Techniques for defining

markets for private healthcare

in the UK
Literature review
Prepared for
Office of Fair Trading

November 2011

Oxera

Draft for Comment: Strictly Confidential

Oxera Consulting Ltd is registered in England No. 2589629 and in Belgium No.
0883.432.547. Registered offices at Park Central, 40/41 Park End Street, Oxford, OX1 1JD,
UK, and Stephanie Square Centre, Avenue Louise 65, Box 11, 1050 Brussels, Belgium.
Although every effort has been made to ensure the accuracy of the material and the
integrity of the analysis presented herein, the Company accepts no liability for any
actions taken on the basis of its contents.
Oxera Consulting Ltd is not licensed in the conduct of investment business as defined in
the Financial Services and Markets Act 2000. Anyone considering a specific investment
should consult their own broker or other investment adviser. The Company accepts no
liability for any specific investment decision, which must be at the investors own risk.
Oxera, 2011. All rights reserved. Except for the quotation of short passages for the
purposes of criticism or review, no part may be used or reproduced without permission

Executive summary

The Office of Fair Trading (OFT) has carried out a market study into privately funded private
healthcare services in the UK (PH). Although it has not been necessary for the purposes of
the market study for the OFT to arrive at conclusions on market definition, the OFT
nevertheless considered that it would be useful, as part of the market study, to undertake an
analysis of different market definition techniques in order to inform any potential future
competition analysis in this sector, including in relation to mergers. In addition, some lessons
from the review may have wider applicability to the healthcare market, including competition
analysis carried out in the NHS sector.
In this context, it commissioned Oxera to provide a critical review of the literature on the
techniques for market definition in PH, and to provide recommendations on their application
to the UK PH sector. This report provides a practical guide for stakeholders by setting out
different approaches to defining markets for PH. It examines the advantages and
disadvantages of the approaches considered and comments on the suitability of the methods
for the UK PH market.1

Market definition in private healthcare

Certain features of the market for PH mean that the standard market definition techniques
are difficult to apply:

the majority of patients pay for their PH through private medical insurance (PMI), which
may mean that they are not sensitive to price changes made by individual hospitals.2
Standard techniques that define markets by imposing hypothetical price rises are
therefore not well suited;

the majority of patients may not have the knowledge or experience to determine which
hospital or consultant will provide them with the best treatment, and may therefore not
be able to determine the correct trade-off between price and quality;

unlike many other markets, each healthcare treatment involves interactions between a
number of parties. These include patients, private medical insurance (PMI) providers,
consultants, private hospitals or private patient units, and GPs.

A market definition exercise is traditionally undertaken as part of assessing competitive

constraints in a market.3 Both the product and geographic dimensions of the market in
question are considered. The starting point for defining the product market is to identify
whether the product in question has any close substitutes from the patients perspective
(ie, demand-side substitution) and from the suppliers perspective (ie, supply-side
substitution).
From the patients perspective, there is likely to be very limited (if any) substitutability across
types of treatment. In some cases there is evidence that supply-side substitution would be a
constraint on a hypothetical monopolist of a particular treatment. However, for certain
1

The report presents Oxeras analysis and does not necessarily represent the views of the OFT. The report should not be taken
as indicating the range of evaluation methods that the OFT may use in future cases.

The OFT may use different terms such as 'PH facilities' for reference in the market study, however, the international literature
reviewed uses various terminology depending on the structure on different healthcare systems. For this reason and for ease of
reference, in this report Oxera refers mainly to 'private hospitals'.
3

Although in certain cases (particularly mergers), there is an ongoing trend towards bypassing the market definition stage, and
testing the competitive effects directly.

Oxera

Techniques for defining markets for

private healthcare in the UK

treatments there may be no such supply-side constraint, or the constraint may be

asymmetric. In the absence of sufficient demand- or supply-side substitution, individual
treatments should be ideally defined as separate markets. For practical purposes, however,
competition authorities in different jurisdictions and US courts (in merger cases in particular)
have often aggregated individual services into clusters based on them being provided by a
common set of competitors.
The literature review in this report does not discuss product market definition in PH in detail.
The product market definition will often draw on clinical expertise and judgement, and may
also depend on the particular attributes of the competition case being considered. Instead,
the report focuses on the techniques for geographic market definition, and comments on the
interactions between geographic and product market definition where relevant.
The geographic market definition in PH is likely to have both national and local aspects.
National contracting occurs between PMI providers and suppliers of PH, but in most cases
patients have to travel to hospitals to receive treatment, and, because consumers prefer to
minimise the distance travelled, there will also be a local element to geographic market
definition.4 Much of the academic literature and case law on PH market definition has
focused on quantifying this local geographic element.
Nevertheless, it is useful to bear in mind that geographic market definition is likely to be
affected by the product market definition. A reasonable hypothesis would be that patients
may be willing to travel different distances depending on the type of treatment.

Techniques for geographic market definition in private healthcare

Techniques for geographic market definition in PH have been examined in great detail in the
academic literature, as well as in government reports, competition investigations and court
cases. The majority of the literature differentiates between the traditional, simpler techniques
developed in the 1980s and 1990s, and the more complex and recent approaches. Overall,
these techniques represent a broad spectrum of approaches that are characterised by
different degrees of theoretical soundness, complexity, data requirements and the extent to
which they have been tested empirically or have established precedent. These are shown in
the figure below.

This assumes that there is no perfect chain of substitution covering the whole of the UK.

Oxera

ii

Techniques for defining markets for

private healthcare in the UK

Figure 1

Spectrum of the main techniques for geographic market definition in the

literature

Elzinga
Hogarty

Time
elasticity

Competitor
share

Critical
loss

Isochrones

Willingness
to pay

Market
definition
approaches

Fully
structural
model

Theoretical soundness, data requirements

Source: Oxera analysis.

The techniques can be broadly divided into two categories. The earlier techniques
catchment area analysis and isochrones/fixed radii; ElzingaHogarty; and critical lossare
often conceptually less well-grounded, but benefit from a simplicity of application and lower
data requirements. The more recent techniquestime elasticity; willingness to pay;
competitor share; and GaynorVogt structural merger simulation model (MSM)
approachesare more sophisticated, but are also complex to apply and characterised by
substantial data requirements. Some empirical studies also use more informal approaches to
explore geographic markets based on the roles of other key market participants, such as
isochrones/fixed radii around consultants or GP practices.
The literature review reveals that earlier techniques typically do not capture certain
characteristics of PH markets. As a result, the use of these techniques may lead to defining
broad geographic markets, and there is therefore some precedent in accepting such broad
markets.
Empirical evidence is increasingly calling these decisions into question by showing that
relevant markets for hospitals can be very narrow, especially in urban areas, and that earlier
acceptance of broad markets in courts may have permitted mergers that led to a significant
increase in market power. This has resulted in the development of more sophisticated
approaches that seek to align the underlying model assumptions with the realities of the PH
market. The assessment of techniques for geographic market definition therefore needs to
account for the following characteristics of PH markets.

Heterogeneity of patients and hospitalsa good geographic market definition

technique would recognise that preferences (such as willingness to pay or willingness to
travel) may differ among patients; such a technique would also recognise that hospital
characteristics can differ (for example, by location or quality of service).

Oxera

iii

Techniques for defining markets for

private healthcare in the UK

Lack of patient price sensitivitythe majority of consumers pay for their PH through
PMI, and are therefore insensitive to immediate increases in the price of treatment.
Therefore, any technique that relies on the patients direct reaction to price is unlikely to
capture the geographic market accurately. In the long run, when high treatment prices
translate into higher PMI premiums, the likely outcome would be a reduction in demand
for PMI rather than switching between hospital; this is known as the payer problem.

Hospital networkscompetition between hospitals in the PH market often takes place

between hospital chains as well as between individual hospitals.

The next step is to consider which of the geographic market definition techniques are wellsuited to the UK market. To do this, it is useful to consider some of the key features of how
this market works and how it may be different from markets such as those in the USA and
the Netherlands, which have received ample attention in the literature. These features
include:

the central role of GPs as gatekeepers for private care;

the presence of the NHS alongside the PH market;
limited data availability because of the separation between the NHS and private
hospitals; and
significant functional separation (and often separate billing of patients and PMI
providers) between the contributions of a consultant and of a private hospital to any
given medical treatment.

These features of the market indicate that some considerations (such as data availability) are
likely to be more important than others for selecting the right geographic market definition
technique for the UK. In addition to data availability, five other criteria have been chosen to
assess which techniques are best suited to the UK:

theoretical underpinning;
data requirements;
complexity;
conceptual suitability for the UK market; and
established case practice.

The assessment against the five criteria of the earlier techniquescatchment area analysis
and isochrones/fixed radii; ElzingaHogarty; and critical lossshows that the techniques
suffer from conceptual shortcomings, in particular having arbitrary cut-off points, not
recognising the heterogeneity of hospitals and patients, and not addressing the lack of price
sensitivity of patients. However, there are practical solutions which could, at least to some
extent, alleviate these problems, such as adopting narrower product market definitions and
undertaking sensitivity checks around the cut-off points. On the other hand, the techniques
score well on the criteria of data requirements and complexity of application, since the data
required to apply the techniques may be accessible in the context of competition
investigations or can be obtained by means of a survey; and all models are simple to apply in
practice. There is also established precedent of using the techniques in competition cases in
the UK (with the exception of ElzingaHogarty) and in other countries.
The assessment of the more recent techniquestime elasticity; option demand/willingness
to pay; competitor share approach; and GaynorVogt structural MSM approachshows that
these techniques have more solid theoretical foundations than the earlier techniques. The
time-elasticity and willingness-to-pay approaches recognise that patients do not pay for
treatment directly, but that the treatment is paid for through their PMI. The willingness-to-pay
approach also has the significant advantage of reflecting the option demand nature of the
market in circumstances where PMI providers hospital networks do not have full coverage.
The GaynorVogt structural model and the competitor share approach both attempt to model
more realistic competitive behaviour between hospitals. All models suffer from some
drawbacks, however, often caused by sensitivity to the underlying assumptions.
Oxera

iv

Techniques for defining markets for

private healthcare in the UK

Furthermore, none of the models successfully takes into account all the characteristics of the
UK PH market discussed above. The advanced techniques also require highly granular data,
which would be difficult or impossible to obtain in the UK due to the effective separation of
the NHS and the individual networks of private hospitals; this limits significantly the extent to
which these methods could be applied in the UK.
Overall, the comparative assessment of the techniques reveals that there is a trade-off
between theoretical soundness and the feasibility of applying a technique in practice. In
general, there is no single technique that scores highly on every one of the five suitability
criteria. The earlier techniques tend to score less on the theoretical underpinnings but more
on ease of application, data requirements and established case practice. The more recent
techniques tend to score more on the theoretical underpinnings, but less well on the other
criteria.

Recommendations
As a result of the assessment, the following recommendations are made.

Advanced techniques based on merger simulation are likely to be useful in the UK only
in rare cases, where data availability is very good (and the competition authority has the
resources/capacity and time to undertake advanced analysis).

In light of the conceptual appeal of the more complex techniques and the fact that the
current level of data does not allow for their application, it may be desirable to put in
place measures that encourage the recording and storage of the data required for these
more advanced techniques, so that they could be used in competition cases.

Earlier techniques are appropriate in many circumstances where the time or budget
available for analysis is more limited, or where information is unobtainable. If the
techniques are applied in the right way, it is possible to avoid, or at least mitigate, the
concerns levelled at these techniques in the academic literature.

Within the set of earlier techniques, ElzingaHogarty and critical loss are likely to be less
appropriate than isochrone-type measures based on catchment area analysis. In the
case of ElzingaHogarty, the lack of a central data source of patient locations and
treatment makes its application more difficult in the UK than in some other countries.
Therefore, the additional benefit from applying this technique compared with the
isochrone-type measures (in terms of increased precision) may be outweighed by the
burden of the additional data requirements. In the case of critical loss, the insurancebased model in the UK creates a fundamental hurdle (as patients are not pricesensitive) that is unlikely to be fully overcome. In cases involving hospitals where there
are fewer PMI-funded patientssuch as those specialising in elective cosmetic
surgerycritical loss would be more appropriate.

When applying catchment area (isochrone or fixed-radius) techniques, the issues raised
above should be borne in mind. As far as possible, it may be sensible to avoid
assessments that bundle together treatments or groups of patients with systemically
different willingness to travel. Assessments should also take into account the potential
heterogeneity of hospitals, so it may be appropriate to apply different-sized isochrones
to different types of hospital.

As far as possible, given the significance of the impact of product market definition on
geographic market definition, when applying the catchment area techniques, empirical
analysis should be undertaken to examine the difference in travel times for patients
undergoing different types of treatment included in the product bundle in order to
prevent bundling together patients with different willingness to travel.

Oxera

Techniques for defining markets for

private healthcare in the UK

In the specific case of merger analysis, it might often be more appropriate to focus more
directly on the likely competitive effects of the transaction rather than on precisely
defining the market and calculating market shares. The local nature of competition
makes the direct assessment of competitive effects in specific local areas attractive.
Assessments that take into account the fact that demand is not symmetric around a
hospital should be used where possible, such as those that use postcode-based patient
discharge data to build a topographic picture of demand for a particular hospital.5

For Competition Act cases (those involving suspected abuse of dominance or

anti-competitive agreements), this direct analysis is less likely to be appropriate. In such
cases, it may be necessary for the OFT to form a more precise definition of the relevant
market (although in some cases the OFT may be able to apply threshold tests to
different candidate markets in order to establish that the relevant legal test is met).

In the case of market investigations, a precise market definition is less essential, but the
nature of the analysis, which must cover many hundreds of local areas, means that a
hospital-by-hospital analysis of local competition is unlikely to be useful or feasible.

The literature refers to, but does not explore in detail, some of the less common
approaches, such as GP- and consultant-based radii, and only limited empirical
evidence is available on these techniques. Given that the more advanced techniques
appear to be less appropriate for the UK due to data availability issues, it may be
desirable to explore these techniques empirically to determine whether they could
provide a suitable alternative to the more complex methods used elsewhere.

See, for example, Office of Fair Trading (2008), Completed acquisition by Spire Healthcare Limited of Classic Hospitals Group
Limited, para 20.

Oxera

vi

Techniques for defining markets for

private healthcare in the UK

Contents

Introduction

1.1
1.2
1.3

Context of the report

Purpose of the report
Structure of the report

1
1
1

Principles of market definition in private healthcare

2.1
2.2
2.3

The concept of market definition in competition analysis

Market definition in the PH sector
The interaction between product and geographic market
definition

3
3

Techniques for geographic market definition in

private healthcare

3.1
3.2
3.3
3.4

Overview
Earlier methods
More recent methods
Other methods

9
12
18
25

Critical assessment of techniques and applicability to

the UK
27

4.1
4.2
4.3
4.4
4.5
4.6

Key features of the UK PH system that affect market

definition
Criteria for assessing market definition techniques
Earlier techniques for geographic market definition in PH
More recent techniques for geographic market definition
Other techniques to measure aspects of hospital competition
Conclusions and recommendations

27
30
31
33
36
36

A1

Bibliography

39

A2

Summary of reviewed literature

42

A3

Summary of reviewed cases

50

List of tables
Table 3.1 Summary of techniques for geographic market definition in PH
Table 4.1 Assessment of techniques in the context of the UK PH market1
Table A3.1Market definition precedents in hospital mergers and acquisitions1

11
37
50

List of figures
Figure 1

Spectrum of the main techniques for geographic market definition in the

literature
Figure 2.1 The impact of defining the product market too broadly in a merger
Figure 3.1 Spectrum of the main techniques for geographic market definition in the
literature
Figure 4.1 Relationships between key players in the PH market

iii
7
9
29

Introduction

1.1

Context of the report

The Office of Fair Trading (OFT) has carried out a market study into privately funded private
healthcare services in the UK (PH). The aim of the study was to examine whether the market
was working well for consumers and, if not, whether there was potential for improvements.
Although it has not been necessary for the purposes of the market study for the OFT to arrive
at conclusions on market definition, the OFT nevertheless considered that it would be useful,
as part of the market study, to undertake an analysis of different market definition techniques
in order to inform any potential future competition analysis in this sector, including in relation
to mergers. In addition, some lessons from the review may have wider applicability to the
healthcare market, including competition analysis carried out in the NHS sector.
In this context, it commissioned Oxera to provide a critical review of the literature on the
techniques for market definition in PH, and to provide recommendations on their application
to the UK PH sector.

1.2

Purpose of the report

The aim of Oxeras study is to:

provide a comprehensive overview of the relevant UK and international precedent

identified by the OFT in relation to the approaches used to define product and
geographic markets for PH;

highlight, and set in a coherent framework, the issues and difficulties encountered in
defining markets in this sector;

assess critically the previous approaches used for market definition, by exploring the
strengths and weaknesses of each approach (in terms of theoretical underpinning or
practical application), as well as clearly stating the assumptions behind them;

provide a commentary on which approaches might be the most appropriate (based on

their relative merits) for defining PH markets in the UK.

Overall, the report aims to provide stakeholders with a practical guide that sets out different
approaches to defining markets for PH. It highlights the advantages and disadvantages of
these approaches and, in so doing, aims to inform the stakeholders approach and decisions
in relation to defining the markets for PH in the UK. In light of its generality, the framework
described in the report is likely to be useful for various stakeholders including the OFT and
other authorities and parties involved.

1.3

Structure of the report

The report is structured as follows.

Section 2 sets out the principles of market definition in PH.

Section 3 provides a comprehensive overview of the literature on the techniques used

for defining geographical markets in PH. This section sets out the advantages and
criticisms of each method that are commonly cited in the literature, describes the data

Oxera

Techniques for defining markets for

private healthcare in the UK

requirements for each technique, and summarises the precedent of using the
techniques in competition cases.

Section 4 provides Oxeras critical assessment of the techniques and examines their
applicability to the UK market for PH. The section concludes with recommendations
resulting from this assessment.

