Tim Helm was previously a senior economist in DTF`s gambling

Submission to the Review of the Gaming Machine Entitlement Term
Dr Tim Helm1
About the author
Tim Helm was previously a senior economist in DTF’s gambling policy group, and has experience in
gambling policy spanning licensing, regulation and taxation for EGMs, wagering, and casino gaming.
He has a PhD in economics from Melbourne University. The views expressed here are his own, and
not those of DTF.
Summary
This submission argues two key points relevant to the Review:


The accepted wisdom following the VAGO report that the auction failed due to insufficient
demand resulting from high regulatory barriers and PACO is misguided. PACO did not alter
the supply-demand balance at auction, in fact, and the low prices at auction are more
sensibly understood as a consequence of an over-supply of entitlements for sale, given the
regulatory processes deemed appropriate by government and the level of EGM availability
deemed profitable by the industry, than a shortfall of demand.
Selling state assets like entitlements should generally be done as late as possible, for as short
a period as possible, because to sell assets that cannot yet be used from a party with low
borrowing costs (the state) to parties with high borrowing costs (venues) is economically
inefficient, destroying value at the expense of future taxpayers.
Two recommendations follow from these points:


Extensions should only be provided for a portion (e.g. 50%) of the 27,500 GMEs currently
allocated, and they should be allocated with a market mechanism (auction) to maximise
competitive tension and revenue for the state. This would allocate extensions only where
genuinely required, and would appropriately retain the option for a future government to
reduce the number of EGMs in the state post-2022.
Licence extensions should be kept as short as necessary to solve any genuine issues with the
current term, and payments should be over the life of the extension (not upfront). Indefinite
entitlements are an economically inefficient way to allocate gaming rights, and they unfairly
favour today’s taxpayers and industry profits over future taxpayers.
If the government wants to issue post-2022 gaming rights, the recommended approach is to auction
13,750 GMEs in 2015 for the period 2022-2032, with each geographical/venue-type market allocated
half the number sold in 2010.
The number of 2022-2032 GMEs to be auctioned in 2020 would be chosen by a future government
in accordance with their social policy goals, but this number would need to be less than 13,750 in
order to ensure competitive tension at auction.
1
Contact: [email protected] or http://au.linkedin.com/pub/tim-helm/40/66/a7b
1.
Supply-demand balance at auction
As the Issues Paper notes, VAGO’s 2011 report identified both institutional (project management)
and auction design failures.
Prominent among the latter were the ideas that the PACO process withdrew demand from auction,
and that the multiple regulatory approvals necessary to operate EGMs further stifled demand.
I submit that VAGO (and DOJ in its audit response) have missed the deep underlying cause of
inadequate tension in the auction.
This cause was the oversupply of GMEs for sale relative to the needs of the existing EGM industry in
Victoria, which meant market-clearing prices reflected not the value of GMEs to established
operators, but the willingness-to-pay of ‘speculative’ operators either entering the industry for the
first time or considering opening new venues.
Offering fewer GMEs for sale would have generated more revenue. The auction failure is thus more
sensibly understood as due to an over-supply of entitlements given the prevailing demand (which is
difficult to alter), than to weak demand given the supply available (which is easy to alter).
1.1.
EGM operations pre-2012
Before 16 August 2012 Tatts and Tabcorp were jointly permitted to operate 27,500 EGMs in Victoria
– 13,750 in clubs, and 13,750 in hotels.
However during the last years of this duopoly these operators elected to operate only around 26,600
machines – 900 short of their entitlement. That is, only 26,600 EGMs were deemed to be profitable
even at the effective zero marginal cost they faced for the basic right to operate each machine.
Although both operators consolidated operations in their final licence years to maximise their
remaining profits, this only affected machine numbers significantly in 2011-12. As far back as 200809, with more than three licence years remaining, Tatts and Tabcorp deemed it profit-maximising to
operate around 800 EGMs less than they could have. This is seen in the graph below.
