Today HM Revenue & Customs and HM Treasury have published evaluations of the impacts of the furlough scheme for employees (the Coronavirus Job Retention Scheme, or CJRS) and the corresponding Self-Employment Income Support Scheme (SEISS) for the self-employed. I peer reviewed drafts of these reports. They contain a very wide range of analyses using various sources of data which I will not attempt to summarise comprehensively in this comment, but I briefly discuss some of the main takeaways.
The Coronavirus Job Retention Scheme
The report on the CJRS provides the only hard quantitative evidence of its kind that the furlough scheme substantially boosted UK employment during the pandemic, as intended and expected. It is extremely difficult to quantify this effect confidently for a number of reasons. The report draws both on evidence from employer surveys and from data on the earnings of employees (from HMRC PAYE records), which give broadly similar impressions. The main estimate comes from a comparison of the outcomes of PAYE employments that began just before, and just after, the ‘cut-off’ date for CJRS eligibility (e.g. for the first phase of the scheme, a job was eligible for furlough if it had started prior to 20 March 2020). Based on this comparison, the report estimates that at peak impact in Spring 2020 there were around 4 million employee jobs in the UK which, without the CJRS, would have been destroyed by that point. Like any conceivable method, this has limitations. For example, it relies on extrapolating from a group of people who began a new PAYE employment in March 2020, whereas the impact of CJRS on the wider population may have been different. All else equal, it will also underestimate the total impact if the CJRS boosted employment even among those not furloughed – which is likely, due to the impacts of stopping whole businesses from folding and in supporting consumer spending.
The report’s primary focus is on the impacts of implementing the CJRS relative to not implementing it – rather than a comparison of the CJRS with alternative options, such as an income protection scheme that did not depend on the preservation of jobs (more akin to the US approach), or with alternative designs for the CJRS or alternative timelines for phasing out the sizeable support that it offered. In its value-for-money assessment, the CJRS is estimated to have had a significant net social benefit. The precise numbers that such calculations provide should be taken with a pinch of salt given the number of assumptions that need to be made, but the conclusion that it was better to have a CJRS than to have none of the income or job protection that it offered would probably surprise few people.
Questions around whether hypothetical alternative approaches, or tweaks to the implementation of CJRS, would have been preferable are naturally even harder to address than the impacts of the actual scheme implemented. The report highlights data that raise some of the key questions here. For example, at around the time when employers were first required to make a financial contribution towards the labour costs of those on furlough in Autumn 2020, there was a large spike in redundancies. Does that mean that the government imposed these contributions too soon, destroying long-term viable jobs? Or does it indicate that it did so too late, supporting jobs that had no long-term future through the summer of 2020 and delaying necessary re-matching of people to different jobs? On the other hand, around one year later when the CJRS was finally withdrawn entirely, there was very little discernible impact on unemployment. Does this indicate that the withdrawal was timed smartly, or that the CJRS was left in place too long, supporting far too many jobs for longer than they needed supporting? Those are very difficult questions, but are of course key for informing policy in similar future scenarios.
The Self-Employment Income Support Scheme
Another report published today looks at the impacts of SEISS for the self-employed. Again, causal impacts of this scheme are very difficult to estimate: simple comparisons between those who did and did not receive support through SEISS will not suffice, since we would expect those receiving support to be in different circumstances from those who did not. The main estimates in the report come from a comparison between those self-employed individuals just below and just above the £50,000 trading profit threshold that was used to determine eligibility. This is a natural strategy to employ, though once again comes with inevitable limitations, including the fact that findings may not generalise to all eligible self-employed people (most of whom have an income well below the £50,000 threshold) who might, for example, be living closer to the edge in terms of having a long-term viable business. This analysis suggests that SEISS increased the probability of business survival in 2020–21 by around 3 percentage points among those claiming it. It does not find statistically significant longer-term effects on business survival, however, perhaps because some businesses were kept on life support temporarily but did not turn out to be the ones that had a long-term future. There is also some evidence that those with trading profits around £50,000 who claimed SEISS reduced their (non-SEISS) income in response, perhaps using the grant to buy themselves a period of lower work intensity (a finding that seems especially unlikely to generalise to the majority of self-employed individuals with lower incomes).
In this light, it is not surprising that most of the estimated net social benefit of SEISS in the HMT/HMRC evaluation comes not from saving businesses and facilitating a faster economic recovery but from what they call ‘equity’. This simply captures the fact that recipients of SEISS tended to have lower-than-average incomes, in combination with the standard assumption in value-for-money exercises that £1 makes a bigger difference to someone with a lower income than a higher income. It is worth noting that this does not include the sort of equity that many might consider more salient for this scheme – namely the provision of an income protection scheme for the self-employed similar to one that had already been put in place for employees. That the non-equity benefits of the scheme were not larger is a reflection, at least in part, of the difficulty in designing a scheme that could target those self-employed people who were most impacted by the pandemic. The report notes that policy here was constrained by the lack of more timely information about the profits of the self-employed, and that improvements in this area could enable more effective policy in future.
The income and job protection schemes put in place during the pandemic are very difficult to evaluate, not least given their size – injecting a total of about £100 billion into the economy – and the extremely unusual context in which they were introduced. We cannot be sure what would have happened to employment, the economy and indeed the path of the virus without these schemes. That said, it is likely that we would have seen a much higher spike in unemployment without them, and that doing nothing of this sort would have been a worse option. The most important remaining questions – even harder to address – are in relation to whether we could have had preferable outcomes with changes to the design, timing or generosity of the schemes.