Oxera

Techniques for defining markets for

private healthcare in the UK

Principles of market definition in private healthcare

2.1

The concept of market definition in competition analysis

Market definition has traditionally been the first step in the assessment of the competitive
constraints in competition cases. The purpose of undertaking this step is to delineate the
market such that the relevant competitive constraints of the case can be isolated, and their
strength assessed. Market definition can be undertaken in different contexts, including
merger inquiries, market investigations and abuse of dominance cases.
A standard approach to market definition that is generally accepted in many jurisdictions is
the hypothetical monopolist test (HMT). The HMT works by defining a set of products or
services as a candidate market and then asking whether a hypothetical monopolist of that
set of products could profitably raise prices by a small but significant amount over the
competitive level.6 If the answer is yes, then that set of products is defined as a relevant
market. The test is iterative, starting with the smallest reasonable set of products (in the case
of mergers, this is usually the products sold by the merging firms) and widening the set of
products each time the question is answered negatively, up to the point where the question is
answered positively.7
In certain competition cases (particularly mergers), however, there is an ongoing trend
towards bypassing the market definition stage, and testing competitive effects directly. For
example, the Competition Commission (CC) and the OFT joint merger guidelines state that
market definition is just a starting point for competition analysis.8
Market definition is a useful tool, but not an end in itself, and identifying the relevant
market involves an element of judgement. The boundaries of the market do not
determine the outcome of the Authorities analysis of the competitive effects of the
merger in any mechanistic way. In assessing whether a merger may give rise to an SLC
[significant lessening of competition] the Authorities may take into account constraints
outside the relevant market, segmentation within the relevant market, or other ways in
which some constraints are more important than others.

Some of the methods discussed in this report focus solely on defining the market, while
others originate in merger simulation analysis and examine the effects of a merger directly.
However, a number of these merger simulation models (MSMs) can be used for both market
definition and the direct assessment of merger effects.

2.2

Market definition in the PH sector

The PH sector covers a very wide range of treatments, from brain surgery to Botox
injections, and from biopsies to heart surgery. As set out in the OFT report, the market for PH
encompasses a range of medical treatments which are privately funded and provided to

The HMT was first popularised following its use in the 1982 US merger guidelines. The test is also known as the SSNIP test,
because it asks whether the hypothetical monopolist could profitably impose a Small but Significant and Non-transitory Increase
in Price (SSNIP).

7
8

Niels, Jenkins and Kavanagh (2011), pp. 3742.

Office of Fair Trading and Competition Commission (2010), Merger Assessment Guidelines, para 5.5.2.

Oxera

Techniques for defining markets for

private healthcare in the UK

patients via PH facilities, through the services of consultants and medical professionals who
work within these facilities.9
The PH services that form the focus of the OFTs market study exhibit certain features that
make the application of standard market definition techniques difficult.10

Separation of payment from consumption. Due to the widespread use of PMI (more
than 60% of patients who are treated by PH providers are PMI-funded), many patients
are not sensitive to price changes made by individual hospitals and so standard
techniques that define market markets by imposing hypothetical price rises are not well
suited.

Asymmetry of information between consumers and providers. The majority of

patients may not have the knowledge or experience to determine which hospital or
consultant will provide them with the best treatment. Similarly, most patients may not be
able to determine the correct trade-off between price and quality. As a result, the
expectation is that many consumers of PH services are relatively passive, in the sense
that they are unwilling or unable to shop around for the best dealsat least when it
comes to choosing how and where to be treated. Instead, as indicated by the OFTs
research into the patient journey, many patients are willing to rely on the advice of health
professionals such as their GP or the consultant to whom they are referred.

Complex interactions between a number of parties. In many markets, the

relationship between parties in the supply chain is clear; however, each healthcare
treatment involves interactions between a number of parties. This is partly a function of
insurance-based systems, which include separate bilateral interactions between:

patients;
private medical insurance providers (PMI providers);
consultants/surgeons;
private hospitals or private patient units.11

In the UK, a further level of complexity is added through the role of GPs, who act as
gatekeepers to acute PH and refer patients to consultants for further examination and
treatment.
As a result, it is not straightforward to determine which interactions are most relevant for the
purposes of market definition. To some extent, all of the interactions are relevant, but the
most relevant ones will depend on the particular competition issue being investigated and the
theory of harm being tested.
Techniques for market definition in PH have been adapted to deal with these features. A
number of techniques have been proposed in the academic literature and, in some cases,
have been employed in merger investigations and court cases.
2.2.1

Product market definition

As with many market definition exercises, the two main elements are the product and the
geographic dimensions. The starting point for defining the product market is to identify
whether the product in question has any close substitutes from the patients perspective

For the purposes of the market study, the OFT defined acute care as short-term treatment via a range of medical and surgical
procedures commonly delivered by PH facilities with in- and outpatient settings. This excludes treatment for long-term
conditions.
10

Some, but not all, of the features listed correspond to features listed in Competition Commission (2000).

11

The OFT may use different terms such as 'PH facilities' for reference in the market study, however, the international literature
reviewed uses various terminology depending on the structure on different healthcare systems. For this reason and for ease of
reference, in this report Oxera refers mainly to 'private hospitals'.

Oxera

Techniques for defining markets for

private healthcare in the UK

(ie, demand-side substitution) and from the suppliers perspective (ie, supply-side
substitution).
From the patients perspective, there is likely to be very limited (if any) substitutability across
types of treatment, since, for example, knee surgery is unlikely to be an adequate substitute
for a hip replacement. However, for a particular treatment type there may be several
approaches that are to some extent substitutable (eg, different types of hip replacement).
From the supply side perspective the question is whether any attempt to raise prices (or
reduce service quality or increase waiting times) by a hypothetical monopolist of a particular
treatment would lead to providers of similar treatments switching to supply the specific
treatment in question. If this would be the case, it may be appropriate to include the capacity
of those providers in the market. However, in some cases there may be no evidence that
supply-side substitution would be a constraint on a hypothetical monopolist, or the constraint
may be asymmetric. These issues were considered in the assessment of a merger between
two public sector NHS Trusts in the UK.12
In the absence of sufficient demand- or supply-side substitution, individual treatments should
be defined as separate markets. For practical purposes, however, courts and competition
authorities often aggregate individual services into clusters based on their being provided by
a common set of competitors. The Federal Trade Commission (FTC) in the USA aggregates
markets on this basis, and has typically defined the product market for hospital services as
inpatient acute care services.13 Similar product market definitions have been defined in the
Netherlands and Germany.14
From both the demand- and supply-side substitution perspectives, the appropriate groupings
for product market definition are likely to require a degree of clinical judgement, so it may be
necessary to rely on medical experts. There may be some scope for economic analysis to
cross-check the assessment of a medical expert, by, for example, comparing the costs and
prices for different treatments. Whether such a cross-check is appropriate and feasible is
likely to be determined by the aspects of the individual case and the availability of data.
There is little discussion of product market definition in PH as the product market definition
will often draw on clinical expertise and judgement, and may also depend on the particular
attributes of the competition case being considered. The literature review in this report does
not discuss product market definition in PH in detail. The product market definition will often
draw on clinical expertise and judgement, and may also depend on the particular attributes of
the competition case being considered. Therefore, the literature review in this report (section
3) focuses on the techniques for geographic market definition, and comments on the
interactions between geographic and product market definition where relevant.
2.2.2

Geographic market definition

Geographic market definition in PH is likely to have both national and local aspects. National
contracting occurs between PMI providers and suppliers of private medical services (PH
providers), but in most cases patients have to travel to hospitals to receive treatment, and,
because it is assumed that they prefer to minimise the distance travelled, there will be a local
element to geographic market definition.15 However, this local element should not be
interpreted as meaning that consumers are willing to travel only a certain distance to receive
12

Co-operation and Competition Panel (2011), Merger of Nuffield Orthopaedic Centre NHS Trust and Oxford Radcliffe
Hospitals NHS Trust, September 30th.
13

See, for example, Chapter 4 of the FTC and DOJ report Improving Health Care: A Dose of Competition, available at
www.justice.gov/atr/public/health_care/204694/chapter4.htm.
14

See, for example, decision of the Netherlands Competition Authority in Case 6424/Walcheren Hospital - Oosterschelde
Hospitals, and Federal Cartel Office 10th Decision Division B 10 123/04 analysing the proposed acquisition by Rhn-Klinikum
AG of the district hospitals (Kreiskrankenhuser) of the Rhn-Grabfeld District, namely Bad Neustadt/Saale and Mellrichstadt
District Hospitals.

15

This assumes that there is no perfect chain of substitution covering the whole of the UK.

Oxera

Techniques for defining markets for

private healthcare in the UK

treatment. Given the positive impact that the treatments covered by the OFT study can have
on a patients quality of life, it is likely that they could be willing to travel very large distances
if there was no hospital close by. The local element to PH competition should instead be
interpreted as the patients preference to minimise their travel time, all else being equal.
Much of the academic literature and case law on PH market definition has focused on
quantifying this local geographic element. This is discussed in detail in section 3.

2.3

The interaction between product and geographic market definition

Geographic market definition is likely to be affected by the product market definition. A
reasonable hypothesis would be that patients may be willing to travel different distances
depending on the type of treatment. For example, one would expect patients requiring
complex treatment for a life-threatening or specialised condition (other than emergencies) to
be willing to travel further for medical care than patients who require only a relatively minor
and commonplace treatment. This points to patient heterogeneity within the wider market of
PH treatments.
In practice, product markets are often defined broadly. For example, in US and UK hospital
merger cases, markets have been defined as acute general hospital care (Gaynor and Vogt,
1999; Gaynor and Town, 2011; Zwanziger, Melnick and Eyre, 1994). By expanding market
boundaries in this way, the authorities avoid the need to determine whether there would be a
substantial lessening of competition in each procedure in which the hospitals overlap.
When the product market is widened through supply-side substitution or clustering of
services based on the common set of competitors, this also has an impact on how
geographic market definition should be considered. Simpler geographic market
methodologies do not account for patient heterogeneity. Indeed, any of the simpler
approaches to market definition generally work best for candidate markets where the
products are relatively homogeneous. The combination of a product market widened through
supply-side substitution (or clustering based on the common set of competitors) and a simple
approach to geographical boundaries could lead to overly broad geographic markets
because differences relating to specific treatments are ignored. The situation could arise, for
example, where a broad product market is defined covering hip operations and heart surgery
(such as a market for inpatient care), but where hip operation patients have a strong
preference for local hospitals and heart surgery patients do not. In this case, the geographic
market boundary would be drawn too wide for hip operations, leading to the possibility of
pockets of market power within the geographic market as defined.
Figure 2.1 demonstrates this point: the merger of hospitals A and B would be allowed based
on aggregate willingness to travel (bringing hospitals C, D and E into the market), but
patients with a lower willingness to travel and requiring a particular type of surgery would
face a reduction in competition from two to one.

Oxera

Techniques for defining markets for

private healthcare in the UK

Figure 2.1

The impact of defining the product market too broadly in a merger

Willingness to travel
for surgery X
Hospital C
Hospital A

Hospital B
Hospital D

Consumer

Hospital E

Aggregate willingness to
travel for all acute services
Source: Oxera analysis.

There is a growing volume of evidence to show that, in practice, patient preferences are
more disaggregated than the broad general care product definition. For example, Capps et
al. (2001) studied hospital choices in California in 1991 and found that patients with severe
health problems were more willing to travel. A study comparing patients aversion to travel
times for orthopaedic care and neurosurgery in the Netherlands also reports that patients
tend to travel further to receive complex treatments (Varkevisser and van der Geest (2007),
p. 7).
In light of this, some sort of clustering may be necessary to group treatments so as to
achieve patient homogeneity in those clusters. The following approaches have been
suggested in the literature as alternatives to defining the product market as acute hospital
care.

Varkevisser et al.(2004), cited in Varkevisser, Capps and Schut (2008), identify five
groups of specialist care in Dutch hospitals: i) high-volume complex specialities; ii) lowvolume complex specialities; iii) high-volume regular specialities; iv) low-volume regular
specialities; and v) specialities that can be provided by general or specialised hospitals
and stand-alone ambulatory surgeries.

Zwanziger, Melnick and Eyre (1994) recommend an alternative disaggregated approach

to reflect differentiation in inpatient care. Factors to consider are: i) the extent to which
treatments for two diseases can be performed by the same personnel and equipment;
and ii) the cost for a hospital to convert from providing one treatment to providing
another treatment. The authors argue that physicians (consultants) are the key inputs
into hospital care, and cluster diseases based on the least-specialised physician
capable of treating them.

Oxera

Techniques for defining markets for

private healthcare in the UK

There is also some precedent for disaggregating acute hospital services into clusters
that form separate product markets for the NHS hospital mergers. In the assessment of
the merger between Basingstoke and North HampshireWinchester and Eastleigh NHS
Trusts, the Co-operation and Competition Panel (CCP) recognises that there are groups
of treatments that face similar competitive constraints as they generally tend to be
provided together. The CCP identifies four such clusters: inpatient elective services,
inpatient non-elective services (eg, emergency and maternity services), community
services and outpatient services. The CCP, however, notes that if there are reasons to
believe that competitive constraints for a particular treatment are different from those for
the overall cluster, then the treatment can be assessed separately (ie, as a separate
product market).16

16

Co-operation and Competition Panel (2011), Merger of Basingstoke and North Hampshire NHS Foundation Trust with
Winchester and Eastleigh Healthcare NHS Trust, August 5th.

Oxera

Techniques for defining markets for

private healthcare in the UK

Techniques for geographic market definition in

private healthcare

3.1

Overview
Techniques for geographic market definition in PH have been examined in great detail in the
academic literature as well as in government reports, competition investigations and court
cases. The majority of the literature differentiates between the traditional, simpler techniques,
which were developed in the 1980s and 1990s, and the more complex and recent
approaches.
These techniques represent a broad spectrum of approaches that are characterised by
different degrees of theoretical soundness, complexity, data requirements and the extent to
which they have been tested empirically or have established precedent. This spectrum is
illustrated in Figure 3.1 below. It can be broadly divided into two categories. The techniques
on the left-hand side (shown in dark purple) are the earlier techniques; these are often
conceptually less well grounded but benefit from simplicity of application and lower data
requirements. The techniques on the right-hand side are more recent; these are more
conceptually sophisticated but their application is complex and they are characterised by
substantial data requirements.
Figure 3.1

Spectrum of the main techniques for geographic market definition in the

literature

Elzinga
Hogarty

Time
elasticity

Competitor
share

Critical
loss

Willingness
to pay

Market
definition
approaches

Isochrones

Fully
structural
model

Theoretical soundness, data requirements

Source: Oxera analysis.

Oxera

Techniques for defining markets for

private healthcare in the UK

The remainder of this section provides a comprehensive overview of the literature relating to
these and other techniques that could be used to define the geographic market for PH. This
literature review sets out the relative strengths and weaknesses of each approach discussed
in the literature, as well as data requirements and the precedent for their use in competition
cases.
The literature review indicates that earlier techniques typically do not capture certain key
characteristics of the PH market (such as heterogeneity of patients, heterogeneity of
suppliers, lack of sensitivity of patients to prices of treatments, and the fact that competition
between hospitals takes place at a network level as well as at an individual hospital level).17
As a result, the use of these techniques may lead to defining broad geographic markets, and
there is some precedent in accepting such broad markets. Empirical evidence is increasingly
calling this into question by showing that relevant markets for hospitals can be very narrow in
urban areas (Dafny, 2009), and that earlier acceptance of broad markets in courts may have
permitted mergers that led to a significant increase in market power (eg, Ashenfelter et al.,
2011). This has led to the development of more sophisticated approaches, which try to align
the underlying model assumption with the realities of the PH market. The assessment of the
techniques for geographic market definition therefore needs to account for the following
characteristics of PH markets.

Heterogeneity of patients and hospitalsa good geographic market definition technique

would recognise that patients preferences (such as willingness to pay or willingness to
travel) may differ between patients; such a technique would also recognise that hospital
characteristics can differ (by, for example, location or quality of service).

Lack of patient price sensitivitythe majority of consumers pay for their PH through
private medical insurance (PMI), and are therefore insensitive to immediate increases in
the price of treatment. Therefore, any technique which relies on the patients reaction to
price is unlikely to capture the geographic market accurately. In the long run, when high
treatment prices translate into higher PMI premiums, the likely outcome would be a
reduction in demand for PMI rather than switching between hospitals; this is known as
the payer problem.

Hospital networkscompetition between hospitals in the PH market often takes place

between hospital chains as well as between individual hospitals.

Table 3.1 summarises how the different techniques for geographic market definition
discussed in detail in the remainder of the section take (or do not take) into account these
characteristics of the market for PH.

17

Although in some cases it is possible to adjust the technique so as to address this issue, at least partially, as discussed in
section 4.3.