Evidently the basic economics of the industry – given the prevailing tax rates and regulatory
requirements (including geographical restrictions) – did not support operating the full quota.
1.2.
Over-supply at auction
When 27,500 entitlements were allocated in 2010, the hotel and club sector was therefore able to
buy entitlements to cover their existing level of EGM operations with around 800 leftover.
Demand by venues that had operated EGMs in the recent past (e.g. in 2008-09) might be labelled
“proven demand” – demand for entitlements to permit operation of EGMs for which the necessary
infrastructure was already in place (physical venue, regulatory approvals, established business
processes etc), and for which future revenue and profitability were relatively certain thanks to an
established customer base and availability of past performance data.
The 800-odd entitlements in excess of that needed to satisfy this proven demand were purchased by
buyers not seeking to continue existing operations, but to open new venues, expand existing venues,
or to hold and sell to other operators. This might be called “speculative demand”.
The auction mechanism revealed the entitlement prices for which demand and supply of
entitlements were equal: the market-clearing price. But since only 26,600 entitlements were needed
to satisfy proven demand, the market-clearing price was effectively determined by speculative
demand, and the price paid by all bidders was determined by the willingness-to-pay of the
speculative bidders.
The auction failed to raise much revenue precisely because the willingness-to-pay of speculative
bidders was extremely low relative to the willingness-to-pay of proven bidders.
With average revenue in 2008-09 of around $100,000 per machine per year, and with taxes and
costs expected to consume four-fifths of this over the entitlement period (according to VAGO), the
average existing operator with a 15% discount rate would have applied a Price/Earnings multiple of
about 5 and hence been willing to pay roughly $100,000 per entitlement.
In contrast, a speculative bidder aware of the regulatory barriers and costs of new EGM operations
would have heavily discounted this amount. It is important to remember that Tatts and Tabcorp,
sophisticated businesses able to generate higher profits from their machines, effectively revealed a
zero willingness-to-pay to operate more than 26,600 EGMs. Smaller operators second-guessing that
decision could hardly have been expected to pay much more. Judging from the number of club
entitlements currently unused and offered for sale at or below auction reserve, many of these
buyers indeed paid too much.
This situation is illustrated in the diagram below.
That the auction generated an efficient allocation – GMEs went where they were most highly
valued – and that it maximised revenue given the GME supply specified to the auctioneer, are both
true, but are not particularly important outcomes.
The more important fact is that the state did not choose the supply of entitlements correctly to
maximise revenue: by restricting the number of GMEs sold it could have created more tension
between bidders, causing the market-clearing price to be set by existing operators fighting to retain
their profitable income stream, and hence generating much more revenue. This hypothetical
outcome is also shown in the diagram.
1.3.
Price discrimination
Economists would recognise the problem the state faced as akin to a monopolist facing two different
customer types with quite different willingness-to-pay.
The profit-maximising response if customers can be identified is to charge different prices to each
type (price discrimination). PACO, the most successful part of the allocation process from a revenue
perspective, did exactly that.
But if customers cannot be identified, the goods can be on-sold, or discrimination is impossible or
undesirable for legal or fairness reasons, then the second-best response is to restrict supply and
raise prices so that low demand types are priced out, rather than let the market-clearing price be set
by low-demand buyers.
1.4.
Multiple markets
The fact that entitlements were not all auctioned in one large market, but rather in 176 geographical
area/venue type markets, nuances this story without altering the underlying message in the
description above. That message is that the auction raised little revenue because many of the
market-clearing prices were determined by speculative bids, not by bids from existing venues
competing to stay in business.
1.5.
PACO
VAGO and DOJ (in its audit response) argued that the decision by the government of the day to allow
clubs to purchase entitlements at a fixed price before the auction in the “Pre-Auction Club Offer”
was a factor suppressing demand and hence prices at auction.
