Steering the UK economy out of the inflationary storm would always require unusual finesse, but the task would be easier if the Bank of England governor had an unemployment number he could trust.
The problem, highlighted by Andrew Bailey in a recent hearing, stems from deeply flawed labour market data from the Office for National Statistics, led by national statistician Sir Ian Diamond.
The situation has cast a shadow over the ONS, an independent government agency tasked with producing reliable economic figures that are critical to BoE decisions about when and how far to cut rates this year.
Lord George Bridges, the chair of the House of Lords economic affairs committee, this month warned that the BoE and Treasury were like pilots trying to plot a course while their instruments were “flashing and whirring around”.
“How are the Treasury and the bank to make critical decisions based on dodgy statistics?” he said in a House of Lords debate.
His view is widely shared among UK economists, business leaders and policymakers as they face as much as a year of uncertainty about the labour force, including questions like how many people are unemployed.
The ONS said this month its new labour force survey would not be ready until September, six months later than it previously said and almost a full year after it first suspended publishing jobs data.
Last week it resumed releasing jobs data, but it warned the figures were still not fully reliable because of the low response rate to its old survey.
The body has not fully explained key decisions taken last year that proved deeply damaging to the reliability of UK labour market data. Diamond himself was not available for an interview, his spokesperson said.
“The response rate to the labour force survey is shocking right now — it is a really poor situation,” said Erik Britton, a former BoE economist now at Fathom Consulting.
“It is a problem,” BoE governor Andrew Bailey told the Lords economics committee last week.
Though Bailey said the central bank could track employment through other sources, there was no real alternative to the labour force survey (LFS) to find out how many people were out of work or why.
“Our staff have advised us that whether unemployment is 3.8 per cent or 4.2 per cent is really pretty hard to judge,” he said.
The crisis besetting the ONS survey has been years in the making.
Valued for its detail, the LFS is time-consuming for respondents. It has morphed so many times in its 50 year history as extra questions have been added that one researcher likened it to the children’s game of drawing monsters by blindly sticking together body parts.
Because of its length, the LFS has been hard hit by a decline seen in many countries in the response rate to household surveys, as people become harder to contact and less willing to give personal details to cold callers.
The LFS was designed to operate with a response rate of 55 per cent. A decade ago it was just below 50 per cent. It has now slid below 15 per cent.
Alarmed by the long-term trend, the ONS had in 2017 already begun developing a simpler “transformed” labour force survey (TLFS) that would run primarily online and produce reliable results with a lower response rate.
When Covid hit, forcing a switch from in-person to telephone interviews, responses to the old survey plummeted, forcing the ONS to plough resources into salvaging it by doubling the number of people contacted.
The ONS also worried the disruption of the pandemic had changed the mix of people answering the survey, skewing the results. Users of the data increasingly noticed its results were at odds with other official figures.
But by mid 2021, the ONS began drawing down the sample size. The agency maintained it was still confident about the headline jobless rate even if there was uncertainty around some of the more detailed breakdowns of the data. Meanwhile, it was taking longer than expected to get the TLFS right.
Darren Morgan, who retired at the end of 2023 as director of economic statistics production, told the Financial Times in November he had to juggle resources between propping up the old survey and ramping up the new one.
“Running two large surveys for longer than we thought — that takes capacity,” he said. The ONS had to redeploy staff while still protecting other surveys “as best we can”, Morgan said.
He insisted the underlying problem was not one of money, but of trying to get people to respond to surveys.
But in July 2023 the ONS made a decision that proved to be a tipping point. It pulled more resources from the old survey, returning it to its pre-Covid size. In July to September the response rate fell to an all-time low of 12.7 per cent, from 14.6 per cent in the previous three months.
When officials crunched the numbers, the results looked implausible — with unusual swings in youth unemployment in particular — forcing the ONS to pull publication with just days’ notice in October.
The ONS declined to answer questions about why it cut back on the old survey, who was responsible for the move, or the exact date of the decision.
But it was clear that scaling back the LFS, while still relying on it because of delays to the new survey, was risky.
On July 10, the Office for the Statistical Regulator warned the ONS about the “sustainability of the current LFS” and noted concerns about the agency’s use of statistical fixes to make up for the small sample size.
The OSR told the FT it was not aware the ONS was reducing the sample size that month when it flagged up this concern in a progress report on the TLFS.
Morgan’s successor, Liz McKeown, now faces an uphill battle to repair the damage. Even with more researchers in the field, a modest boost to the sample size from October and a bigger 50 per cent boost from January, it will take time and money for the LFS-based data to improve. The agency now uses branded notebooks and £10 vouchers to induce people to respond.
Despite the failures at the ONS, the agency’s regulator has said the main problem has chiefly been communication. “The ONS is trying to do some difficult things,” Ed Humpherson, head of the OSR, told a parliamentary committee this month.
He added: “The underlying common thread I would encourage the ONS to really think about is how it responds to users, to challenge, and how it communicates uncertainty.”
Both the ONS and OSR are part of the UK Statistics Authority, which operates at arm’s length from government to ensure its independence from ministers. It is accountable directly to parliament, though the Cabinet Office is involved in appointing non-executive board members.
With tight funding settlements across the government, the UKSA has committed to find efficiency savings of 10 per cent in its baseline budget, which will fall from £225mn to £220.8mn in 2024-25.
However, it has additional funding of £17.6mn allocated in that year for improvements to key economic statistics, including on the labour market.
The ONS has been through rocky patches in the past. It lost close to 90 per cent of its London-based staff after its headquarters shifted to Newport, Wales, in the late 2000s, costing it significant expertise at the time.
Diamond wrote to MPs on the public administration and constitutional affairs committee in November setting out the ONS’s plans to rebuild the jobs data, but has otherwise said nothing publicly on the issue.
The ONS said its work on the area was getting results. “Response rates have been improving since the autumn, with the survey sample boosted by 50 per cent to 24,000 homes. This will continue until we switch to the new survey.
“Its later introduction in September will help it to yield better data for policymakers and address marked changes since the pandemic in public attitudes and behaviours in relation to official surveys.”
But senior lawmakers such as Lord Bridges said progress dealing with the UK’s “shoddy data” is urgently needed.
“This is mission critical.” he said. “This is not a peripheral matter that we can leave just to statisticians.”