Oxera

10

Techniques for defining markets for

private healthcare in the UK

Table 3.1

Summary of techniques for geographic market definition in PH

Heterogeneity
of patients and
suppliers

Lack of patient
price
sensitivity

Insurer and
hospital
networks

Location of
patients of the
focal hospital
and location of
the focal
hospital

Not captured,
but adjustments
can be made to
partly resolve
the problem

Implicitly
captured
through patients
choice of
hospital

Considers
competition
between
individual
hospitals but not
between
networks

Distance
between patient
and hospital

Location of
patients and
location of a
hospital (for
patients treated
at the focal
hospital and
outside of focal
hospital)

Not captured

Implicitly
captured
through patients
choice of
hospital

Considers
competition
between
individual
hospitals but not
between
networks

Critical loss

Measures
reaction of
patients in
response to a
small price
increase

% of customer
switching
following a 5%
price increase.
Hospitals profit
margins

Not captured
(unless detailed
surveys are
used)

The test suffers

from the payer
problem unless
a time- or
quality-based
measure is used
in place of 5%
price increase

Considers
competition
between
individual
hospitals but not
between
networks

Time elasticity

Logit discrete
choice demand
function

Data on
patients
demographics,
diagnoses and
treatment
choices. Data
on hospital
characteristics

Captured

Captured, since
the model does
not rely on price
information

Considers
competition
between
individual
hospitals but not
between
networks

Option demand/
willingness to
pay

Logit discrete
choice demand
function

Data on
patients
demographics,
diagnoses and
treatment
choices. Data
on hospital
characteristics
and profits. Data
on PMI
providers
hospital
networks

Captured

Captures the
option value of
PMI

Captures insurer
hospital network
effects and
insurerhospital
bargaining.
Omits collective
bargaining by
hospital chains

Competitor
share approach

Logit discrete
choice demand
function

Price data for

each insurer
treatment pair.
Data on
patients
demographics,
diagnoses and
treatment
choices. Data
on hospital
characteristics

Captured

Patients are
assumed to be
sensitive to
price changes

Considers
competition
between
individual
hospitals but not
between
networks. Some
insurerhospital
bargaining
captured
through insurer
price data

Underlying
concept/model

Data
requirements

Catchment area
analysis

Distance
between patient
and hospital

ElzingaHogarty

Technique

Oxera

11

Techniques for defining markets for

private healthcare in the UK

Heterogeneity
of patients and
suppliers

Lack of patient
price
sensitivity

Insurer and
hospital
networks

Data on
patients
demographics,
diagnoses and
treatment
choices. Data
on hospital
characteristics.
Data on
hospitals costs,
revenues,
charges to PMI
providers and
information
about the
structure of any
hospital chain
operating
nearby

Captured

Patients are
assumed to be
sensitive to
prices to an
extent

Captured

Distance
between GP
and hospital

Location of GP
and hospitals to
which patients
were referred

Some patient
heterogeneity
captured
through
diagnosisspecific radii

Not captured

Not captured

Hospital HHI
versus system
HHI1

Change in
concentration

Patient
discharge data
by treatment
group for all
hospitals in the
area of interest

Some patient
heterogeneity
captured
through
diagnosis-based
HHIs

Not captured

Not captured

Physician-based
radii

Physicians
willingness to
travel

Location of
hospitals where
the physician
provides
treatments
and/or home
address

Not captured

Not captured

Not captured

Underlying
concept/model

Data
requirements

GaynorVogt
(2003) structural
merger
simulation
model (MSM)
approach

Differentiated
product
oligopoly, logit
discrete choice
demand function

GP referral
mapping

Technique

1
Note: HHI is the HerfindahlHirschman Index, which is used to measure the size of a firm relative to the industry
or the overall level of concentration in the industry as a whole.
Source: Oxera analysis.

The literature review in the remainder of this section is structured as follows: section 3.2
examines the traditional methods; section 3.3 reviews more recent methods; and section 3.4
summarises all other methods.
A number of the more recent techniques have been developed in the context of mergers;
these include time-elasticity, willingness to pay, competitor share and the fully structured
models. This may be due to the fact that the majority of competition cases in PH tend to be
mergers. These techniques are included in the literature review because many of the
approaches can also be used to define markets in other types of competition case; this is
discussed in more detail under the description of the individual models to which this applies.

3.2

Earlier methods

3.2.1

Catchment area analysis and fixed radii/isochrones

The basic approach to geographic market definition that has been taken in a variety of
industries is a customer (or, in the case of PH, patient) catchment area analysis. The
catchment area analysis captures the distance around the hospital from which a certain
Oxera

12

Techniques for defining markets for

private healthcare in the UK

percentage of the hospital patients originate, often taken as 8090%. This method is
commonly used as a starting point for examining the relevant geographic markets.
The catchment area measures the distance that the majority of patients would be willing to
travel to reach a hospital. The geographic market around a hospital is defined on this basis
as either a fixed radius (of, for example, 30 miles) or a fixed drivetime (of, for example,
30 minutes) based on the catchment area analysis. Any hospitals within this distance from
the focal hospital could be regarded as placing a competitive constraint on the focal hospital.
The same radius or drivetime is often used for all hospitals of interest, regardless of their
features and location. The geographic market has been defined using this method in a
variety of retail industries including groceries, cinemas and health foods outlets.18
The techniques are usually simple to apply but are not specifically tailored to healthcare
markets, and therefore do not take into account the unique characteristics of the market,
such as the high heterogeneity of products and consumers, and consumers limited
sensitivity to prices.
The main criticism of this approach that is commonly cited in the literature is that it lacks
economic theoretical underpinnings; as a result, the method for choosing the radius or the
drivetimea crucial driver of the resultsis inherently subject to some discretion, and
potentially contestable. This is because there is no guidance on what the cut-off point should
be for the percentage of patients that should be included in the catchment area analysis, so
this cut-off point selection is arbitrary (to a degree). Section 4.3 discusses the practical
approaches which can be used to at least partly address this issue.
Another criticism is that if the same radius or drivetime is used for all hospitals regardless of
their features and location then the heterogeneity of patients and hospitals might not be
accounted for, meaning that this approach may ignore the preferences and travel patterns of
certain patients. As a result, the geographic market definition based on this method may not
be representative of a patients true travel patterns.
Cooper et al. (2010), for example, argue that the bias in fixed-radius market definition is often
correlated with urban density, such that markets in metropolitan areas where population is
dense are defined too widely when the same fixed radius is applied as in less densely
populated, rural areas. The isochrone method has an advantage over the fixed-radius
approach, since it takes into account the local road networks. The geographic markets are
therefore less likely to be distorted by urban density, since this will be reflected in the fact
that urban areas would have lower speed limits, thereby leading to narrower geographic
markets compared with rural areas (Cooper et al., 2010).
Data requirements
The information needed to estimate patient catchment areas is the location of patients in
relation to the hospital. This would generally be available from hospital records. Alternatively,
when such records are unavailable, it may be possible to establish a typical catchment area
by asking the patients directly where they travel from to get to hospital. This would typically
be achieved by means of a patient survey. It may be possible to target PH patients if their
contact details could be obtained. If not, a national telephone survey could be carried out
(using respondents contact details obtained from a phone directory) to obtain a sufficient
sample of PH patients who have recently had private treatment.
Precedent
The academic literature does not discuss applications of patient catchment area analysis in
detail, although there is some precedent of it being used successfully in court. In particular, in
18

Competition Commission (2008), The supply of groceries in the UK: market investigation, April 30th. Office of Fair Trading
(2005), Acquisition by Terra Firma Investments (GP) 2 Ltd of United Cinemas International (UK) Limited and Cinema
International Corporation (UK) Limited, January 7th. Competition Commission (2009), NBTY and Julian Graves: A report on the
completed acquisition by NBTY Europe Limited of Julian Graves Limited, August 20th.

Oxera

13

Techniques for defining markets for

private healthcare in the UK

US v. Long Island Jewish Medical Center 1997 (Long Island), the merging parties used
patient catchment areas to argue successfully that the US Department of Justice (DOJ) had
failed to define the relevant geographic market and overturn the challenge to the merger
(Gaynor, Kleiner and Vogt, 2011).
Patient catchment areas have also been used in UK hospital merger investigations to inform
choice of isochrone sizes. For example, during the GHGAbbey (Office of Fair Trading,
2010) and GHGNuffield (Office of Fair Trading, 2008a) merger investigations, both the OFT
and the merging parties used the finding that around 80% of private hospital patients come
from areas within a 30-minute drivetime from the hospital to justify using this isochrone size
as a proxy for geographic markets in PH. The 30-minute isochrone was also used by the
OFT in analysing the extent of the overlap between the parties in the SpireCHG merger
investigation. For some hospitals the OFT also considered overlaps based on the 80%
patient catchment areas, observing that sometimes areas from which private hospitals drew
the majority of their patients could be skewed in one direction by the socioeconomic factors
(Office of Fair Trading, 2008b).
There is some evidence of fixed radii being used by UK competition authorities, although
without specific references to patient catchment areas to determine radius length. In
particular, the CC report on the BUPACHG transaction refers to the OFT having defined
geographic markets using 20-mile radii around hospitals as part of its preliminary
investigation (Competition Commission, 2000).
Isochrone analysis has also been used for defining geographic markets in the NHS hospital
mergers. For the most recent NHS merger between Basingstoke and North Hampshire Trust
and Winchester and Eastleigh Trust, the CCP defined the relevant geographic market as
hospital sites within a 3040 minute drive time of each hospital operated by the merging
parties. This isochrone size is based on a range of factors including patient referral patterns
and catchment areas for local hospitals.19
3.2.2

ElzingaHogarty (EH)
The EH test is one of the most widely applied approaches for defining hospital markets, and
it has been frequently used in contested merger cases in the USA. This method uses
hospitals patient flow data to gradually expand the geographic area around the focal
hospital(s) until the inflows of patients from outside the area into local hospital(s) and the
outflows of local patients to external hospitals both fall below an effectively arbitrary 1025%
threshold.
The benefits of using EH are not the primary focus of the literatureindeed most of the
recent research tends to focus on criticising this method. Nonetheless, there is a general
agreement that there are advantages, since it is relatively straightforward to implement and
has moderate data requirementsit requires only data on patient flows to hospitals within an
area, which can be obtained from centralised patient discharge databases.
Despite its attractive simplicity, use of EH to define markets in hospital care has been widely
criticised on a number of grounds; in fact, Elzinga (one of the academics who developed the
test) has testified in court that the approach, originally developed to analyse shipments, does
not address the relevant question of interest in the case of a hospital merger.20
The most frequent and significant criticisms of EH focus on its inability to deal with the
complexity of how patients choose hospitals in practice. For example, the EH method
implicitly assumes that patients who travel further for hospital services have the same
characteristics as those who travel shorter distances, and thus that the currently loyal
19

Co-operation and Competition Panel (2011), Merger of Basingstoke and North Hampshire NHS Foundation Trust with
Winchester and Eastleigh Healthcare NHS Trust, August 5th.

20

Elzinga and Swisher (2011).

Oxera

14

Techniques for defining markets for

private healthcare in the UK

patients would switch if the hospital raised prices. This assumption does not typically hold in
practice; patients are highly heterogeneous, and, while some may travel long distances,
perhaps to receive a one-off complex treatment that is not available in any nearby hospitals,
the majority are likely to have a strong preference for a local hospital. As a result, the
presence of a travelling patient sub-group need not constrain a hospitals market power over
most of the patients in its catchment area, giving rise to the silent majority fallacy of EH
(Capps et al., 2001; Elzinga and Swisher, 2011).
Many commentators observe that this shortcoming of EH is exacerbated by wide product
market definition.21 If product clusters for hospital merger analysis are too general, the size of
the relevant market is likely to be overestimated. This is because a minority of patients may
be willing to travel far for a highly differentiated product (possibly not even offered in the
merging hospitals), which would affect the EH results but which has no relevance for the
market power of the hospitals over the general care patients.
The implicit assumption that patient flows to hospitals are sensitive to prices is another
commonly cited reason why EH may be conceptually less suitable for hospital markets
(eg, Elzinga and Swisher, 2011; Varkevisser et al., 2008). In practice, an anticompetitive
merger would first increase prices paid by PMI providers, and patients would be affected only
when PMI providers pass on the costs in higher premiums. This payer problem implies that
the output reduction will be from consumers or firms not purchasing PMI, not from them
switching between hospitals as is assumed in EH.
EH has been criticised on a number of other methodological grounds. Kemp and Severijnen
(2010) observe that, although 10% and 25% cut-offs for inflows and outflows are commonly
used, these values have no theoretical or empirical foundations, leaving open the question of
the correct threshold where this choice makes a material difference to the case. Similar
scope for discretion exists in the process chosen to expand the geographical area if the
starting point does not satisfy the thresholds (eg, one can add individual zip codes
sequentially, ranked by market share, or gradually expanding the radius around the hospital
of interest). This makes the size and shape of the resulting market sensitive to alternative
implementations of the test (H.E. French in expert evidence aggregated by The Federal
Trade Commission and Department of Justice. 2004). The EH approach is also inherently
backward-looking due to its use of existing patient flow data, which may be suitable for abuse
of dominance cases, which tend to focus on historical behaviour, but not for predicting postmerger behaviour (Varkevisser and Schut, 2009).
Finally, there is a growing volume of empirical evidence that the markets defined by EH tend
to be too broad (Haas-Wilson and Garmon, 2011; Capps et al., 2001; and Geynor, Kleiner
and Vogt, 2011). These findings are also supported by ex post studies of effects of mergers
cleared on the basis of EH analysis. For example, Ashenfelter et al. (2011) report that courts
explicitly relied on EH results to define a broad geographic market in California v. Sutter
Health System 1999 (Sutter), a contested hospital merger in San Francisco. A recent ex post
study of this transaction by Tenn (2011) indicates that this merger had significant anticompetitive effects, with the acquired hospital raising its prices by significantly more than the
control group. This also suggests that the geographic market for the merger was narrower
than suggested by EH.
Data requirements
The limited data requirements of the EH test may help to explain its wide usage despite its
theoretical shortcomings. There are two key data requirements. The first is data on patients
from outside the focal area who attend hospital(s) within the candidate area. Since hospitals
hold details of each patients home address, it is a straightforward task to determine the
geographic area that captures, for example, 75% of a hospitals patients. The second
requirement is for data on patients from within the candidate area who are treated at
21

See, for example, Elzinga and Swisher (2011); Zwanziger et al. (1994); Varkevisser et al. (2008); and FTC and DOJ (2004).

Oxera

15

Techniques for defining markets for

private healthcare in the UK

hospitals outside the focal area. This data may be more difficult to acquire for parties which
are the focus of the competition investigation, but it may be possible for competition
authorities to request this data from hospitals outside the candidate area. In some cases, the
hospitals involved in the case will be part of chains and will be able to provide data on
patients from within the candidate area who are treated at hospitals in the same chain
outside the area.
Precedent
The EH test has been used in a number of contested hospital merger cases in the USA,
including Sutter, FTC v. Freeman Hospital 1995, and FTC v. Butterworth Health Corp 1996.
The tests influence, however, has recently waned. In an ex post litigation of an approved
merger against Evanston Northwestern Healthcare Corporation (ENH), the FTC defined the
geographic market as the area in which a significant number of individuals who seek hospital
care at the three ENH hospitals reside (FTC and DOJ, 2004), evidenced by, among other
things, ENHs ability to profitably impose significant and non-transitory price increases. The
court confirmed the limited market size, using a larger market than that proposed by the FTC,
but smaller than that proposed by the merging parties. Furthermore, based on the testimony
of Professor Elzinga, the administrative law judge concluded that Patient-flow data and the
ElzingaHogarty test are inapplicable to geographic market definition for a differentiated
product such as hospital services.22
The EH test has also been used in the Netherlands in the NMa investigation of the merger
between the Hilversum and Gooi-Noord hospital groups.23 This was the first Phase 2 hospital
investigation in the Netherlands. The NMa identified two separate product markets: one for
inpatient, and one for outpatient general hospital care. In the first assessment, the NMa used
the ElzingaHogarty approach to define the geographic market, leading to the conclusion
that the relevant geographic market was narrow because patients prefer local hospitals and
the merger would therefore lessen competition. The second phase used additional research:
i) interviews with GPs, hospitals and PMI providers; ii) conjoint analysis of patient
preferences; and iii) econometric simulations using patients revealed preferences.
3.2.3

Critical loss (CL) analysis

CL is a standard technique that is widely used to define markets in antitrust and merger
analysis. It provides a formal method for market definition based on the idea that if a
hypothetical monopolist of a set of products would be able to profitably raise prices then the
relevant market is no wider than that set of products. The CL test trades off the two effects of
a price rise: an increase in revenue from the higher price and a reduction in demand resulting
from the higher price. The CL is the percentage of sales at which the hypothetical monopolist
makes the same profit before and after the small but significant increase in price. If the actual
sales loss following the increase in price is higher than the CL then the price increase is
unprofitable and the market is therefore wider than that defined. If the actual loss is below the
CL, the price increase is profitable and the defined market is the relevant market.24 A more
detailed description of the test is set out as follows.

22
23
24

Step 1For a set of firms that are presumed to form a geographic market, calculate the
percentage reduction in demand that would render a fixed price increase (usually 5%)
unprofitable, as a function of the price increase and the gross margin of the firm.25 For
firms with high margins, the loss of even a few consumers would materially reduce
profitability, so the CL is likely to be low.

Federal Trade Commission (2005), p. 30, para. 216.

Niels, Jenkins and Kavanagh (2011), p. 63, footnote 36.
Niels, Jenkins and Kavanagh (2011), p. 57.

25

The gross margin is defined as the difference between price per unit and average variable cost. Its estimation is often a
highly contested issue but, by the way of illustration, margins used in previous hospital cases ranged from 41% in California ex
rel. Lokyer v. Sutter to 66% in FTC v. Tenet Healthcare (Gaynor, Kleiner and Vogt, 2011).

Oxera

16

Techniques for defining markets for

private healthcare in the UK

Step 2Calculate the actual percentage of sales that the firm would be expected to
lose if its price increases by 5% (estimated loss). For hospital mergers in the USA this
estimate is often obtained using the contestable zip code approach. This involves
identifying areas in which at least a fixed percentage (eg, 25%) of patients travel to
hospitals other than those presumed to constitute a market. These patient flows are
assumed to reveal the existence of other substitutes, which is taken as evidence that a
substantial number of currently loyal patients would switch if the hospitals in the
hypothesised market increased prices. The rate of patient switching is assumptionbased (eg, 30% of patients of hospitals of interest in contestable zip codes will switch as
a result of a 5% price increase).

Step 3Compare the estimated loss with the CL. If the estimated loss is higher
(showing that the small price increase is not profitable), expand the market by adding
hospitals that are viewed as the closest substitutes to hospitals already included.