However this logic is flawed, since PACO reduced auction demand and supply equally. For each
entitlement bought via PACO the number available at auction was reduced by one. If there were 800
unused EGM rights in 2008-09, then there would have been 800 entitlements sold at auction that
were surplus to satisfying existing demand no matter how many club entitlements were allocated to
existing operators at PACO.
1.6.
Implications for extending entitlement terms
At the end of December 2013 there were 26,258 EGMs in operation, and 1,242 GMEs unused – 400
more than Tatts and Tabcorp left idle in 2008-09.
If the state allocates rights to operate EGMs beyond 2022 using a market mechanism, and offers
these rights for all 27,500 GMEs, then it faces a real prospect of repeating the mistake of 2010 and
giving up valuable rights at a bargain price.
The industry’s willingness-to-pay will likely be even lower than in 2010 since these extended rights
commence so much further in the future (eight years from 2014, compared to the two from 2010).
And with public opinion having moving significantly against pokies since 2010, how much more will
operators discount their bids to reflect the prospects of stronger regulation in the 2020’s?
These considerations imply that the best approach to targeting entitlement extensions at those
operators which genuinely need them, while maximising value for the state, is to provide extensions
for fewer than 27,500 entitlements with a market mechanism to allocate these extensions.
As the Issues Paper identifies (p22), a competitive allocation has the potential to maximise revenue
contingent on there being sufficient competitive tension. Restricting the number of extensions will
create that tension and will put extensions in the hands of those operators that most value them.
One option would be to auction all post-2022 GMEs now, but to issue (say) 10% fewer than are
currently held (24,750). Legislating the post-2022 statewide cap at this number and thus precluding
the possibility of more being issued later will force venues to bid aggressively to survive. Those that
are forced to (or wish to) exit the industry will have eight years to adjust their business models.
A better option would be to auction 50% of post-2022 GMEs now (13,750) while leaving a future
government the choice over how many to auction in 2020. If the bulk of GMEs are extended beyond
2022 now, it will be more difficult and costly for a government to reduce EGM availability in future
(which is perhaps one reason the industry seeks this).
Venues facing genuine borrowing constraints that would be relieved by an extension (as opposed to
borrowing constraints due to mismanagement or unprofitability), and venues planning investments
with significant payoffs post-2022, will be willing to pay the most for the extensions.
Venues that do not face such issues, but are just joining in the rent-seeking – expecting a quid pro
quo for the recent tax increase, and seeing an election year as an opportune time to buy future
profits at a bargain price – will be priced out of this market by those venues facing genuine need.
The state will solve four problems at once: (1) suboptimal venue investment and credit constraints
caused by short entitlement terms (to the extent this is real); (2) ensuring term extensions are
provided only to those who truly need them; (3) maximising the extension sale price; (4) retaining
flexibility for future governments to reduce the state-wide EGM cap if desired.
Finally, it is noted that Tatts and Tabcorp built a profitable, stable business with 18 year licences, and
were given just four years’ notice that these would not be renewed. A 10-year extension issued in
2015 would give operators 17 years’ certainty, which is as much as Tatts and Tabcorp ever received.
This “short to medium” extension (as labelled by the Issues Paper) would be more than enough.
2. Discount rates and maximising sale value
This section addresses question 14 (section 5.2.3) in the Issues Paper.
The state faces a choice between selling rights to operate EGMs beyond 2022 now or closer to the
time (e.g. 2020), but the state must also choose when to demand payment for these rights.
If the state sells these rights now and demands payment sooner than would have been the case
under a later sale, then value is destroyed. A party with a low time value of money (the state, with a
borrowing cost around 5%) is effectively borrowing money from a party with a high time value of
money (venues, which discount future revenues at perhaps 15%).
This transfer is an inefficient use of society’s capital, and the inefficiency is ultimately borne by the
taxpayer via lower revenue from the sale. It is analogous to privatising a public asset for less than the
present value of its dividend stream, or entering into a PPP where the NPV of the service payments
exceeds the cost of the buy-it-yourself option. In all these cases the state’s balance sheet is
improved today but at a net cost to future generations, a clear intergenerational equity problem.
