Questions Around the ONS Adjustment to Working-Hours Data
The recent ONS revision of UK working-hours and productivity data raises an interesting question about how annual leave is being estimated.
The ONS uses Labour Force Survey data when calculating annual leave for its hours-worked estimates. However, the survey data appears to show that workers have been taking less and less time off, at one point implying levels below the statutory holiday entitlement.
That result seems difficult to reconcile with reality, so the ONS has treated the difference as likely under-reporting and adjusted the figures.
The problem is that this adjustment appears to assume that the missing holiday is broadly equivalent to the workers' annual entitlement. In other words, the original survey result is considered unreliable, but the replacement figure is also partly based on an assumption rather than directly observed holiday-taking data.
This matters because annual leave feeds directly into the estimate of hours worked, which in turn affects productivity statistics. A change in the holiday assumption can therefore change the apparent productivity performance of the UK economy.
I'm not saying that the ONS should never adjust survey data. Correcting for known measurement problems is normal in official statistics. The interesting question is whether there is enough independent evidence to justify this particular adjustment.
If the Labour Force Survey is genuinely under-reporting annual leave, it would be useful to see evidence showing the size of that bias and why the chosen correction is appropriate. Otherwise, there is a risk of replacing one questionable measurement with another estimate that simply produces a more plausible-looking series.
The key issue, therefore, isn't that the ONS changed the numbers. It's how confidently we can say that the revised numbers are closer to reality than the original survey results.
Questions Around the ONS Adjustment to Working-Hours Data
The recent ONS revision of UK working-hours and productivity data raises an interesting question about how annual leave is being estimated.
The ONS uses Labour Force Survey data when calculating annual leave for its hours-worked estimates. However, the survey data appears to show that workers have been taking less and less time off, at one point implying levels below the statutory holiday entitlement.
That result seems difficult to reconcile with reality, so the ONS has treated the difference as likely under-reporting and adjusted the figures.
The problem is that this adjustment appears to assume that the missing holiday is broadly equivalent to the workers' annual entitlement. In other words, the original survey result is considered unreliable, but the replacement figure is also partly based on an assumption rather than directly observed holiday-taking data.
This matters because annual leave feeds directly into the estimate of hours worked, which in turn affects productivity statistics. A change in the holiday assumption can therefore change the apparent productivity performance of the UK economy.
I'm not saying that the ONS should never adjust survey data. Correcting for known measurement problems is normal in official statistics. The interesting question is whether there is enough independent evidence to justify this particular adjustment.
If the Labour Force Survey is genuinely under-reporting annual leave, it would be useful to see evidence showing the size of that bias and why the chosen correction is appropriate. Otherwise, there is a risk of replacing one questionable measurement with another estimate that simply produces a more plausible-looking series.
The key issue, therefore, isn't that the ONS changed the numbers. It's how confidently we can say that the revised numbers are closer to reality than the original survey results.