Next year’s GDP and that puzzle that is still poorly understood

The deficit of 3.1 per cent has dominated the analysis of the figures published yesterday by Istat, fuelling the usual debate over the decimal places that are supposed to authorise new spending. Returning below 3 per cent would have significance for the European infringement procedure, but to imagine that this would throw open the doors to the budget is to ignore both Italy’s public debt and the workings of the new Stability Pact, which constrains the trajectory of net expenditure even after the country has exited the infringement procedure. Moreover, hoping to remain below 3 per cent in order to run a further deficit – the implicit assumption underlying the prevailing view – continues to point to a country suffering from public debt.

Rather, yesterday our attention was focused primarily on the GDP revisions, because these go some way towards explaining the so-called ‘employment and output puzzle’ – a term introduced by this columnist in these pages several years ago.

A brief summary of the issue. From 2022–2023 onwards, GDP has generally grown at a slower rate than employment, resulting in a decline in output per person in employment. This outcome attracted a great deal of attention, particularly because businesses’ ability to hire seemed to coexist with an economy that was virtually at a standstill. Hence, in fact, the term ‘puzzle’ – which, at the time, we had suggested could be explained, at least in part, by an underestimation of output.

First of all, however, it is important to remember that the starting point matters. If we set real GDP and the number of people in employment in 2019 at 100, the former still exceeds the latter today; consequently, productivity per employee is now higher than it was before the pandemic, which contrasts with the widespread perception (hourly productivity would be a separate matter). To put it another way, the decline over the last 3–4 years must be viewed in the context of the previous increase; otherwise, we risk making the mistake made in recent years by many who have confused a temporary dip with a judgement on the entire period.

Against this backdrop, yesterday’s new Istat estimates help to piece together part of the ‘puzzle’. Growth for 2025 has in fact been revised upwards from 0.5 to 0.6 per cent, whilst that for 2024 has risen from 0.8 to 1.1 per cent, with revisions of one and three tenths of a percentage point respectively. In 2024, the gap between economic growth and employment growth narrows significantly, easing the puzzle precisely in a year when it appeared particularly evident. The revisions affect both real and nominal GDP and involve various sectors (primarily construction, but also professional, scientific and technical activities).

This is not a criticism of Istat. On the contrary, as time goes by, the institute obtains financial statements, tax information and survey results that enable it to replace preliminary estimates with more comprehensive figures. Following the pandemic, changes in the composition of demand and output have probably reduced the reliability of the data used in the initial estimates. For this reason, the revisions we have just received come as no surprise; indeed, as already mentioned, we had largely anticipated them.

However, a question should be put to the statistical agencies. The frequency of upward revisions in the post-pandemic period suggests that it is worth checking whether the initial estimates systematically tend to underestimate economic activity. An average error significantly different from zero generally indicates a statistical distortion, and the fact that these errors are predictable on the basis of information already available suggests there is scope for improving the methods. In the case of Italy’s GDP (as in that of other eurozone countries), both factors are present; whilst this does not necessarily demonstrate a flaw in the models used by Istat, it does, in any case, warrant a rigorous review.

In conclusion, we would like to emphasise that these revisions reinforce the view we have long held, namely that the recovery from 2020 onwards has been robust and largely underestimated. This finding leaves the question of future growth open, but requires us to update our assessment of the past with the same diligence with which we comment on disappointing data. To paraphrase a great economist of the past, ‘if the evidence changes, I change my mind’.