1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Simulating the Effects of Regulatory Change in the UK Pubs Industry
Kim Warren
London Business School
Regents Park
London, NW1 4SA, UK
Tel: 071 262 5050 Fax: 071 724 7875
Abstract
A 1989 Monopolies Commission Inquiry into the supply of beer claimed that Tied retail distribution
by brewers’ ownership of retail outlets (pubs) restricted consumer choice, excluding would-be
competitors from brewing and retailing, and forcing up the retail price of beer. The inquiry
recommended reducing large brewers’ ownership of pubs. This was intended enable new entry into
retailing and production, reduce wholesale and retail beer prices, and extend consumer choice.
However, a 1993 Government review found that the outcome had been largely the opposite of these
aims, and a further serious consequence was the closure of large numbers of pubs and a sharp fall in
their values.
These unintended need to be und d if such are to be avoided in other
cases, and firms need good models to help anticipate the effects of regulatory change and discover
appropriate strategic responses. Since in this case there are complex interactions and feedback effects
between three distinct markets - (beer supply, pub retailing, and property) - models are needed that
capture the dynamics of competition in each market as well as the interactions between them. This
paper reports on the structure and results for a system dynamics model built to capture the growth of
pub-owning firms, given different ch istics, incentives and behaviours. This forms the basis of
a further model for the licensed property market as a whole. Data was derived from the Inquiry report
and discussions with industry executives, but all such data was available at the time of the original
inquiry.
Preliminary results d the hani by which firms grow chains of pubs and show how
the impact on property values and numbers might be assessed. The model also demonstrates the
broader value of applying a system dynamic methodology to researching issues of industry structure,
strategic management and business policy.
Industry, page 94
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Simulating the Effects of Regulatory Change in the UK Pubs Industry
The Structure and Regulation of U.K. Beer Supply.
In 1989, the UK beer supply industry was the subject of an inquiry report from the Monopolies and
Mergers Commission (MMC, 1989). This inquiry was mostly concerned with the possible anti-
competitive effects of the common industry practice whereby retail outlets (pubs) were restricted as
to their beer purchases by brewing firms who owned them. This is a form of vertical integration
which has long concerned economists and regulators alike (Williamson, 1975; Perry, 1989; MMC,
1990).
By international standards, the UK beer supply industry was until recently relatively fragmented. In
1985, 6 large firms accounted for 75% of output, with some 50 other small firms. As well as
production facilities, most brewers operate regional or national distribution systems to supply both
shops and the pubs/restaurants sector. Sites licensed for alcohol consumption (mostly pubs)
accounted for about 85% of beer volume sales in 1986. Of the 80,000 such premises, 46,000 (57%)
were owned by brewers. These ‘Tied houses’ source all draught and most packaged beer from the
owning brewer. The remaining pubs are independent ‘Free houses’, free to buy drink from any source.
These are usually owned by a private individual, although chains of Free pubs are increasingly being
assembled by larger firms.
Brewers and others compete in two principal markets - the wholesale market for the supply of beer
and other drinks to pubs, and the market for the retail sale of beer, other drinks and other products.
However, there is an important third market in which integrated brewers, individuals and pub chains
compete - the market for On-licensed property itself. This market is substantial. The large brewers
would all feature in the UK's top 10 property groups if quoted as such, with total assets of some
£20bn. Any change affecting this market is thus of considerable interest, not only to the brewers, but
also to the thousands of publicans whose livelihoods depend on it.
The brewers do not always operate the retail business of their pubs themselves. Only larger pubs,
whose retail profits justify the administration costs, are directly managed by the owning brewer. Most
are run by tenants who pay a rent. Although independent entrepreneurs, they are required to source
beer from their brewer-landlord. Free pub retailers may either purchase a property through which to
trade, or rent one from a brewer or other landlord. Larger firms may run groups of pubs along similar
lines to, say, a chain of supermarkets, or again may rent pubs out to individual publicans.