CL is a classic market definition methodology, which has been widely used in hospital and
other markets, and so has been subject to rigorous examination as it directly implements the
SSNIP test used by competition authorities in defining markets.26 It is also attractive due to its
conceptual simplicity and relatively simple data requirements.
Many of the standard criticisms of the CL approach are valid in the context of defining
markets for hospital care. OBrien and Wickelgren (2003) argue, for example, that most
applications of CL do not take into account substitutability among products on which the price
increase is being considered, whereas in actuality, large cross-elasticities would allow a firm
to profit from a price increase by capturing lost sales through its other products. The authors
also observe that the argument that there would be a material actual loss of sales as a result
of a price increase, which is often accepted in courts, is generally inconsistent with the
existence of high margins.
Academic research identifies further potential problems with using CL in the context of
hospital markets. First, Varkevisser et al. (2008) observe that Elzinga and Swishers payer
problem described for the EH test also applies to the CL analysis, as lack of price sensitivity
among PMI-funded patients suggests that the traditional SSNIP test does not seem to be
conceptually applicable.27 Second, the hospital-specific approach to estimating actual loss as
a result of a price increase using contestable postcodes has attracted criticism similar to the
silent majority fallacy argument against EH by Capps et al. (2001). This approach assumes
that patients currently using the merging hospital are sufficiently similar to the travelling
patients in switching hospitals after the merger, which is often not plausible due to patient
heterogeneity (Gaynor, Kleiner and Vogt, 2011). A study of ex post effects of a merger in the
form of observed post-merger price increases by Simpson (2001) adds empirical support to
this criticism by finding that in all but one area that would have been deemed contestable
using the standard definition, an actual 5% increase reduced the merging hospitals market
share by less that 6%, significantly less than the estimated loss accepted by the courts.
Using hypothetical merger simulations among hospitals in the San Diego area, Gaynor,
Kleiner and Vogt (2011) demonstrate that as a result of these methodological issues, the CL
approach, like EH, tends to define excessively broad markets. This distortion is particularly
pronounced in urban areas.
Data requirements
The standard implementation of CL has two components: data on hospitals variable profit
margins and information on expected customer switching following a 5% price increase. If
the latter element is estimated using the contestable postcode approach specific to hospital
26

For example, Competition Commission (2003); and Federal Trade Commission and Department of Justice (2010).

27

In this context, lack of price sensitivity refers to the fact that the patient chooses the service provider with no knowledge of
the price and no incentive to change its behaviour regardless of the price rather than low price elasticity of demand.

Oxera

17

Techniques for defining markets for

private healthcare in the UK

mergers, predicted actual loss can be estimated instead by combining data on the
distribution of hospital patients across postcodes with assumptions on switching rates. This
requires data on hospital discharges both from the hospitals under investigation and other
hospitals in the area. (Simpson, 2001; Gaynor, Kleiner and Vogt, 2011). The main alternative
way of estimating actual loss uses consumer survey datathe customers can be asked
directly how they would change their behaviour in response to a hypothetical price increase
(or a change in a characteristic which approximates the price increase).
Precedent
There are several precedents of successful use of CL analysis in the US courts. In one
contested hospital merger, FTC v. Tenet Healthcare 1999 (Tenet), the circuit court
specifically gave very great weight to the defendants CL analysis in ruling that the
geographic market was as wide as 65 miles in radius, and reversing the district court
decision to block the merger (Ashenfelter et al., 2011; Gaynor, Kleiner and Vogt, 2011).
Moreover, CL was applied in United States v. Mercy Health Services 1995 (Mercy), in which
the DOJ originally challenged a hospital merger in Dubuque, Iowa, which left only one small
competitor in the 15-mile radius. The defendants used CL to argue for a very wide market,
including hospitals as far as 100 miles away, leading the court to clear the merger
(Ashenfelter et al., 2011). Gaynor, Kleiner and Vogt (2011) also report that CL was used by
both sides in the Sutter contested merger case described in section 3.1.1 above.

3.3

More recent methods

The more recent methods include time-elasticity, willingness to pay, competitor share and
structural merger simulation approaches. These methods specifically simulate effects of
mergers to determine the price increase which is likely to take place following the merger.
However, all four methods discussed in this section can in principle be used in the context of
other competition cases, in which geographic markets could be defined as the smallest set of
hospitals which, in a simulated merger, would materially increase prices while experiencing
relatively few substitutions to other alternatives.
The approaches in this section have been developed over the past decade in response to
methodological differences in the existing market definition methodologies, and growing
empirical evidence of incorrect conclusions that have arisen from their applications in
contested hospital mergers. The new proposed methods use sophisticated econometric
techniques that analyse factors that determine hospital choice for individual patients and
predict changes in their behaviour after the merger.
MSMs, however, have not yet been thoroughly tested (in relation to verifying the predictions
of the models ex post), and where they have been tested this has not been in cases involving
hospitals. There is some limited evidence on the performance of MSMs in other industries.
For example, Peters (2006) and Weinberg and Hosken (2008) in Ashenfelter et al. (2011)
show mixed empirical support for the ability of MSMs to correctly predict direction and,
especially, the size of price changes in other industries. Ashenfelter et al. (2011) report that
there are currently no ex post studies exploring the performance of MSMs in cases involving
hospitals.

3.3.1

Time elasticity
The time-elasticity approach is one of the geographic market definition methods proposed by
Capps et al. (2001) to address the excessively large markets that were often produced by
simpler methods which ignored heterogeneity of patients and hospitals. This new method
was specifically developed to circumvent two other methodological problemsthe limited
price sensitivity of PMI-funded patients, and difficulties in obtaining accurate measures of
prices charged by hospitals.
This approach examines patients willingness to substitute away from the hospitals of interest
by exploring their revealed willingness to travel further to other hospitals with particular

Oxera

18

Techniques for defining markets for

private healthcare in the UK

characteristics. Put simply, the time-elasticity approach allows the geographic market to be
defined according to how many consumers would switch to competing healthcare providers
in response to, typically, a hypothetical 10% increase in travel time to the merging parties.
Using a range of theoretical assumptions, the estimated effects of the merger on timeelasticities of patient demand are transformed into equivalent changes in the pricecost
margin for the hospitals. These results are then used to assess the full competitive effects of
the merger without relying on static consumer behaviour or observed hospital prices.
Since hospital prices are generally immaterial to patients, non-monetary factors such as
travel time can function as prices. This method proceeds as follows.28

Step 1Estimate a logit demand function as the probability that patient i chooses
hospital j, using the patient and hospital characteristics and factors specific to this
patienthospital pair (eg, travel times).

Step 2Use parameter estimates from the discrete-choice model to simulate the effect
of increasing travel time for each patient to a given hospital by a certain percentage
(eg, 5%).

Step 3Compare the effects of increasing travel times to the merging hospitals
individually with the effect of increasing travel times for them jointly. If the time elasticity
for the joint increase is much lower, the hospitals are close substitutes and more likely to
have market power jointly.

Step 4Assuming that consumers trade off time for money at a constant rate, price
elasticities should be directly proportional to time elasticities. Using the inverse-elasticity
pricing rule, this implies that post-merger increases in margins are directly proportional
to the reduction in time elasticity between joint and individual simulations in Step 3.

In proposing this new approach, Capps et al. (2001) argue that it has many advantages over
the earlier techniques for defining hospital markets. Its main aim is to capture adequately the
complexities of patients hospital choice. By estimating the patient demand model, which
explicitly includes a range of patient- and hospital-specific factors, the authors are able to
account for patient and hospital heterogeneity and thus minimise the risk of the silent
majority fallacy. Furthermore, by explicitly modelling factors that affect patients hospital
choice and willingness to travel, this approach produces results that are in line with the
recent empirical evidence that the relevant markets for hospital care may sometimes be very
narrow because in urban areas hospitals often exert competitive constraints on each other
only over short distances (Capps et al., 2002; Dafny, 2009).
Although the time-elasticity approach, like all other techniques discussed in this section,
takes the form of an MSM, it is also suitable for non-merger inquiries. Capps et al. (2002)
illustrate that this method can be used to define markets for general market studies by finding
the smallest set of hospitals, for which a small increase in patients travel times to all
hospitals in the set would result in relatively few substitutions to outside options.
Not using price data in the analysis is an additional advantage that sets this approach apart
from other merger simulation approaches. Scarcity and poor quality of hospital pricing data is
a common theme in the empirical literature on PH, and poor price proxies have often been
found to undermine the reliability of empirical studies (eg, Farell, Pautler and Vita, 2009; Vogt
and Town, 2006). Furthermore, PMI-funded patients are often not sensitive to a hospitals
prices or, as in the Netherlands, they do not observe a variation in prices across hospitals
(Varkevisser et al., 2008). The time-elasticity model has been developed specifically to
address the situations where prices are not directly observable or are not relevant for
patients.
28

The description of the methodology is adapted from Varkevisser et al. (2008).

Oxera

19

Techniques for defining markets for

private healthcare in the UK

The disadvantage of not using price data is that the model cannot allow some patients to be
price-sensitive, even in cases where treatments may involve out-of-pocket expenses in
practice. In presenting this methodology, Capps et al. (2001) acknowledge that, to avoid
distortions, the model needs to be estimated using patients who do not face price differences
across hospitals (eg, who have full PMI cover). Consequently, the merger simulation stage of
the analysis requires an assumption that their preferences with respect to travel distance are
comparable to other patients whose choices may be of interest (eg, self-pay or partially PMIfunded).
Estimating the patients hospital demand using the observed patient choice data requires
further simplifying assumptions. First, this model does not account for the restrictions on
patient choice from a limited insurer network, so patients are assumed to be free to switch to
whichever hospitals maximise their welfare (Capps et al., 2001). Second, simulating mergers
in terms of time elasticities requires assuming that patients trade off travel times and money
at a constant rate. Varkevisser et al. (2008) argue that this assumption needs to be validated
for each case (by, for example, stated-preference research), but even if it does not hold fully,
time elasticities are still broadly indicative of the existence of other available substitutes to
the hospital of interest.
Data requirements
The advantages of accounting for patient heterogeneity using discrete-choice models come
at a price of needing to use highly granular data on patients demographic characteristics,
diagnoses and treatment choices, as well as data on hospital features and quality.
Furthermore, large sample sizes are required due to the complexity of the model. All reported
empirical applications of the time-elasticity analysis carried out use large patient pools: over
27,000 hospital admissions for Capps et al. (2001, 2002) and 5,400 for Varkevisser et al.
(2010).
Precedent
The time-elasticity approach is the only technique in the class of merger simulation
approaches to have a public track record of use in competition investigations for hospital
markets. It was used, albeit with some modifications in Phase 2 of the cleared HilversumGooi-Noord (2005) merger investigation in the Netherlands (NMa, 2005).
3.3.2

Option demand/willingness-to-pay approach

It has long been recognised that US hospital care has strong features of the option demand
market, as patients select their PMI plans while healthy, on the basis of plan features such as
premiums and hospital network coverage, but then are largely insensitive to the price of
treatment (Gaynor and Town, 2011; Elzinga and Swisher, 2011). The first MSM to reflect
these features explicitly was developed by Town and Vistnes (2001), later extended in the
option demand approach by Capps et al. (2003).
The Capps et al. (2003) willingness-to-pay model is based on the idea that patients commit
to a network of medical providers covered by their insurer at the time they choose their PMI
provider, but before they know their medical needs. The value of the network to a consumer
is then based on how well they expect the firms in their insurers networks to meet their
needs when they materialise. The standard logit demand framework is estimated to derive
the value of each hospital to a consumer, conditional on a specific diagnosis and the
consumers demographic features. This makes it possible to derive the ex ante value of a
particular hospital network to all patients, using the probability distribution of diagnoses and
the distribution of consumer characteristics. This aggregate ex ante value reveals how much
consumers are willing to pay to retain a particular hospital in a network. High willingness to
pay suggests higher market power of a hospital over an insurer.
After deriving the willingness to pay for each hospital, Capps et al. regress observed hospital
profits on willingness to pay to find a conversion rate between willingness to pay (which is in
abstract units) and profits. This willingness-to-pay measure is found to be a highly significant
predictor of hospital profits.
Oxera

20

Techniques for defining markets for

private healthcare in the UK

The merger effects are simulated by finding the difference in willingness to pay for a merged
entity versus the willingness to pay for each hospital independently, since two merged
hospitals can increase their market power by coordinating their decision to join an insurance
network. This change in willingness to pay is converted to the change in profits, using the
parameter from the regression in the step above. Price effects are recovered based on an
assumption that quantities supplied remain unchanged. In general market studies, the same
results can be used by defining geographic markets as a set of hospitals, which, if merged,
are expected to increase prices by more than 5%.
The main rationale behind the development of the willingness-to-pay model, and its
frequently emphasised benefit, is its uniqueness among the merger simulation models in
explicitly modelling the insurerhospital bargaining, one of the main competitive dynamics in
the US hospital market (eg, Federal Trade Commission and Department of Justice, 2004;
Gaynor, Kleiner and Vogt, 2011). Furthermore, it avoids the payer problem posed by
Elzinga and Swisher (see above) by modelling patient choice as a two-stage process of
committing to an insurer ex ante before choosing a hospital.
The willingness-to-pay approach also shares the advantages of advanced approaches over
the more basic market definition techniques. By modelling demand using the interaction
between hospital and patient characteristics, this approach can capture heterogeneity and
produces flexible substitution patterns across hospitals and, therefore, plausible own- and
cross-price elasticities of demand (Capps et al., 2003). Moreover, it succeeds in capturing
local market power; comparative simulations of hospital mergers in the San Diego area
presented in Capps et al. (2002) show that, like time elasticity and other semi-structural
models, the willingness-to-pay approach is able to identify local pockets of market power that
can arise in small urban areas even if a large proportion of patients travels outside the area
for treatment.
The willingness-to-pay approach has the attractive attribute that it can capture the bargaining
dynamics between hospital groups and insurer networks, but this comes at the cost of
significant complexity and sensitivity to assumptions. As demonstrated by Vistnes and Town
(2001), the outcomes of the option demand models with hospitalinsurer bargaining depend
on the back-up options the insurer has if one of the hospitals is excluded from the network.
The willingness-to-pay approach, specifically, implicitly assumes that in this situation the
insurer does not replace the lost hospital with the next best alternative, which may be
implausible for many markets where insurer coverage is not universal. Furthermore, the
willingness-to-pay model contains a number of complex departures from the standard merger
simulation models, which have not yet been validated in courts or in retrospective merger
studies.29 There is also no precedent reported in the literature of the willingness-to-pay
method being used in competition investigations.
The literature also raises a number of potential problems related to the models approach to
patient demand. Varkevisser et al. (2008), for example, argue that calculating the ex ante
willingness to pay for hospital for each patient implicitly assumes that patients can accurately
predict their probability of requiring treatment for all possible diseases, which is highly
implausible. Furthermore, the models authors acknowledge that estimating the patients
demand and thus willingness to pay for hospitals using data on observed choices by PMIfunded patients may bias the results, because many of these patients may already be
committed to a restricted hospital choice set. Preliminary investigation, however, does not
find strong evidence of this bias (Capps et al., 2003).
Like other MSMs, the willingness-to-pay approach takes into account the heterogeneity of
patients and hospitals, and it is therefore able to identify localised hospital markets (Capps et
al., 2003). Gaynor, Kleiner and Vogt (2011) note, however, that it can estimate the

29

The paucity of empirical evidence on the quality of MSM predictions is discussed in Ashenfelter et al. (2011).

Oxera

21

Techniques for defining markets for

private healthcare in the UK

percentage increase in joint prices only after the merger, and cannot be used to isolate
asymmetric unilateral effects.
Data requirements
As is standard for merger simulation approaches, the willingness-to-pay demand model
requires a patient-level dataset of hospital choices and characteristics which is sufficiently
large to also draw conclusions about the demographic distributions in the population.
Furthermore, the approach requires data on the network structures of PMI providers in the
area of interest and on each hospitals revenues attributable to payments from PMI.
Precedent
No recorded precedent.
3.3.3

Competitor share approach

The competitor share approach is based on the intuition that the ability of hospitals to raise
prices following the merger depends on the substitutability between the merging hospitals,
which largely depends on the extent of overlaps in the types of patient the merging hospitals
treat (Capps et al., 2001; Varkevisser et al., 2008). Similar to other merger simulation
approaches, the starting point is a logit discrete choice demand function. Next, however, the
mathematical properties of the logit demand function are used to solve for the (implied) price
elasticities for each sub-market (ie, insurerdiagnosis pair) as a function of market shares of
competitors in the same sub-market. In merger analysis, the model simulates the changes in
market shares after the merger and infers the changes in price elasticities (and therefore
prices) for each sub-market, and for the merged hospital in aggregate.
Capps et al. (2002) outline the following steps for implementing the competitor share
approach in practice:

Step 1Define a set of sub-markets (for example, all unique insurerdiagnosis pairs).
There may also be patient-specific dimensions, between which the hospital is unable to
discriminate, but the distribution of these characteristics in the population for each
treatment is likely to be reflected in the overall price.

Step 2Estimate a logit (discrete choice) demand model for each sub-market, using
prices, patient- and hospital-specific characteristics.

Step 3Use the estimated parameters to derive the expression for the hospitals price
elasticities as a function of market shares of competitors in the same sub-market.

Step 4Compare differences in demand elasticities for the two hospitals jointly with
their individual elasticities.

Step 5Convert the reduction in price elasticity into price increase (via an increase in
margins).

