The worst forecasting error in British government history may be unfolding as we speak. While much attention is given to grand projects, such as HS2, which end up costing tens of billions of pounds more than they were supposed to, these at least have a start and finish date – and something tangible emerges at the end. The same can’t be said for forecasting errors involving even more complex, politically contentious issues – such as what migration might actually be costing Britain.
When the full accounting is done, the fiscal impact of migration may well prove to be one of the biggest misestimates of all
The notoriously wobbly Covid epidemiological scenarios are often seen as the archetypal example of the state’s failure to understand reality, but the various immigration fiascos over recent decades may be a better candidate for proving this point.
When the health and care visa was introduced in late 2020, 43,000 people were expected annually. The state then forecast that ‘thousands’ more would arrive after the visa was expanded the following year. But in the government’s own ‘Why do people work in the UK?’
pages, we learn that 118,522 visas were granted to main applicants on the skilled worker–health and care visa, with 209,638 visas granted to dependents on the same route, for the year ending March 2024. In other words, the state was wrong by a multiple of as much as eight.
Another drastic underestimate can be found in a 2003 Home Office-commissioned study that predicted only 5,000-13,000 people would come to the UK annually after the EU enlargement to eastern Europe. The paper’s executive summary noted cockishly that ‘even in the worst-case scenario, migration to the UK as a result of Eastern enlargement of the EU is not likely to be overly large.’ The reality, of course, was rather different.
‘Not likely to be overly large’ is the epitaph for many of these errors, not because they err uniformly on the upside, but because the cadence of the phrase itself – bland, sure, measuring – seems to capture the attitude that led to them. Forecasts can be wrong for all sorts of reasons, and wrong forecasts are not necessarily bad forecasts. But many of these examples were likely to have been bad because they leaned on unrealistic past-will-continue assumptions despite obvious shifts in the policy environment, while offering confidence intervals that were just too narrow to capture the unfolding reality. As a result, they proved to be a nonsense.
There is another way estimates can mislead: that is if the experts are made to answer the wrong question. The best example of this, in my mind, is the estimate for the fiscal impact of migration.
The Office for Budget Responsibility has consistently shown positive impacts to the public finances from net migration. In March 2024, the OBR revised up net migration and estimated this would cut borrowing by £7.4 billion annually in 2028-2029. Yet this measured impact over a five-year window, even though, as Alan Manning’s excellent new book Why Migration Policy Is Hard points out, most economists think the fiscal impact of migration should be measured over a life. After all, working-age people earn income and pay taxes while consuming fewer public services; as they get older, the opposite generally happens.
The government’s Migration Advisory Committee (MAC) has started to do some thoughtful modelling of different migrant cohorts. Recently, this showed that those arriving as spouses in 2022/2023 would cost the state £5.6 billion or £109,000 per person over their lifetime (excluding dependents). But, if you just counted the first two decades after arrival, this figure would be positive £35,000. Positive first, then deeply negative. It is a perfect example of how a five-year window is irrelevant at best and misleading at worst. Of course, the economists at the OBR understand this and have even written about it, even if they have been asked by politicians to answer a different question.
The worst mistakes come from forecasts whose construction is not properly understood. Go back to the OBR’s £7.4 billion additional contribution from migrants. That is constructed from the extra 350,000 people expected to be in the country over the subsequent five years, relative to the November 2023 forecast. That £7.4 billion is made up of £6.5 billion receipts, negligible welfare or public spending and reduced debt interest of £900 million (because of lower borrowing due to the higher receipts). Yet about half of the £7.4 billion could vanish under two simple tweaks.
Another drastic underestimate can be found in a Home Office-commissioned study that predicted only 5,000-13,000 people would come to the UK annually
For a start, should the government spending assumption really be close to zero? The OBR says that because its job is to evaluate the impact of government choices, it cannot automatically assume the government will increase departmental spending. Fair. The OBR then provides sensitivity analyses which shows that if per capita departmental spending was maintained for the 350,000 people, it would remove about a third of the £7.4billion gain.
The OBR also points out that most migrants will be ineligible for welfare spending until they have indefinite leave to remain, something that happens after five years. Yet someone trying to understand the permanent impact on the public finances might want to consider what happens from year six onwards. Moreover, the five-year limit isn’t universal; it doesn’t apply to those with humanitarian or asylum visas; at the time they comprised about 20 per cent of recorded inflows. It doesn’t apply to certain nationalities on certain benefits; Moroccan or Turkish workers, for example, can claim child benefit. Ultimately, if you assumed people ended up claiming the same non-pensioner welfare spending as the average non-pensioner, it would wipe out around £800 million – and combined with the above, just under half of the £7.4 billion number. That ‘very small’ welfare assumption number – OBR language – needs to be unpacked. One could also interrogate the £6.2 billion receipts number.
The true challenge to public finances lies, as ever, beyond the five-year forecast window. The MAC’s estimate of lifetime costs of just the 2022/23 intake of care workers was £2 billion. If we assume that each of the 55,000 care workers in that cohort brought with them one child and one adult dependent, the lifetime cost of that intake would be £7.5billion. If you add the spousal visa and indicative estimates for the humanitarian and asylum visas based on experiences in Australia and the Netherlands cited by the MAC, the figure conservatively comes to £20 billion. Five years of that kind of flow of care workers, spouses and asylum seekers would cost £100 billion, over twice the error on HS2.
When the full accounting is done, the fiscal impact of migration – widely and blithely assumed to be overwhelmingly positive, but already known to be deeply negative in parts – may well prove to be one of the biggest misestimates of all.