Any would-be pub retailer needs two main inputs - products to be sold to consumers, and the property
in which to make that sale. Since the markets for these inputs differ substantially, different bundles of
capabilities and resources would be held by firms who operate in each (Grant, 1991). Indeed, by 1986
certain firms had already chosen to operate largely in one or other of these markets. This paper
reports on a model of the licensed property market, a sector which has come into sharper focus since
the MMC inquiry, with a number of firms aiming to operate solely as licensed property landlords.
Two firms led the way in building activity in this sector, Brent-Walker's PubMaster division and
Grand Met's Inntrepreneur business unit. Both now own many hundreds of sites, and some 100 other
firms now operate anything from 10 to 200 sites.
The MMC inquiry luded that i bent brewers’ hip of Tied pubs was a large barrier to
entry by would-be beer suppliers and by new retailers alike (Bain, 1956), and recommended that
these firms be forced to sell all but 2,000 of their sites. This would affect only the largest brewers.
The Beer Orders (H M Government, 1989) fell short of this proposal, requiring the brewers only to
release the beer supply Tie on half of the pubs they owned above a base of 2,000. A brewer with
6,200 pubs, the average for the majors, would therefore have to free 2,100 of these. The brewer could
retain ownership, but offer long leases on the pubs to individuals or firms. A ively, the pubs
Industry, page 95
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
could be sold outright onto the property market. Both routes have been adopted by the large firms.
What the inquiry did not address, however, was the likely impact on the property market of their
Is, and in particular the prospects for survival of firms who took the opportunity provided by
the releasing of over 10,000 pubs (Williamson and Verdin, 1991)
A MODEL OF THE LICENSED PROPERTY MARKET
This model was developed from an outline of the policies and concerns of the participant firms and
their interaction with the market in which property is traded, derived from the MMC inquiry report
and from di with industry (Richardson et al, 1989; Forrester, 1980).
The licensed property market reflects the combined activities of firms with a variety of strategic
istics (strategy, policy, efficiency, and any interests in pub retailing and beer
supply). The relative size and number of these different types of firm alter through time, and have
certainly been altered by the effects of the Beer Orders. The focus of this model is thus on
understanding the interactions between the different types of firm and the licensed property market
itself. This market exhibits many features to be found in both the residential and general ial
property markets. Buyers are most active when property prices are thought to be low relative to
income. Property values are bid up, and buyers’ activity is stimulated by the belief that asset values
will rise. This enthusiasm may persist beyond the point where property values exceed what is really
justified by income potential, leading to a boom-and-bust, with property transactions falling, both in
number and in value. After a period of falling or stagnant property values, property again becomes an
attractive investment, and the cycle starts again.
‘The licensed property market, however, includes some unique features. The number of sites is largely
limited by licensing and planning restrictions. Most of these sites, generally the better ones, have long
been owned by the integrated brewers, who have tended to hold them indefinitely. The greatest
activity has therefore been amongst a fringe of less attractive sites. Since these sites have not offered
much advantage to the integrated brewers (relative to the sites they already owned), independent
publicans have been the most active participants in the market. Pub-owning chains are a relatively
recent phenomenon, whose activity has been considerably boosted by the Beer Orders.
The dynamic behaviour of the licensed property market depends fundamentally on the growth
strategies being pursued by the different types of owner. Whilst the market was partly stabilised by
the integrated brewers’ estates, the dynamics would have been determined by the investment
i of independent publi A dynamic model of the market therefore needs at its heart a
means of capturing the growth mechanism of a pub-owner (Figure 1)
Figure 1: The growth cycle for a pub chain.
profitability
pubs x required
ought
cash
generated Ty
property
pubs values
owned
Given an initial number of pubs owned, a certain amount of cash is generated. If sufficient cash is
accumulated, the owner buys further pubs, thus boosting the cash generated. The cash generated
depends on the quality of the sites owned. Similarly, pubs are only bought if the quality of those
available is sufficient to generate the profitability required.