The competitor share approach delivers the benefits of other merger simulation models, by
incorporating patient heterogeneity. Using interactions between patient and hospital
characteristics allows for flexible patient-substitution patterns across hospitals, since patients
will choose to switch to different hospitals depending on their illness or income (Capps et al.,
2001). As with other advanced approaches, this allows the competitor share approach to
identify local hospital market power effectively (Capps et al., 2001, 2002). Furthermore,
Varkevisser et al. (2008) suggest that the competitor share approach can be easily used for
non-merger market definition; in more general contexts, markets can be defined, as usual, as
a set of hospitals that, if monopolised, could increase prices with relatively few patients
substituting to outside alternatives.
The additional advantage that sets the competitor share approach apart from other merger
simulation methods is its ability to produce very granular information on competitive

Oxera

22

Techniques for defining markets for

private healthcare in the UK

constraints between hospitals. By breaking down hospitals activity into insurer and
diagnosis-based sub-markets, and estimating the competitor share effects for each, this
method can be used to identify pockets of market power for specific types of service such as
ante natal care (Varkevisser et al., 2008, Capps et al., 2001). This can be particularly useful
if, for example, two of the merging hospitals have few overlapping specialisations and
therefore do not compete with each other on many product dimensions (which would
otherwise mask the potential effects of the genuine overlaps).
One of the main shortcomings of the competitor share approach is that it requires very
granular data in light of its complexity. By producing effects of a merger on prices for every
diagnosisinsurer pair, for example, in their implementation of this approach, Capps et al.
(2002) obtain as many as 1,957 sub-markets using data for five PMI providers. This
potentially unwieldy number of results needs to be aggregated further to be tractable, and the
final outcomes may be sensitive to the choice of aggregation method. Furthermore, patients
are assumed to be sensitive to the variation in prices between the diagnosisinsurer pairs, at
least to some extent.
Capps et al. (2001) note two further conceptual difficulties that arise from the approachs
assumptions about hospitals pricing behaviour. First, in simulating the effects of hypothetical
mergers, the competitor share approach assumes that hospitals set equal prices after
merging. This does not necessarily occur in practice; for example, an ex post study of the
effects of two Dutch mergers reports that only one pair of hospitals had implemented
standardised pricing (Kemp and Severijnen, 2010). Second, the competitor share approach
evaluates hypothetical post-merger competitive constraints using pre-merger prices, in effect
omitting the likely dynamic response by competitors after the merger.
In addition to the above issues, the competitor share approach suffers from the same
shortcoming as the time-elasticity approach. By assuming that patients choice set includes
all hospitals in estimating the hospital demand model, the approach is potentially vulnerable
to bias if, in reality, many patients are constrained by restrictive insurer network coverage.
Data requirements
The highly differentiated results produced by the competitor share model require equally
granular data. The approach assumes that hospitals charge PMI providers or their patients
different prices for each hospital service, and requires price data for each insurertreatment
pair. In practice, this data may not be available, as in many countries at least part of the
remuneration agreements between PMI providers and hospitals is based on a fixed daily
charge basis (eg, Varkevisser et al., 2008; Federal Trade Commission and Department of
Justice, 2004).30 Furthermore, this data needs to be obtained on all hospitals in the area of
interest, not simply the merging parties.
As with other merger simulation approaches, the competitor share method also requires
granular data on individual patients hospital choices and characteristics. For example,
Capps et al. (2001) uses over 27,000 patient episodes to estimate the hospital demand
function.
Precedent
No recorded precedent.
3.3.4

GaynorVogt (2003) structural MSM approach

This is a structural model of consumer and hospital behaviour, based on theoretical
foundations described in the Berry, Levinsohn and Pakes (2004) model for differentiated
product oligopoly. It is the only MSM considered in the literature that attempts to model the

30

This is a particularly important point if prices for staying in the hospital (ie, hospital bed, catering, theatre facilities and nursing
staff) are disaggregated from the prices charged by the consultants for the procedure.

Oxera

23

Techniques for defining markets for

private healthcare in the UK

strategic interaction of the competing hospitals in the market, especially the potential effects
of reduced competition among hospitals belonging to the same chain.
The theoretical set-up for the model describes the dynamics of the PH market as Bertrand
competition with differentiated products.31 It models demand at the level of each patient,
using the standard multinomial logit discrete choice models based on micro data on
individual patients. This demand model allows the prediction of the probability of each
hospital being chosen by each consumer, using the set of consumer and hospital
characteristics, in interaction with the price of the hospital service. The demand faced by
hospital j can be derived by summing the quantities of hospital care demanded by each
consumer and the probability of each consumer choosing hospital j. Because prices are
endogenous, exogenous wages and predicted quantity (using only geographic distribution
and exogenous consumer characteristics) are used as instruments for price, thus enabling
the recovery of ap, the marginal utility of income.
This methodology departs from other merger simulation approaches described above by
including a structural supply-side model of oligopolistic competition among hospitals using
the Bertrand model of price-setting behaviour. This allows Gaynor and Vogt to recognise and
incorporate in the model the fact that hospitals often operate as multi-hospital chains, which
has a significant impact on substitution between hospitals within one chain and pricing
coordination. Using these demand- and supply-side models, the structural approach allows
one to solve explicitly for the own-price and cross-price elasticities faced by each hospital.
Effectively, this structural approach implements a highly sophisticated version of the CL
analysis. The amended SSNIP criterion proposed by the authors states that for a given
hospital, j, a SSNIP market is the smallest set of hospitals for which an increase in price at
this set of hospitals (including hospital j) would increase the collective profits in the systems
of which these hospitals are members (Gaynor, Kleiner and Vogt, 2011, p. 18). This has the
benefit of being consistent with the new US horizontal merger guidelines (2010), which, in
defining a market, require a hypothetical monopolistpossibly a chain of hospitals operating
in many geographic marketsto impose a SSNIP in at least one location, at least one of
which is the location of one of the merging parties.32 The algorithm used to define the
markets using the proposed structural model is as follows.
Step 1Begin with a hospital of interest for which the market needs to be defined.
Step 2Find the competing hospital that is the closest to the hospital of interest.
Step 3Raise the price of only these hospitals by 5%, and simulate the resulting
change in demand.
Step 4If total profits (given diversion to other hospitals in the same chain) increase as
a result, this constitutes a market and the SSNIP test stops here. If not, add the next
hospital that is closest to the hospital of interest.
All MSMs discussed in this section share an advantage of being grounded in theoretical
fundamentals of consumer demand, but in addition to this the GaynorVogt approach is
31

The description of the theoretical framework for the model draws on exposition in Gaynor and Vogt (2003) and Gaynor,
Kleiner and Vogt (2011).

32

The article illustrates this criterion through a useful example, quoted here in full: ... consider 4 hospitals, A, B, C and D, and
let A and B be members of the same hospital system. Suppose hospitals A and C act as a hypothetical monopolist and
engage in a coordinated price increase of 5% (holding the terms of sale constant at all other locations), resulting in a decrease
in demand at both hospitals and a decrease in profits at the combined hospital entity of A and C. Suppose, however, that B is a
sufficiently adequate substitute for care at these hospitals so that the increase in profit as a result of the increase in demand for
hospital Bs services is greater than the decrease in profits at the combined hospital entity of A and C. Hospitals A and C would
be a market under the SSNIP criterion, as the collective profits in the systems of which these hospitals are members has
increased. Likewise, if hospital D is a close substitute for the care rendered at A and C while hospital B is not, hospital B would
see little or no increase in demand or profits and thus hospitals A and C would not be considered a market according to the
SSNIP criterion. (p. 19).

Oxera

24

Techniques for defining markets for

private healthcare in the UK

unique in also having a theoretical underpinning for the supply of hospital services. This
feature of the model is important in capturing the fact that hospitals often operate in chains,
so a realistic implementation of the SSNIP test needs to allow for potential coordinated price
setting and some of the profits diverted from merging hospitals being captured by other
hospitals in the chain. Unlike the willingness-to-pay approach, this method also allows for the
prediction of different price increases for each of the merging hospitals rather than only the
joint effect overall (Gaynor, Kleiner and Vogt, 2011).
The GaynorVogt approach also shares the advantages of other MSMs, such as being
sensitive to patient heterogeneity, in particular willingness to travel, allowing for plausible
cross-price substitution effects (Gaynor and Vogt, 2003). A comparative empirical study by
Gaynor, Kleiner and Vogt (2011) shows that this fully structural approach can identify local
hospital market power, in contrast with the simple patient flow-based techniques. Finally, the
empirical results from recent studies (eg, Dafny, 2009) imply that the hospital markets are
similar in size to those obtained through the structural merger simulation approaches, such
as Gaynor and Vogt (2003) or Capps et al. (2003).
Generally, the GaynorVogt approach is not frequently discussed in the reviewed literature,
so only a limited range of critical academic assessments of this method is available. Besides
the models complexity, which may be a major obstacle for practical implementations, the
main shortcoming of the GaynorVogt approach is its treatment of patients as pricesensitive. The price sensitivity arises from including the total price paid by the insurer to the
hospital as an argument in the patients utility function, which, in turn, drives hospital choices
(Gaynor and Town, 2011). Although Gaynor and Vogt (2003) do report some empirical
evidence that prices affect patients hospital choice, the theoretical foundations of the model
seem to assume that the effect arises from PMI providers ability to channel patients to
hospitals. This assumption may not hold in healthcare markets where insurer networks are
not very selective and their influence over patients choices is weak.
Data requirements
As usual, estimating the hospital demand function requires data on patient and hospital
characteristics and patient discharges. In addition to this standard dataset, the approach also
requires data on hospitals costs, revenues and charges to PMI providers, as well as
information about the structures of any hospital chains operating in the area of interest. The
data-collection burden of the method is considerable since a robust estimation of supply-side
features is likely to require using a large number of hospitals in the study. The original article
by Gaynor and Vogt (2003), for example, implements the model using data on 374 hospitals
and over 900,000 patients.
Precedent
No recorded precedent.

3.4

Other methods
Oxera encountered a number of other non-standard approaches in the literature which, while
not necessarily designed to define markets for competition investigations, can still be used to
gain an understanding of hospitals areas of operation and the competitive constraints they
are facing. These approaches are outlined in this section.

3.4.1

GP referral mapping
Cooper et al. (2010), one of the few UK studies, albeit not primarily focused on market
definition, constructs another interesting measure of competition in an empirical study based
on GP-centred radii. The geographic market is defined as a distance around a GP practice
which corresponds to the 95th percentile of distance travelled by a patient from this practice
to a hospital. HHI concentration is then measured for every GP-centred area and diagnosis

Oxera

25

Techniques for defining markets for

private healthcare in the UK

combination, for which the given hospital is in the choice set, and the results are aggregated
into a prevalence-weighted HHI for the whole hospital.33
3.4.2

Hospital HHI versus system HHI

This method draws on the disaggregated product market view proposed by Zwanziger et al.
(1994). For each hospital, this method calculates an HHI in each postcode from which it
draws patients, separately for each major type of care, and aggregates the results into a
prevalence-weighted hospital-HHI. Capps and Dranove (2004) extend this approach to
estimate the effects of mergers on concentration by calculating the same measure for a
system of hospitals taken together (eg, merging parties), and comparing the two HHIs to
analyse the increase in market concentration if hospitals act in a coordinated manner. This is
used to analyse effects of mergers on concentration, but does not provide clear cut-offs for
defining markets (or for assessing competitive impact) specifically.

3.4.3

Physician-based radii
Luft and Maerki (1984), cited in Morrissey et al. (1988), use an alternative approach to
market definition, based on physicians, not patients, willingness to travel to carry out
treatments in hospitals. They consider, for example, a fixed radius of 15 miles to define the
maximum distance a physician would be willing to travel. Morrissey et al. (1988) criticise this
approach, on the basis that even if physicians are limited to a particular set of hospitals, the
patients choice set simply consists of hospitalphysician pairs, and the distance they are
willing to travel to high-quality hospital-physician offerings need not be constrained by the
physician-centred radii. As a result, even when physicians are tied to a very local set of
hospitals, patient flows may exercise competitive pressures over a wider area.

33

The HHI stands for the HerfindahlHirschman Index. This index is used to measure the size of a firm relative to the industry
or the overall level of concentration in the industry as a whole.

Oxera

26

Techniques for defining markets for

private healthcare in the UK

Critical assessment of techniques and applicability to the UK

The literature review in section 3 shows that there is a spectrum of techniques available to
define the geographic market for PH. The techniques are characterised by different degrees
of theoretical soundness, complexity, data requirements and the extent to which they have
been tested empirically or have established precedent in court cases. The next step is to
consider which of these are suitable for the UK market.
The conclusion that emerges from the literature review is that the more recent techniques are
conceptually more appealing than the earlier ones, since they have been developed to
account for specific characteristics of the PH market such as heterogeneity of patients and
suppliers, lack of price sensitivity of patients, and the fact that competition between hospitals
takes place at a network level as well as at an individual hospital level.34
However, this theoretical appeal needs to be weighed against the need for extensive and
detailed data, on which these techniques tend to rely. Other relevant considerations for
whether a technique is suitable for the UK would be the complexity of the technique
(ie, whether it allows for the calculations to be carried out with sufficient ease within the given
timescale), conceptual suitability of the technique for the UK (ie, whether the underlying
assumptions of the model reflect the characteristics of the UK PH system), and existing case
practice.
In addition, the theory of harm that is being considered is likely to play an important role in
the choice of geographic market definition technique. Some techniques focus purely on
market definition, while others have been designed specifically to simulate the effects of a
merger. The latter models may therefore be appropriate in mergers, potentially putting less
weight on the market definition stage.
This section assesses the available techniques in order to determine which are most
applicable for use by the OFT or other parties in defining the geographic market in the UK.
The section proceeds as follows. Section 4.1 discusses the key features of the UK PH
market, since these determine the criteria for assessing the applicability of the techniques to
the UK market. Section 4.2 discusses the criteria against which the techniques are assessed.
Sections 4.3 to 4.5 present the assessment of the techniques against the criteria listed in
section 4.2. Section 4.6 makes a recommendation resulting from the assessment of the
techniques.

4.1

Key features of the UK PH system that affect market definition

To identify which techniques for geographic market definition are suitable for the UK PH
market, it is useful to consider some of the key features of how this market works, and how it
may be different from markets such as those in the USA and the Netherlands which have
received considerable attention in the literature. This section therefore summarises Oxeras
understanding of the main steps of the patient journey, and the roles of the main players
therein.
In most cases, the patient journey begins when a patient experiences symptoms and seeks
advice from a GP.35 In cases that require specialist knowledge or further investigation, the

34

Although the literature review shows that no single method so far has succeeded in capturing all of these market
characteristics.

35

The Opinion Leader (2011).

Oxera

27

Techniques for defining markets for

private healthcare in the UK

GP makes an initial diagnosis and formally refers the patient to a specific consultant at a
hospital or, in certain cases, simply to any relevant specialist in a selected hospital.
The extent of GP involvement in the selection of private or NHS referral, or in the choice of
consultant, varies from case to case, but evidence from patient surveys undertaken by the
OFT suggests that GPs have significant influence over the initial choice of the consultant and
the hospital.36 The GPs choice of private hospital PH facility is normally not led by the price
of treatment, especially for PMI-funded patients, but is primarily based on the reputation of
consultants and facilities, factors that are likely to favour information obtained through local
relationships.37
If further consultation with the consultant reveals that medical treatment is necessary, the
consultant plays an important role in the patients decision about a PH facility where
treatment occurs. Almost half of the consultants in the OFT survey indicated, for example,
that they never give a patient a choice between the different private facilities in which they
operate, and a majority of consultants responded that over 75% of the treatments they carry
out occur in the same private hospital in which they first see the patient.38
Following the treatment, the typical patient journey diverges between self-pay and PMIfunded patients. The self-pay patients are billed directly by the hospital for the price of
consultant services and the hospital facilities provided during the course of treatment. Due to
these out-of-pocket expenses it appears that these uninsured patients are more likely to be
price-sensitive in choosing consultants and hospitals, and are more likely to view NHS
hospitals as a competitive substitute, trading off the costs of being treated privately against
longer waiting times and potentially lower quality. Evidence collected as part of recent
merger investigations suggests that approximately 15% of private hospital patients are
currently self-pay (Office of Fair Trading, 2010).
For patients with PMI, the next stage of the patient journey involves minimal expenses (other
than any policy excess) if the treatment is carried out by an authorised consultant and PH
facility, since both the hospital and the consultant settle their costs directly with the PMI
according to pre-determined remuneration agreements. PMI-funded patientsamounting to
as much as 60% of private patientsare therefore unlikely to be sensitive to treatment prices
as long as the providers are within their PMIs network. In fact, the OFT patient survey
reveals very limited awareness of costs of treatment among PMI-funded patients.39 These
patients choices may be more likely to be driven by quality, travel times or recommendations
of GPs or consultants. Figure 4.1 summarises Oxeras understanding of the relationships
between the main parties in the market for a typical PMI-funded patients journey.

36
37
38
39

The Opinion Leader (2011), p. 20 and p. 42. GHK (2011), p. 24.

GHK (2011), p. 22.
GHK (2011), pp. 534.
The Opinion Leader (2011), p. 38.

Oxera

28

Techniques for defining markets for

private healthcare in the UK

Figure 4.1

Relationships between key players in the PH market

GP

GP makes diagnosis
and refers to consultant

Consultant refers to
hospital, and
produces services

Consultant

Joint supply of
services

Insured
consumers
Hospitals provide the
location, facilities, and staff
facilitating the production of
services by consultants

Ex ante: consumers pay

premium to be included
in health plan
Ex post: PMI authorises
and covers costs of
services

PMI

Ex ante: negotiate prices of

services; inclusion in PMIs
health plan
Ex post: PMIs pay hospitals
for delivered facilities

Private
hospital
Ex ante: negotiate conditions of being recognised (including charges)
Ex post: PMIs pay consultants for delivered services

Source: Oxera analysis.

The limited sensitivity of patients to prices due to the widespread use of PMI adds another
important feature to competition in the UK PH marketthe ex ante price negotiations
between hospitals and PMI providers. Evidence submitted in the GHGAbbey merger
investigation (Office of Fair Trading, 2010) shows that there is a small number of large PMI
providers (eg, BUPA and AXA) and large nationwide hospital chains (eg, GHG, Spire,
Nuffield) that negotiate remuneration agreements centrally, adding a potential national
dimension to competition in the UK PH market.
4.1.1

Emerging conclusions on the UK market

The review of the typical patient journey in the UK highlights several features that are
important to note in assessing the applicability of market definition techniques to the UK
market.
First, there are three types of market participant in addition to the hospital and the patient
that have a significant effect on the market: consultants, GPs and PMI providers. Whereas
the important roles played in hospital choice by PMI providers and consultants have been
noted in the literature on US markets before (eg, Federal Trade Commission and Department
of Justice, 2004; Capps et al., 2003), the UK appears to be different in terms of the central
role of GPs as gatekeepers and traffic controllers for private care. Their involvement is
particularly significant in light of the survey findings that consultants tend to treat private
patients in the hospitals where they first see them.
Second, the UK appears to be unique in that the public healthcare sector (NHS) exists
alongside the PH market. NHS private patient units (PPUs) and the existence of a free public
healthcare service in practice may provide a competitive constraint on private hospitals, at
least with respect to self-pay private patients.
The effective separation of the NHS and the individual network of private hospitals
significantly restricts the availability of statistical data in the UK for the purposes of
undertaking market definition (as PH is outside the data reporting requirements applied to the
NHS). The majority of the more recent and more complex market definition methodologies
that require highly granular data were developed and applied in the USA and the
Netherlands, where private hospitals are the core of the healthcare industry and are subject
Oxera

29

Techniques for defining markets for

private healthcare in the UK

to reporting requirements from government organisations. In the UK, however, centralised

hospital supervision is largely focused on the NHS; consequently, no analogous central
patient discharge databases or standardised financial reporting systems appear to exist for
private care. This is therefore likely to significantly limit the choice of the geographic market
definition techniques that can currently be applied to the UK market for PH.
Finally, in the UK markets for private care there is significant functional separation (and often
separate billing of patients and PMI providers) between the contributions of a consultant and
of a private hospital to any given medical treatment. This feature of the market raises a
number of issues for product market definition, since the two contributions to treatments
significantly differ in terms of the possibilities for supply-side substitutability. In particular,
whereas consultant services are generally highly heterogeneous across treatments, many of
the core hospital services, such as overnight stay, food and operating theatre facilities, may
be the same. The interaction between these two components suggests that more clustered
product markets might be appropriate for purely private hospital services insofar as an
argument can be made that they are separable from consultants contributions in practice.

