David McNamara: Are economic statistics still trustworthy?
Data has been signalling a cooling of Irish jobs growth in 2025.
The firing of a little-known labour statistician by US President Donald Trump has brought to the fore an ongoing debate amongst economists — that is, whether macro statistics are becoming unreliable as the response rates to key surveys decline.
In the US, major revisions to the benchmark payroll jobs survey are quite common and are typically shrugged off by investors and grumpy forecasters, whose models then require tweaking. In some cases, the revisions can completely shift the narrative of an economy, as was the case when US payrolls were recently revised sharply lower for May and June, precipitating a fall in interest rate expectations and a decline in the dollar.
Response rates to this key survey have fallen sharply to 30-40% at present from as high as 70% in the period before the pandemic. However, despite President Trump’s assertions of political bias, the revisions in the past few years to the payroll survey have been in both directions, with few signs of bias in the data.
Nonetheless, covid appears to have left a lasting impact on the ability of statisticians to collect data across the world, with another example being the UK’s labour force survey, which the Office for National Statistics has placed a large health warning on.
The response rate to this survey has fallen from a pre-covid rate of 50% to just 10% at present. While the revisions in the US payrolls survey still appear to be random, the UK survey is demonstrating what economists might call ‘systematic error’. In the absence of robust national statistics, a new cottage industry of private labour market data has emerged, which policymakers in the Bank of England, who have been struggling to get a gauge on wages and inflation, now rely on.
In Ireland, the concentration of a small number of very large multinationals often renders initial estimates of GDP useless, but we have also seen large revisions to our labour force survey, including the sharp upward revision in unemployment recently. Emerging ‘frontier’ series, which use administrative data rather than relying on surveys, might become more reliable and prevalent in the future. For example, a new payroll employment series pulls data directly from the Revenue’s PAYE system in Ireland, which gives a timelier estimate of the number of employee numbers.
This series has been signalling a cooling of Irish jobs growth in 2025, which the official labour force survey might now belatedly show when the latest revisions are published later this week. The key takeaway is that economies are becoming more complex and harder to measure in real-time, with emerging trends such as AI having the potential to upend how we live and work in the years to come. Spare a thought for the beleaguered statistician in an ever-changing world.
