As UK business and society have been provided with their pathways out of lockdown in H2 2021. Regulus Partners states that UK gambling is in for a rude awakening as covid triggers the industry’s generational change.
Over the course of the last week, Britain has been given its (likely) path out of lockdown, statistics on online gambling activity which provides a (relatively) clear view of how 2020 evolved, and some headline evidence on problem gambling rates. Each of these provides valuable insight, and equally valuable hope, for a beleaguered industry.
However, while Britain may be trying to establish trading ‘normality’ from H221, the level of damage done to retail products especially will be profound and the unfolding implications are not visible yet. Further, the British gambling sector is entering a period of regulatory-political scrutiny unprecedented in a generation, while the Treasury will be looking hard for sectors which can help pay the bill. The timing of online’s impressive resilience may therefore be tricky, especially if key stakeholders are not focussed on clear longer-term trends.
The Gambling Commission’s monthly GGY guidance during Covid-19 allows us to piece together the out-turn for a highly distorted 2020 with a fairly high degree of accuracy. As we have pointed out before, capturing c. 85% of GGY still has plenty of issues (betting and poker, for example, over-index in capture vs. casino; how operators report GGY is still potentially subject to error). Nevertheless, it appears that 2020 has provided the following pattern:
- Remote betting is up c. 12% YoY and up 2% on what we expected prior to the pandemic unfolding; a c. £250m impact in Q2 has been more than offset by the Q3 sports calendar and very strong December margins (bets placed + 12% MoM; GGY + 53% MoM)
- Remote gaming is up 22% YoY and up 13% on what we expected prior to the pandemic; after normalising as the first lockdown unwound, growth picked up again in December (+12% MoM, continuing a Q4 trend)
- Overall remote growth is therefore c. 18%, which is 8% bigger than we expected prior to Covid-19 policy impacts, with c.33% of this being December betting margins
Given the relatively long road out of lockdown, Q120 is likely to look positive YoY vs. comps, with Q220 a very easy comp period for betting, but a tougher one for gaming (likely returning to single-digit growth). From H221, comps get very tough and absolute decline is likely as another round of ‘normalisation’ takes place. What will be critical from a policy perspective (very sensitive in terms of tax and Gambling Act Review), is whether focus is on ‘’Covid-19 beneficiaries’’ or far more pedestrian longer-term trends.
In that context, the publication of the latest Gambling Commission ‘problem gambling’ estimates was meaningful – but not perhaps in the way that some have tried to claim. The finding that the national estimate of problem gambling (from the PGSI short-form) had fallen from 0.6% to 0.3% of the population should be taken with a large dose of salt.
The Commission quarterly telephone surveys are not really designed for estimating population-level problem gambling rates (instead they fulfil an intermediate role between the Health Surveys). Telephone surveys are not considered as reliable as household surveys; and the PGSI short-form is a blunt tool (known for overestimates). The sample size (around 4,000 respondents each year split into four quarterly tranches) is small enough to warrant caution – the actual number of ‘problem gambling’ respondents appears to have reduced from around 24 to around 13.
That said, some have tried to make larger mountains out of smaller molehills – and those who have expressed outrage at convenient results from smaller sample surveys and yet stayed silent (or found other reasons for histrionics) this week forfeit credibility. At the same time, those who have tried to claim that the reduction is significant should be prepared for the possibility of regression to the mean.
The Commission’s estimates aside, only the most blinkered could have suggested that in a year when a large part of the gambling industry was shuttered ‘problem gambling’ prevalence would do anything but decline. The blinkered were however in full throat last year – in Parliament, in the press and in academe – and are sadly unlikely to be abashed by the publication of inconvenient statistics.
The decline in problem gambling rates is consistent with observations in Australia. One study by the Australian National University (Biddle, 2020) found both that the rate of problem gambling fell under lockdown and that those people who continued gambling (and were not ‘problem gamblers’) exhibited higher life satisfaction scores. It is often forgotten in policy debates on gambling that most people choose leisure pastimes that they find enjoyable.
Interestingly, the Gambling Commission appears to be coming around to the view that the prevalence of problem gambling may in fact be reducing. This week at a Westminster E Forum event, executive director of the Commission, Tim Miller claimed that “it does appear that there is an emerging trend showing a decline in overall rates of problem gambling”. This is a welcome and sensible reflection on the evidence-base, in line with the government’s own request for evidence-led thinking and balance. We hope that this measured position evolves positively as the Commission plays a key role in shaping policy at a critical time.
There is however a risk that some may draw very dangerous conclusions from the data – namely that the reduction of opportunities for adults to gamble has been effective in reducing the rate of problem gambling. Some in the public health lobby may in fact see the latest prevalence figures as vindication for an even more restrictive (or even prohibitive) approach.
One area of potential gambling policy intervention which probably does need a return to the evidence-base is in-play betting. The sub-product was referred to in Miller’s Westminster E Forum speech as an ‘intense’ product (also a recent one; which it isn’t), in the GC consultation on affordability as an ‘indicator of harm’ regardless of context, in the Gambling Harms APPG report as something to restrict to in-venue and telephone (like Australia – see blogs), and in the House of Lords report on gambling as a de facto riskier product. While in-play betting is certainly more likely to pose risks than draw-based lottery, its characteristics become more complex than clearly riskier thereafter.