4.2

Criteria for assessing market definition techniques

The OFT may define relevant markets in a variety of contexts, including merger
investigations, investigations under the Competition Act 1998 and market studies. Given the
differences in the statutory timelines and information-gathering powers arising from different
types of investigation, it is unlikely that a single market definition method would be suitable
for all circumstances. It is therefore important to determine a set of features that would allow
all the available techniques to be compared in a structured way that gives weight to the most
important aspects of the case.
The following five criteria are used to assess the suitability of techniques for defining the
market for PH in the UK. These criteria have been chosen to allow a balanced assessment of
theoretical and practical considerations.

Theoretical underpinningany model is necessarily a simplification of reality;

however, to obtain reliable results, an appropriate method for defining markets in
hospital care needs to be in line with economic theory, internally consistent, and not
contradict the established facts about how agents in the relevant market behave in
practice.

Data requirementsthe issue of methodological difficulties in defining key variables for

many techniques (eg, hospital prices and profits) is a recurring theme in the literature.
As discussed in section 4.1, unlike the countries from which more sophisticated hospital
market definition techniques originate, the UK does not have a centralised private
patient discharge database, and centralised data on private hospital features and
finances is similarly hard to access. This makes data availability a key criteria for
selecting the appropriate geographic market definition technique for the UK.

Complexitythe available methods range from simple approximations to

methodologies that require significant time and highly specialised resources to evaluate
and interpret model performance and results. The models at the more complex end of
the spectrum are less likely to be useful for cases with short timescales for the analysis,
or where it is a preliminary stage of an investigation.

Conceptual suitability for the UK marketthis criterion tests whether a techniques

assumptions are in line with how the UK system operates. For example, techniques that
cannot adequately capture the mainly insurance-based model and the GP referral
system will be of little value for UK cases.

Oxera

30

Techniques for defining markets for

private healthcare in the UK

4.3

Established case practicethe experiences of competition authorities, courts and

regulators with using a particular technique are important for selecting appropriate
methodologies. These precedents, combined with ex post studies, also reveal whether
predictions of the methods were consistent with observed outcomes.

Earlier techniques for geographic market definition in PH

This section assesses the earlier techniques for geographic market definition in healthcare
markets against the five criteria described above.

4.3.1

Conceptual/theoretical underpinning
The literature review in section 3 highlights that the three earlier techniquesisochrones, CL
and EHhave weaker theoretical underpinnings than the more advanced models. The main
criticisms of these earlier models are as follows.

The models rely on arbitrary cut-off points that are not justified by the economic theory
and are therefore open to challengethe applications of catchment area analysis for
isochrones generally adopt a threshold of around 80% of patients; the EH test uses a
threshold of 10% or 25% for patient inflows and outflows; and the CL test uses an
arbitrary cut-off point to define contestable areas and predict switching rates.

The models appear not to take into account the heterogeneity of hospitals and
patientsthe same isochrone size is used for different hospitals; and EH and CLs
contestable postcode approach assumes that the presence of some travelling patients in
area indicates that substitutable external hospitals are available even for the currently
loyal patients.

The models do not address the lack of price sensitivity of patientswhile isochrones do
not rely on the price changes, CL and EH both assume that patients flows respond to
increases in prices.

There are, however, practical solutions that could address these problems, at least to some
extent.
In relation to the arbitrary cut-off point, a practical solution might be to flex the cut-off
percentage, for example to between 70% and 90% for the isochrone analysis, to test whether
this makes a significant difference to the result. Only where this shows that the result of the
analysis is sensitive to the level of the cut-off threshold does the choice of this threshold
become problematic. For the CL analysis, the contestable postcode approach appears
unsatisfactory, despite its parsimony, since it requires arbitrary cut-offs to define contestable
areas and predict switching rates. A more empirically robust approach to calculating
estimated lossfor example, using carefully designed conjoint surveysmay avoid many of
these criticisms.40
If preliminary analysis indicates that the heterogeneity of patients and hospitals is likely to be
an issue then a practical solution would be to define the product markets in a more granular
fashion. Treatment-specific catchments can then be used in the isochrone analysis to
address patient heterogeneity. Similarly, different sizes of isochrones could be used for
different hospitals depending on their particular characteristics (such as size, range of
treatments offered, and location). This approach is used in the groceries market, where the
CC defines geographical markets using different-sized isochrones depending on the size of
the store (convenience, mid-sized or one-stop) and its location (urban or rural).41 Defining the
product market more granularly may also alleviate the same problem for the EH test. In fact,
40

Conjoint, or stated-preference, surveys ask patients to choose between a range of options multiple times, and alter the
parameters of choice in order to estimate elasticities of demand with respect to those characteristics.

41

Competition Commission (2008).

Oxera

31

Techniques for defining markets for

private healthcare in the UK

Varkevisser and Schut (2009) argue that disaggregating the product market may improve EH
results, although distortions from other types of patient heterogeneity will still remain.
Another criticism of the EH approach is the backward-looking nature of this test, which might
pose a problem in the context of a competition case that is forward-looking in nature.
However, the test can be used where the investigation focuses on historical events such as
antitrust investigations.
Overall, although the earlier methods may suffer from some conceptual drawbacks (in terms
of arbitrary cut-off points and not explicitly taking into account the heterogeneity of patients
and hospitals), it appears that the problems associated with the methods can be mitigated or
resolved.
4.3.2

Complexity of application
As a general rule, all the classical and other non-formal techniques for market definition are
relatively simple to understand and to apply, and appear feasible even in the Phase 1
investigation.

4.3.3

Data requirements
All three methods have moderate data requirements. The kind of data necessary for the
isochrone analysis and (in part) CL and EH should be held by merging hospitals, and may
therefore be accessible in the context of mergers. In the context of other investigations, the
relevant data may be obtained through a survey. In cases where patient contact details are
known, specific patients may be targeted. Where the contact details of the patients are
unknown, an alternative would be to conduct a nationwide telephone, online or postal survey,
targeting PH patients who have recently undergone private treatment.
The data on patient flows to hospitals other than those assumed to constitute a geographical
marketas is required in the EH and CL contestable postcode approachmay be more
difficult to obtain, since it may require discharge data from third-party hospitals or PMI
providers. For CL, an alternative source of information on likely patient diversions could
come from approaching patients directly by means of a survey. This may be an expensive
exercise, however, because the contact information of specific patients is not available in the
UK. A national survey may need to be carried out to obtain the data (with a low expected
response rate, since PH patients who have recently undergone treatment would need to be
identified). In a limited range of cases it may also be possible to survey patients directly
outside hospitals.

4.3.4

Suitability for the UK market

Fixed-radius and isochrone techniques have previously been used in the UK in the context of
merger cases. For example, in GHG/Abbey, the OFT used an isochrone of 30 minutes
around each merging hospital to determine the local areas in which the merging parties were
likely to be rivals. That analysis did not identify any reasons why the techniques could not be
applied in the UK.
The prevalence of PMI and GP referrals in the UK (leading to a lack of patient response to
changes in prices of treatments, and the payer problem) can be seen as potential barriers to
the application of the CL analysis and EH test to the UK. For the CL analysis, this problem
can be overcome by using other dimensions such as quality of service, and waiting times can
be hypothetically flexed to gauge the likely reaction of patients in an alternative to a
hypothetical price rise. However, flexing dimensions other than price leads to further
questions around the appropriate degree of flexing to approximate a 5% price rise, whether
the dimension being flexed is actually important to consumers, and whether flexing non-price
dimensions is realisticie, whether a hospital faced with reduced competition would be
expected to flex quality or waiting times, or would simply negotiate a more lucrative deal with
PMI providers. In principle, this quality-driven interpretation of patient flows to hospitals may
also be applied to EH, albeit with the same methodological issues. Despite these

Oxera

32

Techniques for defining markets for

private healthcare in the UK

improvements, none of basic methods can capture the second component of the payer
problemthe price competition between hospitals and PMI providers.
4.3.5

Established case practice

As summarised in Table 3.1 in Appendix 3, all of the early techniques (EH, CL and
isochrones) have been used in courts for contested mergers. CL and isochrones are also
standard techniques that have been used in other industries in a variety of contexts.42

4.4

More recent techniques for geographic market definition

This section assesses the more recent techniques for geographic market definition against
the five criteria set out in section 4.2.

4.4.1

Theoretical underpinnings
All four MSM modelstime elasticity, willingness to pay, the structural MSM model, and
competitor sharedeliver large theoretical benefits by capturing the underlying
characteristics of the PH markets to a greater extent than the earlier techniques. In particular,
by reflecting the heterogeneity in sizes of geographic markets for hospitals, these models
deliver results consistent with the recent empirical finding that geographic markets in urban
areas can be very narrow (eg, Dafny, 2009).
The time-elasticity and willingness-to-pay approaches recognise that patients do not pay for
treatment directly, but pay for it through their PMI. The willingness-to-pay approach also has
the significant advantage of reflecting the option demand nature of the market in
circumstances where PMI providers are able to undertake selective contracting.43 However,
both approaches are sensitive to the underlying assumptions. The results of the timeelasticity approach are fairly sensitive to the assumption about the relationship between time
elasticity and price elasticity, which is not well-established. In the willingness-to-pay analysis,
it is necessary to define the counterfactual choices that PMI providers would have made in a
scenario in which a hospital was dropped from a network, which makes this method difficult
to apply in practice and vulnerable to challenge.
The GaynorVogt structural model and the competitor share approach both attempt to model
more realistic competitive behaviour between private hospitals. The GaynorVogt structural
model recognises that hospitals operate in large chains, which has major implications for
coordinated price-setting and for some of the diverted demand being captured by other
members of the chain in case of a price increase. The benefits of the competitor share
approach are more relevant to merger investigations, as this method is sensitive to the
potential differences in market power that a set of hospitals might have over a particular
insurer or for a specific treatment.
The disadvantage of both approaches is that they rely on the assumption that patients are
price-sensitive to some extent. This is implausible in many contexts where PMI providers are
unable to channel patients to hospitals, and difficult to implement in empirical studies due to
data constraints.
The GP-centred radii approach, although used in empirical research on the UK hospital
market, is not a formal market definition technique and has not been explored in the literature
to any great extent. Conceptually, defining markets based on GP-centred radii that capture
the overwhelming majority of referrals in the practice does not have clear economic
foundations. However, recent survey evidence shows that patients are often not aware of the
42

See footnote 10 for examples of the use of isochrones in a variety of retail contexts. For examples of the use of CL, see
Niels, Jenkins and Kavanagh (2011), chapter 2.
43

The extent to which this assumption is an advantage depends on the structure of the market. If the assumption mirrors the
actual structure of the market of interest, it can be a significant advantage, although in cases without selective hospital
contracting the willingness-to-pay model structure may result in significant distortions.

Oxera

33

Techniques for defining markets for

private healthcare in the UK

prices of treatments, and do not have sufficient information about the availability and quality
of the treatments available to them.44 Patients therefore often rely on their GPs when making
a choice of hospital. The breakdown of referral patterns by treatment captures indirectly,
albeit imperfectly, the heterogeneity of patients travel preferences across treatments and
may alleviate (but not fully resolve) the silent majority fallacy problem for patient flow-based
approaches. Furthermore, constructing hospitals catchment areas practice by practice can
directly identify areas where two hospitals of interest are directly competing for patients or
GP referrals. Additional empirical evidence on this approach would be desirable in order to
explore whether it can provide an alternative to the other four, more complex, techniques.
Overall, all five methods build on the earlier approaches in that they aim to capture specific
characteristics of the market for PH such as patient and hospital heterogeneity, the lack of
price sensitivity of PMI-funded patients, and the fact that GPs may have a role in a patients
choice of hospital. However, none of these approaches successfully incorporates all of the
characteristics of the market.
4.4.2

Complexity of application
In contrast to the earlier methods used to define the geographical market, the four MSM
approaches are very complex, requiring substantial data collection, cleaning and analysis, as
well as specialist econometric skills to calculate and interpret the results. These approaches
define geographic markets by comparing the simulated effects of hypothetical mergers
between subsets of hospitals, which can quickly become cumbersome when considering
large numbers of hospitals or many hypothesised markets at once. As a result, these
methods are likely to be appropriate only in the case of in-depth targeted investigations.
In comparison with the MSM approaches, the GP-centred radii method is more feasible to
implement. Calculating referral radii for each GP practice and treatment group and
constructing the catchment areas for each hospital require considerable data analysis and
aggregation, but this procedure is not theoretically complex.

4.4.3

Data requirement
All the approaches are data-intensive and require substantial amounts of detailed information
on individual patients (age, gender, diagnosis, location in relation to the hospital) and hospital
characteristics (quality, teaching status). Unlike in the USA and the Netherlands, where this
approach has been applied, the UK authorities do not collect this data for patients receiving
treatment in private hospitals. The data requirement to estimate the model exceeds the
samples that can be achieved using surveys; the time-elasticity analysis used in the NMa
investigation of the HiversumGooi-Noord (2005) merger used a national database of over
800,000 patients, although academic research has been carried out on samples with as few
as 5,400 patients (NMa 2005; Varkevisser et al., 2010). If available, however, data from a
large PMI may be sufficient for time-elasticity regressions, and there is precedent of PMI data
being used in academic time-elasticity research in the Netherlands (Varkevisser et al., 2010).
In addition to individual patients data, some of the more recent approaches require data on
hospitals and the overall market. For example, the willingness-to-pay approach requires
detailed data on insurer network coverage, as well as data on the profitability of all hospitals
in the area of interest. Hospital profit data, disaggregated by insurer, is difficult to obtain
(especially for third-party hospitals in the affected area). The GaynorVogt structural model
requires detailed price, profitability and cost data across a range of hospitals, which is
unlikely to be available with sufficient breadth and granularity. The competitor share
approach requires highly granular prices on specific treatments set by each hospital for each
specific PMI provider. Considering that remuneration agreements between hospitals and PMI
providers in the UK appear to be a mix of granular and per-day rates, these prices may not
exist. Even if they do, third-party hospitals in the investigated area and PMI providers would
have little incentive to disclose them. As a result, this method is unlikely to be feasible.
44

The Opinion Leader (2011); GHK (2011).

Oxera

34

Techniques for defining markets for

private healthcare in the UK

The GP-centred radii approach requires data on private referrals, which is not centrally
collected in the UK. However, given that detailed data is available on NHS referrals, this may
be used to infer the distance for private referrals. This would depend on the assumption that
the distance to which GPs tend to refer does not differ between private and public referrals.
NHS referral data can be used to construct a radius containing an effective choice set for
patients in the area (because GPs play a very large role in selecting consultants). This may
help to circumvent a lack of availability of patient data.
Overall, all the more recent methods have high informational requirements. For some
methodsGP-centred radii and time elasticityit may be possible to collect the required
data, but such an exercise would be expensive and time-consuming, and might require the
involvement of third parties with limited incentive to disclose. For other methods, such as the
GaynorVogt competitor share approach and, even more so, willingness to pay, the data
might not be feasible to collect in the UK.
4.4.4

Suitability for the UK market

In light of the majority of PH patients in the UK being PMI-funded, the time-elasticity and
willingness-to-pay approaches, which recognise that patients do not directly pay for their
treatments, may be the most appropriate for the UK. The general price-less framework of the
time-elasticity model can be very useful if an investigation requires NHS hospitals to be
added into a patients choice set to study the competitive constraint from the public sector.
On the other hand, the benefit of the willingness-to-pay approach is that it is the only method
to recognise explicitly the PMI providerhospital bargaining aspect of competition. This
aspect is particularly relevant in the UK, where the majority of private patients have PMI.
However, it does require insurer networks be to viable even if they do not cover all available
hospitals, which is not the case in the Netherlands. The UK appears to satisfy this
assumption, but it is not clear whether the practice of restricted insurer networks is
sufficiently widespread in the UK to be the main driving force of the competitive dynamics in
the market.
For both the GaynorVogt structural model and the competitor share model, the assumption
of sensitivity of patients to treatment prices does not appear to be consistent with the fact
that the majority of PH patients in the UK are PMI-funded. However, both models also have
attractive features in relation to the UK market for PH. For example, the GaynorVogt
structural model has the advantage of explicitly modelling the effects on competition of the
large hospital chains, which are a significant feature of the UK hospital market (eg, findings in
Office of Fair Trading (2010), Competition Commission (2000) and the current OFT market
study). The disadvantage of both models is that they require extensive granular input data,
which is not centrally collected in the UK.
The benefit of the GP-centred radii method is that it captures the crucial role that GPs play in
the choice of consultants and hospitals in the UK.45 Relying solely on the GP data, however,
does not take into account the fact that a consultant can treat a particular patient in any of a
number of hospitals where they practise, although recent evidence indicates that the majority
of consultants treat patients in the hospital where the first consultation occurs.46 The GPcentred radii technique may not be as suitable for self-pay patients, some of whom might
play a more active role in choosing their hospital and consultant, or might approach
consultants directly without referral. However, this approach has not been considered to a
great extent in the literature since the UK is unique with regard to the role of the GP in the
selection of consultants and hospitals. Given the potential importance of this method for the
UK, it would be interesting to examine this analysis in more detail empirically by way of a
survey. This would determine whether the method could offer a suitable alternative to the
other, more complex methods.

