Rishi Sunak’s furlough scheme comes to an end this week, at least in its original form.
Initially a three-month offer to those who would lose their jobs to Covid restrictions, the Job Retention Scheme became a six-month spending binge on a scale never seen before.
It is being followed by a Job Support Scheme, which after a few last-minute tweaks will look remarkably similar to furlough.
So did the initial scheme work? Was it worth the cost - and where does it leave the economy?
Who used it?
A total of 9.6m jobs have been furloughed at one point or another since March, according to HMRC.
At the peak in early May the best part of 9m were on furlough at once.
On top of that, 2.7m claims were made by July under the first tranche of the Self Employed Income Support Scheme, and another 2.3m under the second wave in October.
This varied enormously by industry, with some businesses able to shift staff to working from home easily, while others found sales completely dried up or required the sort of human contact that was banned under lockdown.
Retail, wholesale and motor repair - wrapped together in a slightly awkward category for the purposes of official statistics - led the way with a peak of more than 1.8m in May.
The closure of non-essential retail was particularly punishing.
Accommodation and food services were the next biggest, with more than 1.6m jobs placed on the scheme.
This pattern changed a little over time. As retail reopened ahead of hospitality, there were still more than 1m furloughed food and accommodation workers by the end of July, even as the number of retail and wholesale staff paid by the state was down to around 800,000.
Other sectors used it far less. The peak in finance and insurance was 70,000, while in public administration the numbers never rose above 15,000.
There were more modest differences by region, with workers in London typically a little more likely to be furloughed than their counterparts in other areas.
By age group, the youngest and oldest workers were most likely to end up being paid by the state.
Did it work?
So millions used the scheme - but did it "save" their jobs?
At a time when the economy shrank by around one-quarter almost overnight due to lockdown, the immediate impact on unemployment was stunningly muted.
From March to August, the number of payroll employees was down by 673,000, according to the ONS, or about 2.4pc of employees - a fall in jobs amounting to one-tenth of the fall in GDP.
Looked at another way, the total number of hours worked per week in the three months to August was down by about 15pc compared to the three months before the pandemic, yet employment fell far more slowly.
On the face of it, this is a ringing endorsement of the policy.
The clearest example of a similar nation which did not have a furlough equivalent is the United States. The world’s largest economy did not try to preserve jobs but instead ramped up unemployment benefits.
Its unemployment rate surged almost instantly into double figures, but has since fallen sharply.
“The truth is the job support schemes, if they had not been in place you probably would have seen European unemployment peak at about 20pc in the second quarter, if you had used a similar gauge to what has been happening in the US,” says Janet Henry, chief economist at HSBC.
As a plan to bridge the gap between the economic crunch and recovery, the JRS has worked well - at least on the initial assumption that the economy would be well on its way to recovery by the end of August.
Ultimately the test will come when the support ends. ONS surveys indicate around 9pc of business employees are still on furlough. That means just over 2m people remain dependent on the scheme, so if their bosses do not want to take them back on, their jobs may not have been ‘saved’ after all.
The picture for the self-employed is rather muddier. Typically their pay is less reliable than that of employees, making them more vulnerable to downturns.
On the other hand, a proportion were able to keep at least some money coming in on top of the self-employed support scheme cash from the Treasury. An estimated £1.3bn went to those whose incomes were not affected by the pandemic.
The number of people in self-employment between June and August was down by 472,000 compared with the months before the pandemic. This is a fall of almost 10pc, closely matching the drop in the economy in August compared to February.
In addition, not everyone was eligible. Efforts to target the scheme and avoid fraud meant up to 2.9m workers could not claim across the various schemes, the National Audit Office (NAO) estimates. This includes 400,000 short-term contractors moving between jobs, 700,000 limited company directors paid by dividends, and 1.6m self-employed, often because their income from other activities was higher than their trading incomes.
How much did it cost?
HMRC puts the cost of the furlough scheme at £41.4bn, plus an extra £13.7bn for the self-employed.
The Office for Budget Responsibility estimates these will rise to £54.5bn and £15.2bn respectively, taking the total to almost £70bn. For context, the UK's entire defence budget for the year was £55bn.
Surveys by the NAO indicate 9pc of those furloughed worked anyway at the request of their employer, in breach of the scheme. If that equates to 9pc of all claims, it could indicate that the best part of £5bn has been dished out via rule-breaking businesses.
On top of that may come businesses which told HMRC it was furloughing staff, but did not let workers know.
HMRC itself has a “working assumption” of a fraud rate of between 5pc and 10pc, and has had more than 10,000 calls to its fraud hotline that need investigating.
What comes next?
The plan was to keep people in work through the pandemic, enabling businesses to ramp up again without having to re-hire staff once Covid was gone.
Clearly the pandemic is not over and the recovery is stalling, so those millions still on furlough risk unemployment at the end of the month without further action.
The Chancellor has acted with a new Job Support Scheme for workers returning for at least a fifth of their usual hours.
This should cover those already on furlough part-time - at the end of August, that was almost 1m people, compared to just over 2m who were fully furloughed.
But the tough choice will come for those employers who have not been able to bring back staff at all, such as events businesses who are stuck with no conferences to organise.
Sunak will hope to support jobs which are sustainable and viable in the long-run, while those whose positions will never return have to find a new career. But judging this is tricky.
“Over one million jobs were being supported by the CJRS [furlough scheme] in sectors relatively unaffected by social distancing – for example in manufacturing, construction and professional services,” warned the Institute for Government.
“Many of these jobs are likely to be unviable in the longer-term if they are still reliant on the furlough scheme at this stage. But the new Job Support Scheme will continue to prop them up too.”
As Gertjan Vlieghe at the Bank of England says: “I do not know which of the very large reallocations that we have just experienced will be temporary, and which will persist. Neither does anyone else.”
That indicates pain delayed rather than avoided for those workers, and for the economy which will have large numbers of workers subsidised to stay in relatively unproductive jobs.
“These schemes will have to be phased out over time, otherwise there is a risk that there is a degree of zombification,” says Henry at HSBC.
“If people remain in these zombie jobs as there is a persistent postponement of companies restructured even once the public health situation is under control, that will have an even worse effect on medium- to long-term productivity.”
That calls for more training to ease the transition for those without viable jobs.
At the same time others can expect to lose furlough support even when their industries could come back. Most theatres, for instance, remain closed and will struggle to bring staff back even part-time, yet demand is likely to return once the pandemic is over. They may soon be unemployed.
All of this is expensive for the Government. Analysts at Capital Economics think the deficit this financial year could reach close to £400bn, compared to a peak of less than £160bn in the worst year of the credit crunch.
Forecasting at a time of such uncertainty is particularly difficult. Most predictions by economists expect unemployment in the final months of the year to rise to between 7pc and 9pc, double the 3.9pc at the start of the year and potentially matching the 8.5pc peak in the financial crisis.
This unprecedented policy and its unprecedented costs are not over yet.