentration, controlling for other factors
– Four food products – wheat products, red
meat, poultry, fruit and vegetables – drawn
from nine countries
Correlation coefficients are generally low, whether spreads are measured in
absolute amounts or as a ratio of the two prices (relative spread)
Highest correlation coefficient obtained for cereals of about 0.3
Source: London Economics 2004
When changes in concentration are correlated with changes in spreads, in
some cases the correlation is negative
Correlation coefficients are sensitive to the time period chosen, but even
when time period is divided, coefficients remain low
Source: London Economics 2004
Example London Economics SCP
model
• Purpose is to determine the influence of
changing concentration ratios on the farmretail price spread
• Model is
Spread = f(lagged spread, SCCOSTS,
CSHARE, LIP, EXRATE, C5, Trend)
Example London Economics SCP
model
Spread
Price spread (including lagged values)
SSCOSTS
Economy-wide indicator of costs sustained along
the supply chain
CSHARE
The share of the product used directly for human
consumption in total domestic supply. This variable
controls for any impact of demand and supply
shifts on farm-retail spreads.
LIP
Log(Intervention Price)
EXRATE
Vector of exchange rates against the euro
C5
Share of food retail market accounted for by top 5
firms
Example London Economics SCP
model
• Concentration in the retail domestic
market does not seem to have a
significant impact on the evolution of
spreads.
• Confirms conclusions from correlation
analysis
The Marion et al study of monopsony
power in US beef packing
• Tries to determine influence of packer
concentration on cattle prices and packer
margins
• Uses model
P = f (B, S, PG, R, NSD)
• Estimates using pooled cross section data
from 13 regional markets for time period
1971-1986
The Marion et al study of monopsony
power in US beef packing
P
Price of beef cattle
B
Structure of regional buying markets. Packers’ shares of
total cattle slaughtered using various concentration
measures
S
Structure of regional selling markets. Per cent of cattle
coming from feedlots with capacity of 1000 head or more
PC
Packer costs measured by three variables: employee
wages; economies of scale binary variable if large plant
existed; and distribution costs, distance from major market
R
Rivalry or market turbulence. Measured as change in CR
between years or relative share instability
NSD
National supply-demand, measured by either national
prices or yearly dummies
The Marion et al study of monopsony
power in US beef packing
• Conclusions
– Evidence that higher concentration is
significantly related to lower cattle prices paid,
other factors controlled for
– Relationship is less clear during period 197986 when packer concentration was increasing
sharply and there was excess capacity and
considerable competition for supplies