45
46

The Opinion Leader (2011), pp. 20 and 42.

GHK (2011), pp. 534.

Oxera

35

Techniques for defining markets for

private healthcare in the UK

4.4.5

Established case practice

In general, there is limited precedent for the more recent geographic market definition
methods being tested in practice.
On occasion, methods broadly similar to time-elasticity analysis have been used in
competition investigationseg, in the Netherlands (NMa, 2005). Outside hospital markets,
MSMs similar to the GaynorVogt approach have been used in some competition
investigations (eg, Volvo-Scania),47 but this does not provide sufficiently robust precedent
since the MSMs have been adapted significantly to be suitable for markets.
There is also no established case practice of using GP radii in competition investigations.

4.5

Other techniques to measure aspects of hospital competition

Given the importance of referrals to a specific consultant and the tendency of consultants to
treat patients in hospitals where they are first seen for an appointment, and tendency to treat
from one main facility, it may be appropriate in the UK to consider geographic market
definitions based on consultants willingness to travel. This can be captured using the
physician-based radii approach discussed in section 3, or physician-based isochrones. In
either case, a radius of a particular distance or travel time can be constructed around
consultants primary NHS hospitals, to measure their willingness to travel to private hospitals.
Using the radii and data on NHS hospital locations, a pool of consultants available to each
hospital in the area can be calculated. Moreover, this can identify the extent to which
hospitals of interest compete for consultants or, for example, appear to be locked out of
providing a particular type of care owing to the unavailability of relevant specialists.
This approach, like most radii, is relatively simple and does not have firm theoretical
foundations. Nonetheless, it provides valuable insights into one of the main steps of the UK
private patients journey, and is relatively straightforward to implement in the UK, especially
in narrower inquiries. This is because the method requires limited data: the locations of NHS
hospitals, lists of consultants in each hospital, and lists of consultants practising in all private
hospitals in the area. An estimated willingness-to-travel radius is also needed, which can be
estimated by means of a survey. There is also some established precedent from the UK
competition investigations for looking at consultants working patterns, although not
specifically in terms of willingness to travel. For example, in the BUPACHG investigation,
the CC estimated the hospitals share of the consultant market by calculating, for all
consultants employed by each hospital, what proportion of their treatment was carried out in
the hospital of interest instead of in its competitors in which the same consultants also
practised (Competition Commission, 2000).

4.6

Conclusions and recommendations

This section presents a table that allows a simple, high-level comparison of the different
techniques and how they perform against the criteria set out in section 4.2. Based on the
analysis in sections 4.34.5, a number of recommendations are then made.
Overall, the comparative assessment of the techniques reveals that there is a trade-off
between theoretical soundness and the feasibility of applying a technique in practice. As can
be seen from Table 4.1, there is no single technique that scores highly on every one of the
suitability criteria set out in section 4.2. The earlier techniques tend to score less on the
theoretical underpinnings but more on ease of application, data requirements, and
established case practice. The more recent techniques tend to score more on the theoretical
underpinnings but less on the other criteria.

47

European Commission (2000).

Oxera

36

Techniques for defining markets for

private healthcare in the UK

Table 4.1

Assessment of techniques in the context of the UK PH market1

Technique
Isochrone

Theoretical
soundness

Not complex

Data available
in the UK

Established
case practice

Conceptual
suitability
for the UK

99

GP-radii

99

Consultant
isochrone

99

ElzingaHogarty

9/3

99

99

9/5

Critical loss

Time elasticity

9/3

Willingness to pay

99

Competitor share

99

GaynorVogt

1
2
Note: The scores for each criterion range from 99(ideal fit) to xx (major failings identified). Data on NHS GP
referrals is publicly available and could be used as a proxy (if it is established that NHS and private referral
patterns are similar); alternatively, direct GP surveys may be an option. 3 In principle, the patient-level data
required for the time-elasticity analysis and calculation of patient outflows in EH can be held by PMI providers,
4
although access to it may be restricted in many circumstances. The contested zip code CL approach requires
similar data on patient outflows to EH, and is therefore equally difficult to implement; survey-based estimates of
actual loss may be more feasible to obtain. 5 The standard CL analysis using 5% price increases is not
conceptually suitable in the UK due to the prevalence of price-insensitive insured patients; however, time- or
quality-based alternatives may be more applicable.

Based on this critical review of the literature and merger cases, the following conclusions can
be drawn.
1)

Advanced techniques based on merger simulation are likely to be useful in the UK only
in rare cases, where data availability is very good and the competition authority has the
resources/capacity and time to undertake advanced analysis.

2)

In light of the conceptual appeal of the more complex techniques and the fact that the
current level of data does not allow for their application, it may be desirable to put in
place measures that encourage the recording and storage of the data required for these
more advanced techniques, so that they could be used in competition cases.

3)

Earlier techniques are appropriate in many circumstances where the time or budget
available for analysis is more limited and where information is unobtainable. If the
techniques are applied in the right way, it is possible to avoid, or at least mitigate, the
concerns levelled at these techniques in the academic literature.

4)

Within the set of earlier techniques, ElzingaHogarty and critical loss are likely to be less
appropriate than isochrone-type measures based on catchment area analysis. In the
case of ElzingaHogarty, the lack of a central data source of patient locations and
treatment makes its application more difficult in the UK than in some other countries.
Therefore, the additional benefit from applying this technique compared with the
isochrone-type measures (in terms of increased precision) may be outweighed by the
burden of the additional data requirements. In the case of critical loss, the insurancebased model in the UK creates a fundamental hurdle (as patients are not pricesensitive) that is unlikely to be fully overcome. In cases involving PH facilities where
there are fewer PMI-funded patientssuch as those specialising in elective cosmetic
surgerycritical loss would be more appropriate.

5)

When applying catchment area (isochrone or fixed-radius) techniques, the issues raised
above should be borne in mind. As far as possible, it may be sensible to avoid

Oxera

37

Techniques for defining markets for

private healthcare in the UK

assessments that bundle together treatments or groups of patients with systemically

different willingness to travel. Assessments should also take into account the potential
heterogeneity of PH facilities, so it may be appropriate to apply different-sized
isochrones to different types of PH facility.
6)

As far as possible, given the significance of the impact of product market definition on
geographic market definition, when applying the catchment area techniques, empirical
analysis should be undertaken to examine the difference in travel times for patients
undergoing different types of treatment included in the product bundle in order to
prevent bundling together patients with different willingness to travel.

7)

In the specific case of merger analysis, it might often be more appropriate to focus more
directly on the likely competitive effects of the transaction rather than on precisely
defining the market and calculating market shares. The local nature of competition
makes the direct assessment of competitive effects in specific local areas attractive.
Assessments that take into account the fact that demand is not symmetric around a PH
facility should be used where possible, such as those that use postcode-based patient
discharge data to build a topographic picture of demand for a particular PH facility.48

8)

For Competition Act cases (those involving suspected abuse of dominance or

anti-competitive agreements), this direct analysis is less likely to be appropriate. In such
cases, it may be necessary for the OFT to form a more precise definition of the relevant
market (although in some cases the OFT may be able to apply threshold tests to
different candidate markets in order to establish that the relevant legal test is met).

9)

In the case of market investigations, a precise market definition is less essential, but the
nature of the analysis, which must cover many hundreds of local areas, means that a
hospital-by-hospital analysis of local competition is unlikely to be useful or feasible.

10) The literature refers to, but does not explore in detail, some of the less common
approaches, such as GP- and consultant-based radii, and only limited empirical
evidence is available on these techniques. Given that the more advanced techniques
appear to be less appropriate for the UK due to data availability issues, it may be
desirable to explore these techniques empirically to determine whether they could
provide a suitable alternative to the more complex methods used elsewhere.

48

See, for example, Office of Fair Trading (2008), Completed acquisition by Spire Healthcare Limited of Classic Hospitals
Group Limited, para 20.

Oxera

38

Techniques for defining markets for

private healthcare in the UK

A1

Bibliography

Ashenfelter, O., Hosken, D., Vita, M. and Weinberg, M. (2011), Retrospective analysis of
hospital mergers, International Journal of the Economics of Business, 18:1.
Baker, L.C. (2001), Measuring competition in health care markets, Health Service Research
36:1, pp. 22351.
Berry, S., Levinsohn, J. and Pakes, A. (2004), Differentiated Products Demand Systems
from a Combination of Micro and Macro Data: The New Car Market, Journal of Political
Economy, 112:1, pp. 68105, February.
Blackstone, E.A. and Fuhr, J.P. (1992), An Antitrust Analysis on Non-Profit Hospital
Mergers, Review of Industrial Organization 8:4, pp. 47390.
Canoy, M. and Sauter, W. (2009), Hospital mergers and the public interest: Recent
developments in the Netherlands, TILEC Discussion Paper No. 2009035, Tilburg
University.
Capps, C.S. and Dranove, D. (2004). Hospital consolidation and negotiated PPO prices,
Health Affairs, 23:2, pp. 17581.
Capps, C.S., Dranove, D., Greenstein, S. and Satterthwaite, M. (2001), The silent majority
fallacy of the ElzingaHogarty criteria: a critique and new approach to analyzing hospital
mergers, Working Paper 8216, National Bureau of Economic Research (NBER), Cambridge
(Mass.).
Capps, C.S., Dranove, D., Greenstein, S. and Satterthwaite, M. (2002), Antitrust policy and
hospital mergers: recommendations for a new approach, Antitrust Bulletin, 47:4,
pp. 677714.
Capps, C.S., Dranove, D. and Satterthwaite, M. (2003), Competition and Market Power in
Option Demand Markets, Rand Journal of Economics, 200334, pp. 73763.
Co-operation and Competition Panel (2011), Merger of Nuffield Orthopaedic Centre NHS
Trust and Oxford Radcliffe Hospitals NHS Trust, September 30th.
Competition Commission (2000), British United Provident Association Limited and
Community Hospitals Group Plc: A report on the proposed merger; and British United
Provident Association Limited, Salomon International LLC and Community Hospitals Group
Plc; and Salomon International LLC and Community Hospitals Group Plc: A report on the
existing mergers. Inquiry report Cm 5003, December 7th.
Competition Commission (2003), Merger references: Competition Commission Guidelines.
Competition Commission (2008), The supply of groceries in the UK market investigation,
April 30th.
Connor, R.A., Feldman, R.D. and Dowd, B.E. (1998), The Effects of Market Concentration
and Horizontal Mergers on Hospital Costs and Prices, International Journal of the
Economics of Business, 5:2.
Cooper, Z., Gibbons, S., Jones, S. and McGuire, A.l. (2010), Does hospital competition save
lives? Evidence from the English NHS patient choice reforms, LSE Health working papers.

Oxera

39

Techniques for defining markets for

private healthcare in the UK

Dafny. L. (2009), Estimation and Identification of Merger Effects: An Application to Hospital

Mergers, Journal of Law and Economics, 52:3, pp. 52350.
Dranove, D. and White, W. (1994), Recent Theory and Evidence on Competition in Hospital
Markets, Journal of Economics and Management Setting, pp. 169209.
Elzinga, K. and Swisher, W. (2011), Limits of the ElzingaHogarty Test in Hospital Mergers,
International Journal of the Economics of Business, 18:1.
European Commission (2000), Commission Decision of 14.03.2000 declaring a
concentration to be incompatible with the common market and the functioning of the EEA
Agreement (Case No COMP/M. 1672 Volvo/Scania), March 14th.
Farrell, J., Pautler, P.A. and Vita, M.G. (2009), Economics at the FTC: Retrospective Merger
Analysis with a Focus on Hospitals, Review of Industrial Organization, 35:4.
Federal Trade Commission (2005), In the matter of Evanston Northwestern Healthcare
Corporation: Initial Decision, Docket No. 9315.
Federal Trade Commission and Department of Justice (1992), Horizontal merger guidelines,
Issued April 2nd 1992, revised September 2010.
Federal Trade Commission and Department of Justice (2004), Improving Health Care: A
Dose of Competition: A Report by the Federal Trade Commission and the Department of
Justice, Washington DC, July.
Federal Trade Commission and Department of Justice (2010), Horizontal merger guidelines.
Gaynor, M., Kleiner, S.A. and Vogt, W.B. (2011), A Structural approach to market definition
with an application to the hospital industry, Working Paper 16656 NBER working paper
series.
Gaynor, M., and Town, R.J. (2011), Competition in healthcare markets, Working paper
17208, NBER working paper series.
Gaynor, M. and Vogt, W.B. (1999), Working Paper 7112, NBER working paper series.
Gaynor, M. and Vogt, W.B. (2003), Competition Among Hospitals, Working Paper 9471,
NBER working paper series.
GHK (2011), Programme of Research Exploring Issues of Private Healthcare Among
General Practitioners and Medical Consultants, August.
Kemp, R. and Severijnen, A. (2010), Price effects of Dutch hospital mergers. An ex-post
assessment of hip surgery, NMa Working Paper 2, Netherlands Competition Authority.
Meghrigian, A. (2003), Physician product and geographic market definition, presentation to
the Joint FTC/DOJ Hearings on Health Care and Competition Law and Policy, California
Medical Association.
Morrisey, M., Sloan, F. and Valvona, J. (1988), Defining Geographic Markets for Hospital
Care, Law & Contemporary problems, 165.
NMa (2005), Ziekenhuis Hilversum Ziekenhuis Gooi-Noord, Zaak 3897.
Niels, G., Jenkins, H. and Kavanagh, J. (2011), Economics for Competition Lawyers, Oxford
University Press, March.

Oxera

40

Techniques for defining markets for

private healthcare in the UK

OBrien, D.P. and Wickelgren, A.L. (2003), A Critical Analysis of Critical Loss Analysis, US
Federal Trade Commission.
Office of Fair Trading (2008a), Completed acquisition by General Healthcare Group of
assets of Nuffield Hospital.
Office of Fair Trading (2008b), Completed acquisition by Spire Healthcare Limited of Classic
Hospitals Group Limited.
Office of Fair Trading (2010), Completed acquisition by General Healthcare Group of control
of four Abbey hospitals and de facto control over Transform Holdings Limited previously part
of the Covenant Healthcare Group, Merger Decision. No. ME 4560/10, October 11th.
Simpson, J. (2001), Geographic markets in hospital mergers: a case study, International
Journal of the Economics of Business, 10:3, pp. 291303.
Sorensen, A. (2003), Insurerhospital bargaining: Negotiated discounts in post-deregulated
Connecticut, Journal of Industrial Economics, 51:4, pp. 46990.
Tenn, S. (2011), The price effects of hospital mergers: a case study of the SutterSummit
transaction, International Journal of the Economics of Business, 18:1.
The Opinion Leader (2011), The Patient Journey, August.
Town, R. and Vistnes, G. (2001), Hospital competition in HMO networks, Journal of Health
Economics, 20, pp. 73353.
Varkevisser, M., Capps, C.S. and Schut, F. (2008), Defining hospital markets for antitrust
enforcement: new approaches and their applicability to the Netherlands, Health Economics,
Policy and Law.
Varkevisser, M. and Schut, F. (2009), Hospital merger control: An international comparison,
iBMG Working Paper W2009.01, Erasmus University Rotterdam.
Varkevisser, M. and van der Geest, S.A. (2007), Why do patients bypass the nearest
hospital? An empirical analysis for orthopaedic care and neurosurgery in the Netherlands,
European Journal of Health Economics, 8:3, pp. 28795.
Varkevisser, M., van der Geest, S.A. and Schut, F.T. (2010), Assessing hospital competition
when prices don't matter to patients: the use of time-elasticities, International Journal of
Health Care Finance and Economics, 10, pp. 4360.
Vogt W. and Town, R. (2006), How has hospital consolidation affected the price and quality
of hospital care?, Research Synthesis Report from the Robert Wood foundation.
Zwanziger, J., Melnick, G. and Eyre, K.M. (1994), Hospitals and antitrust: defining markets,
setting standards, Journal of Health Politics, Policy and Law, 19:2, pp. 42347.

Oxera

41

Techniques for defining markets for

private healthcare in the UK

A2

Summary of reviewed literature

Techniques
discussed

Techniques
used

Literature review

EH, critical loss,

option demand
(willingness to pay)

None

The article gives an overview of recent hospital merger

challenges lost by the FTC, citing market definition and
not-for-profit hospital defence as the main reasons. It
also summarises two retrospective studies of the effects
of the consummated mergers, which suggest that the
markets defined by the EH test are too broad.

Measuring Competition
in Health Care Markets

Analytical paper

Geographical
boundaries, fixedradius,
EH

None

The study reviews key issues and data sources of

hospital competition measurement in the USA for
researchers and policy-makers. Data scarcity, careful
product and geographic market definitions and regard for
endogeneity in econometric studies are identified as the
main problems for competition studies.

Blackstone and Fuhr (1992)

An Antitrust Analysis on
Non-Profit Hospital
Mergers

Analytical paper

None

None

The paper presents four qualitative case studies of

contested not-for-profit hospital mergers in the USA,
concluding that the relevant geographic market for the
mergers depends on the level of care: small local
markets for simple hospital care and significantly wider
markets for complex operations. It also finds that not-forprofit status, by itself, does not change the effects on
competition of hospital mergers.

Canoy and Sauter (2009)

Hospital mergers and the

public interest: Recent
developments in the
Netherlands

Analytical paper

None

None

The paper discusses the experience of hospital mergers

in the Netherlands, focusing on issues of market
definition, vertical integration and efficiency defence. The
authors criticise the competition authorities for failure to
define robustly geographic markets in past mergers and
welcome development of more robust structural market
definition approaches.