The motives of the various types of pub-owner differ considerably in the extent to which they reflect
this simple model. Individual Free traders may seek to add to their single property, and attractive
property values encourage new individuals to enter the market. Pub chains with retail operations most
directly reflect the dynamics of the model given above. Many such firms started during the 1980s,
Industry, page 96
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
and have been boosted by the effects of the beer orders. Pub chains who lease to tenants reflect the
model in part, but their leases generally allow for rent revisions only every 3-5 years, so their cash
generation is lower and more stable than that of retail pub chains. However, they incur few of the
overheads of their retail I d brewers can participate fully in the model, but for
various reasons may choose not to do so. They may suffer a relatively poor average quality of sites,
so that the cash generated appears not to justify any expansion. This effect may be exacerbated by the
overheads they incur in controlling relatively large numbers of sites. They do have the benefit of up-
stream production profits on the beer supplied to their sites, but this may distract them from
optimising the retail operations. Finally, these firms may have other uses for the funds generated by
their pubs.
The practical behaviour of the basic model is thus considerably influenced by the policies of the
particular management of each firm. As a simulation model was developed from this causal loop
structure, these differences were tested with data appropriate to the different categories of firm.
More aggressive firms can use a further reinforcing effect to drive growth. The asset values of owned
sites can be used as security for borrowings. Those borrowings can be added to the cash resources
available, and thus boost the rate at which the firm can grow. The firm's cash generation from those
additional sites is, however, reduced by the interest payments to be made on the incurred debt.
Adding interest rates, pub profitability and various external features produces the model shown in
Figure 2.
Figure 2: The effect of borrowing and external factors on the model of pub chain growth.
growth
policy
profitability
required
related
business
interests
i | property
interest values
consumer
demand
debt
policy
The chain growth model
The central model for the growth of a pub-owning firm reflects this feedback structure and was
modelled in Powersim™™M | Figure 3 shows the structure of this model before the debt effect is added.
The ‘pub profitability’ sector computes the retail profit of the chain's pubs, given the average size in
terms of beer volumes sold. the discount on beer purchases, and other revenue and cost data derived
from the MMC's inquiry report. This pub profit, multiplied by the number of pubs owned and
adjusted for chain overhead costs produces the chain's profit. After deducting tax in the year
fe ing, the ‘cash a ation’ process is modelled. Cash d is added to the reserves, and
money spent on pubs is deducted. Growth is determined in the ‘investment decisions’ sector by both
the average return on each pub above a required return and by the management's growth ambitions set
by a maximum fraction of cash reserves they are prepared to spend on buying more pubs.
P rsim is the trade mark of AS, Bergen, Norway.
Industry, page 97
Investment decision
L
beer_bris_pa
retum_per_pub
ROA target e
Invest_potential ( v=
a
S|
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
The performance of the firm is largely d ined by the five highli iables. 'Pubs owned! sets
the initial size of the chain. A larger chain of pubs achieves better recovery of overheads and so
accumulates cash faster than a small chain. ‘Beer bris p.a.' gives the size of the chain's average pub.
Larger pubs generate disproportionately better returns than small ones, but there are diseconomies of
management control in pub retailing. Larger pubs would also benefit from better buying terms.
‘Discount per brl' reflects the buying power of the chain, and simply adds to the unit profitability of
each pub. ‘Cash spend max fraction’ sets management's growth aims, being set low for conservative
management or 1.0 (spend all available cash) for the most aggressive scenario. 'ROA target’ sets the
rate of return above which management is prepared to invest. The higher the excess return, the larger
the proportion of the i cash limit is prepared to spend. If returns are 10% above
this target or more, all of the maximum cash spend is committed to buying pubs.
Figure 3: The chain growth model in Powersim with no debt.
