Gambling with the Social Market Foundation

A new report shows how to start with a conclusion and work back

The Social Market Foundation (SMF) has published a dog’s breakfast of a report calling for big rises in most forms of online gambling duty. As I mentioned at The Critic in May, online casinos are taxed at 21% whereas online betting is taxed at 15%. Until 2019, they were both taxed at 15% but the government needed some extra money to make up the shortfall created by effectively banning fixed odds betting terminals.

The current government is considering ‘harmonising’ these tax rates which, with the fiscal black hole getting bigger by the day, can only mean setting them both at 21% or more. The SMF, which has been generously supported by Derek Webb over the years, used to want remote gaming duty (i.e. for online casinos) to be increased to 42% (for no other reason that 21 x 2 = 42) but it now wants it to be hiked up to 50% (a nice round number). It used to want affordability checks but it now accepts that they have been a disaster because they were not implemented as the SMF wanted (“real affordability checks have never been tried!”).

The SMF wants taxes on betting to be hiked from 15% to 25%. For horse racing betting, this would be divvied up between a 5% duty and a 20% Horserace Betting Levy (which goes to the racing industry and is currently 10%). Bookies would therefore pay no more tax on their profits from racing than they do now, and the racing industry itself would get a bigger handout from the bookies. That would lose the government a bit of money (£100m according to SMF) but that would be dwarfed by the extra revenue from higher taxes on sports betting and massively higher taxes on online casinos.

This is an attempt at triangulation that may find favour with some politicians. Horse racing is the acceptable face of gambling in Britain as far as a lot of people are concerned and so anti-gamblers pretend to approve of it in the same way that the temperance lobby pretends to approve of pubs. Indeed, the whole thing reminds me of when the anti-alcohol lobby hopped into bed with the pub trade to campaign for minimum pricing. That didn’t end well for pubs in Scotland which closed at twice the rate of pubs in England once minimum pricing was introduced (probably for the reasons I predicted).

Anti-gamblers can afford to be nice about racing because, like pubs, it is in decline. With the threat of a tax hike, the racing industry might be prepared to throw the rest of the gambling ecosystem under the bus for the prospect of more money from the bookies.

Going from 15% to 19% to 50% in a few years sounds a bit extreme, doesn’t it? On the face of it, it sounds like the SMF are asking Rachel Reeves to shoot yet another golden goose. Britain doesn’t produce many world leading industries, but gambling is one of them. (The UK excels in producing popular products that moral busybodies disapprove of, e.g. tobacco, alcohol, oil and gambling.) The SMF don’t think a huge increase in taxation will be a problem because some countries tax remote gambling at similarly high rates. They point to places like Delaware, Pennsylvania, Austria and the Netherlands which apparently have tax rates on gambling between 38% and 57%.

This is an interesting comparison but I would like to know more. I’ve never been on an Austrian online slots website and I doubt anyone at the SMF has either. Do customers have a good experience? Are the odds attractive? Does the Austrian government make a lot of money out of it or do punters go elsewhere? I do know that some US states operate an effective private gambling monopoly and a high tax rate is the price operators pay for getting the golden ticket. I also know that betting in the USA is, by and large, a miserable experience and that corporation tax tends to be a lot lower (it is 8.7% in Delaware and 8.99% in Pennsylvania compared with 25% in the UK). You’d need to look at the overall tax burden before you could make a reasonable comparison.

I don’t know what the revenue-maximising tax rate is on online casinos, but I am quite confident that a tax hike on sports betting would damage what’s left of the UK bookmaking industry and would harm horse racing (for the reasons the bookmaker Geoff Banks explains in this post). I also know that the SMF’s assumptions in this report are woeful and that’s what I’d like to focus on here. It piles one anti-gambling factoid on top of another to create a trifle of layered nonsense.

Its starting point is this:

At present, remote gaming in the UK is under-taxed, meaning that it is in effect subsidised by the state.

I don’t know exactly how much money remote gaming duty brings in but the OBR expects £3.8 billion from all gambling taxes this financial year. The majority of this will come from the online sector and most of that is taxed at the highest rate, so it is clearly a lot. The SMF don’t attempt to calculate how much tax revenue comes from the online sector, nor does it attempt to calculate what costs it supposedly puts on the state. Instead, it looks at the total amount of gambling tax revenue in 2022/23 (£3.3 billion) and uses two estimates of how much ‘gambling harm’ costs the state. The highest of these estimates comes from OHID, the successor of Public Health England.

The Office for Health Improvement and Disparities (OHID) has estimated that the direct financial cost to government due to harmful gambling was £413 million in 2022 and that, when combining the annual societal value of health impacts, the total annual cost of harmful gambling was between £1.05bn and £1.77bn.