Author

Title

Type of article1

Ashenfelter, Hosken, Vita

and Weinberg (2011)

Retrospective Analysis of
Hospital Mergers

Baker (2001)

Oxera

42

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Author

Title

Type of article1

Techniques
discussed

Techniques
used

Capps and Dranove (2004)

Hospital consolidation
and negotiated PPO
prices

Ex post merger study

None

Hospital HHI

The paper studies confidential data from hospitals

contracts with preferred provider organisations to identify
the effects of recent US hospital mergers on prices. It
finds that, in most cases, consolidation enabled hospitals
to increase prices.

Capps, Dranove and

Satterthwaite (2003)

Competition and Market

Power in Option Demand
Markets

New technique

Option demand
(willingness to pay)

Option demand
(willingness to
pay)

The authors develop a new willingness-to-pay approach

to defining geographic markets for hospital care, where
the price-sensitive decision-makers are often insurers,
not individual patients. The new measure is then applied
to hospitals in the San Diego area, suggesting that in
some cases suburbs with as few as two to three
hospitals may be a well-defined market.

Capps, Dranove,
Greenstein and
Satterthwaite (2001)

The silent majority fallacy

New technique

EH, competitor
share, time
elasticity

Competitor
share, time
elasticity

The paper articulates one of the main theoretical

challenges to the EH test: the silent majority fallacy. It
proceeds to develop two alternative market definition
approaches for contexts when both patients and
hospitals are heterogeneous. Merger simulations using
the two new approaches show that the silent majority
fallacy can lead the EH test to significantly overstate
geographic markets.

of the ElzingaHogarty
criteria: a critique and
new approach to
analyzing hospital
mergers

Main relevant findings

Capps, Dranove,
Greenstein and
Satterthwaite (2002)

Antitrust policy and

hospital mergers:
recommendations for a
new approach

Empirical study

Competitor share,
time elasticity,
option demand
(willingness to pay)

Competitor
share, time
elasticity, option
demand
(willingness to
pay)

The paper provides an overview of the three new

methodologies to market definition in mergerstimeelasticity, competitor share and option demand
approachesand illustrates all three by providing
simulation results for hypothetical mergers of San Diego
hospitals. Simulations using all three approaches
produce very similar qualitative predictions and identify
local geographic markets in San Diego suburbs.

Connor, Feldman and

Dowd (1998)

The Effects of Market

Concentration and
Horizontal Mergers on
Hospital Costs and
Prices

Ex post merger study

None

Geographical
boundaries

The study investigates the effects of market

concentration and hospital mergers on hospital costs and
prices. Overall, hospital mergers are found to reduce
hospital costs and, in turn, lower prices to customers.
There is some evidence that price reductions are smaller
in more concentrated markets.

Oxera

43

Techniques for defining markets for

private healthcare in the UK

Techniques
discussed

Techniques
used

Empirical study

Fixed-radius,
isochrones, EH,
GP-centred radii

GP-centred radii

NHS reforms introduced in 2006 have meant that public

hospitals had to compete for patients (largely on the
basis of published quality metrics) while the prices were
set. The authors use this as a natural experiment to
examine the effect of competition on the quality of care in
the UK, finding that, in this fixed-price setting, hospital
competition appears to improve patient outcomes.

Estimation and
Identification of Merger
Effects: An Application to
Hospital Mergers

Empirical study

Option demand
(willingness to
pay), fully structural
approach

None

The study explores the impact of a hospitals rivals

merging on that hospitals price, using rivals co-location
as an instrument. A merger by nearby rivals is found to
lead hospitals to increase prices by as much as 40%.
The findings suggest that markets for hospital care are
very local, far smaller than those typically considered in
courts and similar in size to predictions of the new
merger simulation models.

Dranove and White (1994)

Recent Theory and

Evidence on Competition
in Hospital Markets

Literature review

Fixed-radius, EH,
hospital HHI

None

The paper reviews the theoretical literature on

competition under imperfect information to define the
nature of hospital competition and offers empirical
evidence on the extent to which hospitals can comply
with the predictions of traditional IO literature on
competition. Both for-profit and not-for-profit hospitals
are considered.

Elzinga and Swisher (2011)

Limits of the Elzinga

Hogarty Test in Hospital
Mergers

Analytical paper

EH

None

This paper discusses the main methodological problems

in applying the EH test in the hospital merger context,
focusing on its failure to reflect heterogeneity and priceinsensitivity of hospital patients. It also examines FTC v.
Evanston (2007), in which the courts concluded that the
EH test was not applicable to hospital markets.

Farrell, Pautler and Vita

(2009)

Economics at the FTC:

Retrospective Merger
Analysis with a Focus on
Hospitals

Literature review

None

None

The article summarises the findings of three ex post

studies that explore the effects of mergers. This research
confirms that mergers between not-for-profit hospitals
can have anti-competitive effects and shows that
geographical markets for hospital care are more
localised than suggested by approaches traditionally
used in courts.

Author

Title

Type of article1

Cooper, Gibbons, Jones

and McGuire (2010)

Does hospital
competition save lives?
Evidence from the
English NHS patient
choice reforms

Dafny (2009)

Oxera

44

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Techniques
discussed

Techniques
used

Literature review

Geographical
boundaries, fixedradius,
EH, fully structural
approach

None

The paper summarises the main issues in the healthcare

competition, the related case law and empirical research
published up to1999. It highlights how optimality of
competition is affected by ways in which healthcare is
different from other industries. Based on the identified
differences, the authors outline a new structural
approach to evaluating the effects of hospital mergers.

Competition among
Hospitals

New technique

Fully structural
approach

Fully structural
approach

The authors propose a fully structural approach to

modelling the demand and supply sides of competition
between hospitals. This model is developed to explore
the effect of ownership type on hospital conduct and to
define geographical markets in hospital care more
accurately. The new model is also used to simulate
mergers among Californian hospitals, identifying some
firms whose consolidation creates significant local
market power.

Gaynor and Town (2011)

Competition in
healthcare markets

Literature review

Fully structural
approach, option
demand
(willingness to
pay), hospital HHI

None

The paper reviews empirical and theoretical literature on

markets for healthcare services produced between 2000
and 2011. It summarises the main empirical findings on
the effects of competition in healthcare services and
presents key theoretical models that best describe the
option demand nature of US hospital markets.

Gaynor, Kleiner, and Vogt

(2011)

A Structural Approach to
Market Definition with an
Application to the
Hospital Industry

Empirical study

EH, critical loss,

option demand
(willingness to
pay), fully structural
approach

EH, critical loss,

option demand
(willingness to
pay), fully
structural
approach

The paper describes the structural approach to hospital

market definition developed by Gaynor and Vogt (2003)
and compares its predictions to traditional approaches
used in courts and to the willingness-to-pay model. Both
advanced merger simulation methods are found to
deliver very similar market structure predictions. Critical
loss analysis and EH, however, are found to significantly
overstate the size of geographic markets.

Author

Title

Type of article1

Gaynor and Vogt (1999)

Antitrust and competition

in health care markets

Gaynor and Vogt (2003)

Oxera

45

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Author

Title

Type of article1

Techniques
discussed

Techniques
used

Kemp and Severijnen

(2010)

Price effects of Dutch

hospital mergers. An expost assessment of hip
surgery

Ex post merger study

EH, critical loss

None

The study examines the effects of two hospital mergers

in the Netherlands on the price of hip surgery and
patients travel patterns. It finds statistically significant
price increases after the Ziekenhuis Hilversum
Ziekenhuis Gooi-Noord merger, which was originally
delayed by the NMa due to failing the EH test, but
cleared upon further investigation. The study also finds
that patients may have overstated their willingness to
travel in response to price increases in the revealed
preference study conducted as part of the merger
investigation.

Meghrigian (2003)

Physician product and

geographic market
definition

Analytical paper

None

None

The paper does not analyse specific market definition

techniques. Instead, it argues for a lenient antitrust
approach to physicians, because their bargaining
position vis--vis healthcare insurance providers is said
to be very weak.

Morrisey, Sloan and

Valvona, J. (1988)

Defining Geographic
Markets for Hospital
Care

Empirical study

Fixed-radius,
geographical
boundaries, EH

EH

The authors apply the EH approach to define geographic

markets in hospital care. They find that hospital markets
for both rural and urban hospitals are much larger and
less concentrated than implied in earlier studies. On this
basis, the authors conclude that antitrust concerns in
hospital mergers are far less likely than conventionally
believed.

OBrien and Wickelgren

(2003)

A Critical Analysis of
Critical Loss Analysis

Analytical paper

Critical loss

None

The authors use a series of theoretical arguments to

criticise the standard applications of the critical loss
analysis in courts. The main criticisms relate to the
internal inconsistency of accepting large estimated
losses for high-margin firms and the failure to consider
cross-price elasticities.

Simpson (2001)

Geographic markets in
hospital mergers: a case
study

Ex post merger study

Critical loss
(contestable zip
code approach)

Critical loss
(contestable zip
code approach)

The paper investigates the assumption often made by

courts in the critical loss test that patients in contestable
zip codes would switch to other hospitals after a price
increase. Using the acquisition of Community Hospital in
1990, the author finds that in all but one area that the
court would have deemed contestable for the purposes
of the critical loss test, the market shares declined very
little as a result of price increases after the merger.

Oxera

46

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Author

Title

Type of article1

Techniques
discussed

Techniques
used

Sorensen (2003)

Insurerhospital
bargaining: Negotiated
discounts in postderegulated Connecticut

Empirical study

None

None

The paper finds that the ability of insurance providers to

obtain discounts from hospitals depends on their ability
to channel patients to other hospitals, and on their size.
Seen from another perspective, this ability also
influences the scope for hospitals to raise prices.
Hospitals can raise prices more easily when insurance
providers are not able to channel patients to other
hospitals.

Town and Vistnes (2001)

Hospital competition in
HMO networks

New technique

Option demand

Option demand

The authors develop an empirical framework that models

competition between hospitals and insurers, and
examines the effects of insurers selective networks on
hospital prices. The hypothetical merger simulations in
the study suggest that mergers between neighbouring
and closely substitutable hospitals can lead to significant
price increases, even in urban settings where there are
many other nearby hospitals.

US Federal Trade
Commission and
Department of Justice
(2004)

Improving Health Care: A

Dose of Competition: A
Report by the Federal
Trade Commission and
the Department of
Justice

Analytical paper

EH, critical loss,

option demand,
isochrones

None

The report provides an in-depth overview of the structure

of the US health industry and its implications for
competition law. The experiences of antitrust authorities
in challenging historical mergers, and specifically the
major role played by market definition, are discussed in
detail.

Varkevisser, Capps and

Schut (2008)

Defining hospital markets

for antitrust enforcement:
new approaches and
their applicability to the
Netherlands

Analytical paper

EH, critical loss,

competitor share,
time elasticity,
option demand
(willingness to pay)

None

Time-elasticity, competitor share and option demand

approaches are the main formal market definition
techniques that were developed to address theoretical
failings of the existing methodology. The applicability of
an advanced approach is found to critically depend on
the specifics of the market structure: how patients
choose hospitals and how hospitals contract with
insurers.

Oxera

47

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Techniques
discussed

Techniques
used

Analytical paper

EH, critical loss,

isochrones

None

The article describes the current practices and legal

precedents for private hospital merger control in the
USA, Germany and the Netherlands. It also presents
case studies of a landmark merger case for each
country. The main findings of the international
comparison are that geographic market definition is a
key vulnerability for merger challenges, and that
geographic markets for hospital care are small in
practice.

Why do patients bypass

the nearest hospital? An
empirical analysis for
orthopaedic care and
neurosurgery in the
Netherlands

Empirical study

None

None

Many Dutch patients bypass their nearest hospitals. The

choice to travel beyond the nearest hospital depends on
travel time and hospital quality. Patients are found to
have lower aversion to extra travel time for complex
treatments. The study concludes that both patient and
hospital heterogeneity should be taken into account
when assessing hospital substitutability.

Varkevisser, van der Geest,

and Schut (2010)

Assessing hospital
competition when prices
dont matter to patients:
the use of timeelasticities

Empirical study

Time elasticity

Time elasticity

The time-elasticity approach is applied to the Dutch

hospital markets using a dataset from a large insurer.
The paper explores factors that affect hospital choices
and simulates the effects of artificial increases in travel
times. Overall, all hospitals time elasticities are found to
be high, suggesting the existence of at least one close
substitute for each.

Vogt and Town (2006)

How has hospital

consolidation affected
the price and quality of
hospital care?

Literature review

Time elasticity,
competitor share
and option demand

None

The paper provides a summary of the research on

hospital consolidation to assess the likely effects of
hospital mergers on healthcare prices, costs and quality.
Overall, hospital mergers are found to increase prices by
5% or more, with competitors raising prices as well as
the merging parties. The findings regarding mergers
effects on quality and costs are inconclusive.

Zwanziger, Melnick, and

Eyre (1994)

Hospitals and antitrust:

defining markets, setting
standards

New technique

Geographical
boundaries, EH,
hospital HHI

None

The authors propose a highly disaggregated approach to

defining the product market, based on physician
specialties. The corresponding geographic markets are
defined by constructing a weighted HHI using patient
flow data. The article also argues that in the context of
selective insurer networks, the markets for hospitals are
very local in nature.

Author

Title

Type of article1

Varkevisser and Schut

(2009)

Hospital merger control:

An international
comparison

Varkevisser and van der

Geest (2006)

Oxera

48

Main relevant findings

Techniques for defining markets for

private healthcare in the UK

Note: 1 The broad types of papers identified in the table are defined as follows: literature reviews primarily provide a broad overview of articles on a particular topic; ex post merger
studies investigate effects of actual mergers, primarily using standard econometric techniques; empirical studies include all other quantitative studies on competition in hospital
markets; new technique articles introduce and test new market definition techniques; and analytical papers cover all other types of qualitative or theoretical discussions of hospital
mergers or broader issues in healthcare markets.

Oxera

49

Techniques for defining markets for

private healthcare in the UK

A3

Summary of reviewed cases

Table A3.1 Market definition precedents in hospital mergers and acquisitions1

Case

Year

GHGAbbey

Inova

2010

Country

UK

2008

USA

2008

UK

Methods used to define geographic markets

Evanston
FTC v. Evanston Northwestern

Medium: Market definition was undertaken by

the OFT and was not challenged

Cleared by the OFT

subject to undertakings

Hospitals

Isochrones

Isochrones

Patient catchment areas

Patient catchment areas

EH

No information

Medium: Market definition was undertaken by

the FTC and was not challenged

Hospitals withdrew from

the merger after the FTC
injunction

Isochrones

Patient catchment areas

Medium: Market definition was undertaken by

the OFT and was not challenged

Cleared by the OFT

Patient catchment areas

GHGNuffield

Merger outcome

Competition authority

FTC et al v. Inova Health System

Foundation et al
SpireCHG

Relevance of market definition to outcome

2008

UK

Isochrones

Isochrones

Medium: Market definition was undertaken by

the OFT and was not challenged

Cleared by the OFT

2007

USA

Observed post-merger price

increases

Patient catchment areas

Low: Evidence of substantial lessening of

competition after the merger; use of
EH explicitly rejected

Ex post merger challenge

by FTC upheld by the
court

EH

No information

High: Merger cleared due to inconclusive

evidence of the relevant geographic market

Cleared by the NMa after

Phase 2 investigation

Geographical borders

Geographical borders

Prohibited by the CC

Isochrones

Isochrones

Medium: Market definition was undertaken by

the CC and was not challenged

Critical loss

High: Merger cleared due to insufficient evidence

of the relevant geographic market

Challenge to the merger

overruled in court

Critical loss

High: FTC found to have failed to identify the

relevant geographic market

Challenge to the merger

overruled in court

Ziekenhuis Hilversum and

6
Ziekenhuis Gooi-Noord

2005

BUPACHG

2000

Netherlands

Isochrones

Time-elasticity approach
UK

Fixed radii
Sutter

1999

USA

Critical loss

California v. Sutter Health System

Poplar Bluff

EH

1998

USA

EH

FTC v. Tenet Health Care Corp.

Oxera

50

Techniques for defining markets for

private healthcare in the UK

Case

Year

Country

Methods used to define geographic markets

Relevance of market definition to outcome

Merger outcome

Long Island

1997

USA

Insurer testimonies

Patient catchment areas

High: DOJ found to have failed to identify the

relevant product and geographic market

Challenge to the merger

overruled in court

1996

USA

EH

EH

Low: Not-for-profit merger defence

Challenge to the merger

overruled in court

1995

USA

EH

Critical loss

High: DOJ found to have failed to identify the

relevant product and geographic market

Challenge to the merger

overruled in court

1995

USA

EH

EH

High: FTC found to have failed to identify the

relevant geographic market

Challenge to the merger

overruled in court

1994

USA

EH

EH

High: FTC found to have failed to identify the

relevant geographic market

Challenge to the merger

overruled in court

US v. Long Island Jewish Medical

Center
Grand Rapids
FTC v. Butterworth Health Corp.
Dubuque
United States v. Mercy Health
Services
Joplin
FTC v. Freeman Hospital
Ukiah
Adventist Health System/West

Sources: 1 This table is an extended version of US-focused Table 2 in Gaynor, Kleiner and Vogt (2011), which is a source for all data unless stated otherwise. 2 Office of Fair Trading
3
4
5
6
(2010). Elzinga and Swisher (2011). Office of Fair Trading (2008b). Office of Fair Trading (2008a). Varkevisser and Schut (2009). NMa (2005).

Oxera

51

Techniques for defining markets for

private healthcare in the UK

Park Central
40/41 Park End Street
Oxford OX1 1JD
United Kingdom
Tel: +44 (0) 1865 253 000
Fax: +44 (0) 1865 251 172
Stephanie Square Centre
Avenue Louise 65, Box 11
1050 Brussels
Belgium
Tel: +32 (0) 2 535 7878
Fax: +32 (0) 2 535 7770
Thavies Inn House
7th Floor
3/4 Holborn Circus
London EC1N 2HA
United Kingdom

www.oxera.com

Tel: +44 (0) 20 7822 2650

Fax: +44 (0) 20 7822 2651