Pub profitability
ba
pubavg_value
4
os
ub_avg_vaiue
>
accumulation
‘pub_value_increase
The model was tested with data to reflect the different size and trading profiles of different categories
of pub operator. As an example, the growth of a 50-pub chain is given in Figure 4. Variables were set
at levels that are realistic for a typical pub chain - average pub size of 300 bris, discount of £40/brl,
ROA target of 8% at the pub level and the most aggressive use of available cash. The figure shows
the sensitivity of growth to the size of pubs owned. Small pubs simply do not generate sufficient cash
to be reinvested into additional outlets.
Industry, page 98
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
Figure 4: Sensitivity of chain size and profit to average pub size.
Gi in pub ir fit £0
irowth in pub numbers size of pubs Chain profit £000 p.a.
400 bris size of pubs
Bt ST HOO te
$00 [400 brs
200 [500
Gebers so gunn cero BOP ae SE i SS See
500} 200
T + t t + et
0 1 2 3 4 5 oO 1 2 3 4 5
years years
This model does not generate the rates of growth that have been achieved by the most aggressive pub-
acquirers, since it is lacking the further key driving mechanism of debt. This effect is added simply
with a stock variable that holds the debt of the firm (Figure 5). This level is set by a key policy
variable - gearing target - that sets desired debt as a fraction of the asset value currently held. The
increase in debt from period to period is added to cash resources, to be spent on more pubs according
to the firm's growth aims. Interest charges are derived from the debt level multiplied by the current
interest rate.
Debt also accelerates growth through the effect of inflation. Although inflation may often obscure the
simplicity of system dynamics (SD) models, its role in property firms is critical, since income may
rise with inflation, whilst the cost of interest on any given debt level may not.
Figure 5: Adding debt to the chain growth model.
chain_profit
Somat tin
inflation_inc
interest_charge
cash_resvs
interest_rate
debt
= debt_increase
5) I et i
pib_avevae( oN ‘
wO4 os
-- A sexing. scion deste
-——
pusaronrea "fee ae dacivuearea
This model can illustrate the powerful impact of debt with sensitivities to the policy variable -
‘gearing fraction desired’. Figure 6 shows the impact that gearing has on the growth of a 50-pub chain
of 300-barrel pubs.
Industry, page 99
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
The powerful impact of debt leverage on chain growth is clear, but the profit chart illustrates the
steep price that must be paid. The firm's profit is depressed by the interest charges on the debt the
firm takes on, to an extent that cannot be by the i i ies of chain scale. Note,
though, that this result is only true for the particular values that have been taken for the average size
of pub, discount levels and initial chain size. These combine to create average ROA for the chain that
is below the rate of interest on debt. The importance of this effect is illustrated in Figure 7, which
show chain size and profitability for different scales of average pub, given a standard gearing level of
50%.
Figure 6: Chain growth and gearing.
Growth in pub numbers vs. gearing . Chain profit £000 p.a. vs. gearing
gearing
years years
Figure 7: Impact of pub size on chain growth and profitability with 50% debt.
Growth in pub numbers vs. pub size Chain profit £'000 p.a. vs. pub size
Pub size 1.9
400 bris al
300
200
z : a t : T T t t +
0 1 2 3 4 5 0 1 2 3 4 5
years years
The licensed property market
Once the behaviour of the model for a single pub-owning firm has been modelled, the final stage is to
represent the interaction between these firms and the property market (Figure 8).
Figure 8: The link between pub returns and market values.
cash
generated —
potential
returns
bid
premum profitability
required
Pub-owning firms compare current property market values with the profitability they believe they can
achieve. If the market values are lower (higher) than they can justify, they will bid a premium
Industry, page 100
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
(discount) against market values. If many firms offer a premium, there will be upward pressure on
market values and, after some delay, these will rise. Higher average values reduce the returns that
pub-owners can achieve, so bid premia fall and upward pressure on values eases or may reverse.
The model represents typical market values by a simple stock variable whose inflow/outflow reflects
the weighted average of firms’ bids. This value for an average pub is fed back into the growth model
for each category of firm. If the profitability of pubs falls too far, none of the firms will be able to
justify running those pubs and they will close.