A simple example illustrates the potential cost from taking early payment for post-2022 GMEs.
Suppose in 2020 the 2022-2032 entitlements are valued by the industry at $4bn, and an improved
auction structure means the state captures three-quarters of this instead of one-quarter. The receipt
of $3bn in 2020 has a present value to the state of $2.4bn in 2015.
Suppose instead these same 2022-2032 rights are auctioned in 2015, with the payment schedule
also brought forward 5 years. Rights worth $4bn to the industry in 2020 are valued by the industry at
$2bn in 2015. If the state again captures three-quarters, then it makes $1.5bn from the sale.
Thus $0.9bn in 2015 dollars is lost to the state (and society) by demanding earlier payment than
would otherwise be the case. From the perspective of taxpayers in 2020 the balance sheet is $1.1bn
short of what it would have been had the auction and payments been delayed.
The recommendation from this is twofold.
If post-2022 entitlements are to be issued, then payment should occur no earlier than it would have
otherwise (i.e. no earlier than 2020, assuming the next auction would have been held at that time).
If for political reasons the government must seek early payment, then the post-2022 entitlements
should be as short as possible to minimise the net cost to future taxpayers arising from the
inefficient allocation of capital described above.
3. Other comments
3.1.
Administrative allocation vs auction
The Issues Paper contemplates an “administrative allocation”, presumably motivated by fear of
another auction failure. While this appears attractive, there are several counterpoints worth noting.
First, the process of devising the formula for an administrative allocation will be subject to continual
lobbying from industry that will occupy time and resources and pervert the outcome. Even if an
administrative allocation appears simpler than a competitive allocation from a budgetary point of
view, the cost from society’s point of view including resources the industry will expend trying to
influence the process might well be higher. Moreover, in selling billions of dollars of state assets, any
additional costs of running an auction process pale in comparison to the potential revenue losses
from a formula that sets extension prices either too high or too low.
Secondly, there is an asymmetric information problem that means the public service will invariably
be fooled by industry when it comes to the estimation of extension values (and hence the pricing
formula). The advantage of a market mechanism – if properly designed – is that this information
asymmetry becomes irrelevant.
Thirdly, there is an argument that a running an auction process much like the 2010 one could be
relatively cheap and straightforward (for participants and government alike) since the industry and
public service have so recently been through the experience. Instructional materials have already
been developed, most bidders have already been through the process, and the same auctioneer,
software and venue could be used, for example. If the rights being auctioned are broadly the same
as those sold in 2010 (e.g. 13,750 or 24,750 GMEs in the same geographical/venue-type proportions
as before) the auction could be repeated almost identically.
3.2.
Eligibility to buy post-2022 entitlements
Section 5.2.1 of the Issues Paper gives three options for “who could apply for an extension”.
In fact, only option (a) constitutes an extension as such. Options (b) and (c) are equivalent to issuing
new gaming machine entitlements for the post-2022 period. Option (c) envisages an allocation open
to new entrants just as the 2010 auction was, while option (b) envisages restricting future EGM
operations only to current participants.
Both fairness and competition reasons justify keeping the allocation open to incumbents and new
entrants alike, and options (a) and (b) should be rejected outright.
Moreover, what is being proposed in options (b) and (c) amounts simply to issuing new pokies
licences, and should be labelled as such. GMEs are merely ‘pieces of paper’, and there is no reason
to continue the present numbering system beyond 15 August 2022 with an ‘extension’ when what is
important to industry is continuation of the underlying right.
Explicitly issuing new post-2022 entitlements rather than ‘extending’ existing ones would be simpler
and more transparent to industry and the public. It would also be necessary if the number of post2022 GMEs is to be reduced, which is necessary to avoid repeating the failure of 2010 (as argued in
section 1 of this submission).