I’ve written at length about this cost estimate before. The main point about it is that it is not a cost-benefit analysis because it ignores the benefits, and the main costs involve what OHID calls the ‘intangible societal cost’ of suicides which, as the term suggests, are not monetary costs to anyone, let alone the government.¹ OHID say explicitly that the direct costs to the government are only £412.9 million. The SMF says it explicitly in the quote above. So even if OHID’s estimate of how many suicides are ‘caused’ by problem gambling was credible - and it really isn’t - it still wouldn’t justify the claim that gambling is “subsidised by the state”.

This is doubly true because the £3.3 billion received by the government from the gambling industry in 2022/23 is a bigger number than the £1.77 billion that is at the top end of OHID’s estimate. Quite clearly, therefore, gamblers are subsidising non-gamblers.

But SMF have an answer to that.

Using Health Survey for England (HSE) data, both the OHID and NIESR studies used relatively low estimates of problem gambling, assuming it affects around 0.5% of adults.

The estimates are relatively low because the rate of problem gambling is relatively low and has been ever since it started being measured in the 1990s.

The more recent Gambling Survey for Great Britain (GSGB), using an experimental method first released in 2024, has estimated that the prevalence of problem gambling is five times as high, at 2.5%.

Ooh, an experimental method! Let’s throw out everything we know about problem gambling prevalence in Britain and rely on that instead, shall we?

That may represent an upper bound, but it would substantially increase the estimated fiscal cost of gambling, to £7.2bn using the OHID method and £5.0bn using the NIESR method.

It’s not a question of picking an upper or lower bound. Either the GSGB estimate is wrong or every other estimate from surveys conducted in the last 30 years is wrong. It’s a binary choice and it is pretty obvious what the correct answer is. The GSGB is an internet survey done of the cheap, it has a pitifully low response rate and is known to attract gamblers, and problem gamblers in particular, by its very nature. It cannot be taken seriously.

By contrast, last month saw the publication of the high quality Adult Psychiatric Morbidity Survey which reported a problem gambling prevalence of 0.4% which is in line with all the Health Survey for England statistics and those that preceded it. The GSGB is a risible outlier.

You can tell that the SMF are slightly embarrassed about having to lean on the GSGB, but needs must.

Our purpose in this report is not to determine which of the HSE or GSGB methodologies of measuring harm is the correct one or not…

Well, it should be.

…but rather to demonstrate to the Treasury the lower and upper bounds of fiscal costs using both these studies.

But if you look at the lower bounds, i.e. the only ones that reflect reality, the whole argument about externalities and subsidisation collapses.

While there is uncertainty over the size of the externalities generated by gambling harm, it is clear that the costs are likely to be in the billions, rather than hundreds of millions, of pounds.

OHID says it’s in the hundreds of millions.

By comparison, gambling duty generated £3.3bn for the Treasury in the 2022-23 financial year, meaning the sector is likely failing to cover its external costs.

Having made this assertion - for it is nothing more - the SMF turn to a further assertion, which is that there are certain gambling activities that ‘cause’ more ‘harm’ than others. The nature of the ‘harm’ is never defined. Instead, the SMF say things like “online slots have a higher-than-average proportion of people with a Problem Gambling Severity Index (PGSI) score of 8 or more”, as if having a high(ish) score in a problem gambling survey necessarily means that the person has suffered ‘harm’, and that the games they played must have ‘harmed’ them.

These bone-headed notions have floated around in anti-gambling circles for years and have really come into their own now that gambling is perceived as a ‘public health’ issue. A report from the Fabian Society (written by someone who seems to have been grown in a laboratory specifically to annoy me) recently said that we should simply “prohibit the most harmful gambling products”. The author admitted that she didn’t know what these ‘products’ were, but was confident that researchers could find out. The chances are that these researchers will just look at the proportion of problem gamblers among players of different games and conclude that correlation is causation.

The alternative and more plausible view is that there are a small but more or less fixed number of problem gamblers in society and they will seek out gambling ‘products’ that don’t interest most people until they grow out of it. These people will spend all their money on gambling until they get a grip and if you ban one form of gambling because it is especially ‘harmful’ then they will just lose all their money on something else. (Since the problem gambling rate hasn’t fallen since 2019, this must be what happened with fixed odds betting terminals - if you accept the claim of anti-gambling spivs that they were the ‘crack cocaine of gambling’.)

The SMF don’t actually want to ban “the most harmful types of gambling” - not yet, at least - but they do think the companies that make them available should pay more tax to cover the imaginary costs to the state.

By their own rationale, the “most harmful” form of gambling is betting on non-sports events in person. I’m not even sure I can picture what that means. Are these the people who will bet on the two proverbial flies climbing up a wall? Answers in the comments please.

1. SMF claim that “estimates of the societal costs of gambling also tend to focus relatively narrowly on tangible fiscal costs.” If only that were true.