The model includes the option of setting a number of different categories of firm, each with its own
characteristics of pub size, chain size, gearing, growth targets, and so on. As an example of early data
used to assess the impact of regulating to remove Tied pubs, the model was run with the following
sets of rival pub firms.
Type of firm Pubs per firm Average size of Number of
pub (barrels p.a.) chains
Small independent pub chain 50 350 50
Large brewers’ tied pubs 6200 initially, 400 5
then 4,100
Large brewers' leased pubs O initially, then 200 5
2,100
Small brewers’ pubs 240 400 52
Note that ‘large brewers’ leased pubs’ is the number of pubs required by regulation to be freed from
the beer supply Tie. Naturally, such a brewer would release its smaller pubs first, and the average size
of 200 bris p.a. indicates the typical size of the smaller pubs owned by major brewers. Figure 9 gives
the results of running the model with these data.
Figure 9: The effect on pub market values and numbers from reducing the number of Tied
pubs owned by large brewers.
Total pubs in market. Value of standard 500 bri pub £'000
40, + + +
years
Cumulative total of pubs closed.
years
In the early years, the p ility of ‘ing pubs chains to grow. In year 2, though, the
enforced release of 2,100 pubs for each major brewer is modelled by switching this number of pubs
into the new category ‘Large Brewers’ leased pubs’. This depresses property values by some 5%,
which should lead to the take-up of pubs by non-brewing firms. However, without the up-stream
profits on beer supply, the profitability of these small pubs is insufficient to provide a return, even on
Industry, page 101
1994 INTERNATIONAL SYSTEM DYNAMICS CONFERENCE
the depressed property value, and many of the pubs leave the market altogether at the same time as
values are falling. Not all the 10,500 released pubs close, though, since lower values do allow some
to be taken on by other operators.
Conclusions
The model d here has imp implications for lation of the UK pub market.
Although freeing pubs from vertical ownership by major brewers would seem at first sight to
encourage new entry by pub op there is an imp - that the pubs
released fail to be economically viable when removed from the production interests of their former
owner. These pubs then depress the value of remaining pubs in the market. However, this process
does not result in all the released pubs closing, so the inder are absorbed by the ‘ining firms
in the market.
More lly, the paper indi the value of ing models of industry dynamics as a means
of testing out the possible effects of regulatory (or other) changes. This should be of interest both to
the regulators themselves and to firms whose prosperity may be affected by such changes and who
wish to experiment with alternative strategic responses.
Finally, the paper has broader implications for research into competitive strategy and business policy.
It suggests that, as in the physical sciences, it should be possible to conduct research not by looking
backwards at historic events, but by setting up software models of unknown future conditions and
testing those models against alternative sets of events.
References
Bain, J S, 1956, Barriers to new competition, Cambridge Mass.: Harvard University Press.
Forrester, J W, 1980, Information Sources for modelling the National E Journal of the A
Statistical Assn., 75 (371): 555-574.
Grant, R M, 1991, The ‘based theory of itive ad ge; implications for strategy
i California Me Review, Spring: 119-135
HM Government, 1989, The Beer Orders, London: House of Commons.
House of Commons, 1993, Agriculture Committee fourth report - Effects of the beer orders on the brewing
industry and consumers, London: HMSO.
MMC, 1989, The Supply of Beer. London: HMSO.
MMC, 1990, The Supply of Petrol. London: HMSO.
Perry, M K, 1989, Vertical I ion: D ination and Effects. In Handbook of Industrial Or
Eds. I and Willig. A North-Holland.
Enaeon G P, Vennix, J A M, Andersen, D B Rohrbaugh, J, Wallace, W A, 1989, Eliciting group
for model-building. In C be M of Complex Systems, Eds. Milling P and
Zahn E. Berlin: Springer-Verlag.
Williamson, O E, 1975, Markets and hierarchies - analysis and anti-trust implications. New York: Free
Press.
Williamson, P, and Verdin, P, 1991, From barriers to entry to barriers to survival, Unpublished
working paper, London Business School and Insead.
Industry, page 102
