Ryan Bourne is Chair in Public Understanding of Economics at the Cato Institute.

It will be tougher for many in Westminster to swallow than the subsidised food, but it increasingly looks as if the August Eat Out to Help Out scheme (EOTHO) was a costly economic and public health mistake.

Evidence now suggests that restaurants are important vectors in our current case uptick. More than that, the scheme has entrenched dining behaviours that threaten more transmission today. It is bizarre then that Rishi Sunak has avoided more critical scrutiny of the policy.

The Government’s most recent weekly coronavirus surveillance report says “eating out was the most commonly reported activity in the two to seven days prior to symptom onset” for infected individuals in the contact tracing system.

Figure 23: Events and activities reported by people testing positive, prior to symptom onset (enhanced contact tracing), England, NHS Test and Trace (as at 04:30am on 24 September 2020)

People do a lot of things during a week, so this doesn’t necessarily show where infection occurred. But that “eating out” appears more than “shopping,” “living alone or with family,” “holiday,” “visiting friends or family,” or “travel or commuting” suggests a high relative risk of restaurant dining.

This chimes with evidence from American states and the U.S. Center for Disease Control too. The latter’s recent study found adults testing positive for Covid-19 were twice as likely as those testing negative to have reported dining in a restaurant before becoming ill. Subsidising that activity in a pandemic seems a huge error.

That indicative data won’t convince everyone. When I Tweeted it last week, people demanded firmer evidence that (a) areas with more EOTHO meals had more cases, or (b) cases were actually seeded in August, when EOTHO was running.

But this level of precision may hide more than it reveals. In areas where prevalence of the disease was low in August, the risk of any activities would be low, meaning correlations between cases and meals in the scheme may not be particularly informative (as it happens, there is a correlation between meal numbers and Covid-19 cases by region). What matters is whether restaurants led to more transmission between infected and susceptible people in areas where prevalence was already there.

Accurately thinking through that counterfactual is tough. You have to (1) disentangle the scheme’s impact on dining numbers from any pent up demand returning after restaurants reopened in July; (2) account for the scheme’s longer-term impact of normalising eating out again. Examining August data alone therefore risks understating the scheme’s significance.

Basic network economics suggests activities like indoor dining might “link” more people in riskier circumstances who otherwise wouldn’t cross paths, particularly if packed into, passing through, or queuing on subsidised meal nights. Most restaurants took appropriate precautions to mitigate these risks. But sustained time indoors likely worsens the spread of the virus compared with other activities undertaken absent the subsidies. Case studies from Thailand, China, and Korea show restaurant’s risks. The scheme’s design also encouraged “superspreaders” – infected people visiting many outlets over time to take advantage of discounts.

Figure 22 of the Surveillance report shows that the overwhelming majority of “named contacts” given by infected people are those in their households or household visitors. Many faultily read this, and other information about where most transmission occurs, as showing that restaurants are insignificant vectors compared to domestic settings.

Figure 22: Contacts by exposure/activity setting in week 38, England
(Data source: NHS Test and Trace)

But this is misleading. People don’t know strangers in restaurants to give them as named contacts. And once the disease gets into a household, those most likely to be in contact with it are obviously others living or visiting that household. The problem is that restaurants seem more likely to be a place where a household member might catch the disease and then bring it home than other places they might otherwise spend lunchtimes or evenings. As the chart below shows, as EOTHO went on (Weeks 32 through 35), the share of “Covid-19 incidents” in food outlets or restaurants increased significantly.

Figure 20: Number of COVID-19 incidents by institution from week 27, England

What’s more, the Government didn’t just want to give restaurants a temporary boost in August. They wanted to encourage more economic normality. Data from OpenTable and others suggests they were successful, on this basis.

Diner numbers averaged 28 percent below last year for the week before EOTHO began. By the end of EOTHO, people were eating out excessively relative to previous years – averaging 44 percent per day above last year’s “normal” levels in EOTHO’s final week. Now numbers are back to around last year’s – i.e. dining is back at the good old days pre-Covid days of last year, despite the pandemic. This is in stark contrast to the U.S., where dining levels are still down over 40 percent.

It’s difficult not to conclude that, because of the message the scheme sent out or the habits it entrenched, EOTHO proved far more than a gimmick to give restaurants a temporary fillip. Instead it made people think restaurants were a-ok for people to party like it was 2019. That becomes more problematic now schools are open and more people are back to work too, bringing clear evidence networks of transmission have densified.

Unfortunately, people respond to this as if pointing it out means critics wanted restaurants permanently shuttered until a vaccine was available. But there’s a wide range of options between enforced closure and actively subsidising restaurants, including, well, not subsidising them, or subsidising outdoor dining and takeaways, tax breaks for investments for patios or delivery, and more.

What’s baffling economically is the thought process behind actively subsidising indoor gatherings. Social interactions right now impose negative externalities – risks on others beyond the diners themselves, for which the affected cannot receive compensation. Basic economics, if anything, suggests imposing taxes rather than handouts on these activities, to account for this social cost. Social distancing protocols and regulations seek to proxy for these taxes, of course. But it made no sense to undo this by overcoming people’s voluntarily choices and risk preferences in this world through taxpayer incentives.

Covid-19 debates, sadly, are more defined by culture wars and crude commercialism than economics these days. Many who usually oppose state subsidies backed this scheme loudly, not least because its use generated “buzz” about getting people out and about.

But that’s the problem. During the summer, many deluded themselves that the virus’s threat was deterministically and consistently falling, and we were on a one-way street to economic normalisation. Rather than adapting to live with the virus at the lowest overall cost, people thought that “restoration” should be the aim of government policy.

That was a grave error, of which EOTHO was Exhibit A. And with restaurants now facing restricted hours and a potential lockdown, even those short-term commercial benefits look a pyrrhic victory for the businesses who lauded the Chancellor at the time.