Perhaps unhelpfully, in-play betting tends to get talked up as a contributor by operators which want to look ‘modern’ and data providers who want to sell services. There is certainly a higher volume of lower margin bets than ‘multiples’ in football (structurally), which when combined with the long-play tournament characteristics of many sports (betting on tennis, golf or cricket is almost inevitably ‘in-play’), can give an image of considerable volume (c. 80% + stakes ex horseracing). In reality, it represents c. 50% of football betting net revenue (70% of stakes) and c. 40% of total net betting revenue in GB, including horseracing (where in-play is tiny). It is also the case that customers can in theory bet c. 1x per minute given the volume of betting opportunities presented and UX capabilities, but that is not much more than the pre-match total of ‘traditional’ racing products (GB + International + dogs + virtual), certainly during retail trading periods. Customers only bet to this intensity extremely rarely, and if they do then the behaviour can be picked up. Equally, one bet per minute is not unusual as a pattern in gaming (even cash line bingo), and so the context and safeguards matter a great deal more than the bald number.
Nevertheless, it is far more likely that even an engaged customer will have only 2 or 3 in-play bets over a match-session, while the sizeable majority of sports customers (the long occasional annual tail) do not place in-play bets at all (amounting to 75-80% of the c. 8-10m annual bettors). It is unsurprising therefore that sportsbetting not only has low levels of recorded problem gambling prevalence but also that this prevalence has not increased during the growth of in-play (broadly from 2007).
The behavioural characteristics surrounding in-play are not structurally intense, just as they are not with horseracing unless someone chooses to bet on every available option, in which case the risk parameters are only broadly similar to traditional products and many established gaming products. In play betting therefore retains a policy intervention risk without the evidence to back up a specific intervention:
- high-frequency betting is not the same as in-play betting and the two are only loosely linked
- the growth in in-play betting has not driven any growth in problem gambling amongst online sports bettors
- the structural characteristics of sports and bet types, combined with some industry hype, drive an appearance of intensive activity which is not borne out by reality
However, if these points are not made clear, or if they become victims of political points scoring and inter-sector rivalry, then ill-considered but highly damaging policy responses cannot be ruled out.
Retail gambling faces brutally different trading and policy challenges to online. For landbased gaming, the issue of an online alternative has been developing for over two decades and the extent to which lockdown is driving structural channel shift is not clear, given the maturity of the GB market (it is certainly not pronounced). The unwinding of lockdown is therefore principally a cash flow issue, albeit a degree of long-term damage is bound to have been done to demand, especially to supply with limited scope for innovation. The risks for LBOs and their product stakeholders are more subtle, however.
In the first lockdown, LBO closures coincided with the significant disruption of substantially all mainstream sports. The period of re-opening demonstrated strong overall resilience, but established three key trends:
- SSBTs were by some margin the strongest performing product, with total revenue growing vs. pre-Covid policy impact levels (+ c. 40%) despite footfall disruption and a reduced estate
- Gaming machines proved resilient, with GGY at c. 95% of pre disruption levels; reflective far more of post-B2 estate reductions than Covid-19 policies, in our view
- OTC fell by 20% vs. pre Covid-19 policy impact and additional closure levels, undermining the hope that reducing shops would be fully redistributed, but also most heavily impacted by consumer behaviour (lower dwell times, more likely to use machines protected by screens)
However, LBO re-opening from 12 April (in England) is likely to be phased, with the initial rules reflecting old ‘Tier Three’ trading restrictions. This allowed LBOs to reopen, but without pictures (racing, virtual, soccer-led sports channels), dwell time restricted to 15 minutes, only two gaming machines in operation and 8pm closure. The impact of this could be dramatic from both a supply and a demand perspective, in our view, even if it only lasts for a few months.
First, SSBTs are the clear beneficiaries of reduced dwell times, no pictures, full content operation and evolving customer safety habits. SSBT revenue mix is therefore likely to grow further, including capturing more customers who might historically have been OTC-led. Absolute revenue trends could be impacted by dwell time, but this may be mitigated by product shift. To what extent these customers revert to traditional betting patterns when restrictions are lifted is open to question.
Second, gaming machines are likely to be impacted by dwell time and evening trading restrictions, but this at least probably means the reduced number of machines is likely to satisfy most demand. Any cash flow generated by machines would be welcome and is still likely to be more than 50% of pre-Covid-19 impact levels. However, this is more of an unwelcome short-term cash flow hit with an operational light at the end of the tunnel rather than a structural issue. The extent of the pain is more driven by timing and balance sheets, than shifting customer behaviour therefore.
OTC is in a different and much more dangerous position. Before tiering, OTC had been reduced to 36% of GGY. Without pictures to drive interest in the racing betting data, and without the ability to hang around, a number of customers (also typically older) are likely to not turn up at all.
Further, the decision to hold the Grand National in lockdown is also likely to encourage a much greater proportion of older customers to try digital channels for the first time. How many then come back to the LBO environment is open to question. Those that do will be given the opportunity to bet on racing without live TV coverage, however. Live TV coverage is one of the biggest single costs of an LBO: never as big as staff costs but very often similar to or bigger than the rent; also, the biggest single product-attributable cost by some margin. With OTC driven below 35% of revenue, the economic pressures on racing pictures is likely to be acute. This is a very dangerous time to give LBOs the opportunity to test picture-free economics, therefore.
Featured article edited by SBC from ‘Winning Post’ Sunday 28 February 2021 (click on the below logo to access a full unedited version)