Net migration, defined as the difference between immigration and emigration, is a key driver of population growth and hence labour supply. This makes it a key driver of our forecast for the level of potential output and, therefore, potentially an important driver of the fiscal outlook.¹ We follow the ONS definition that a person should be in the country for over 12 months to be considered a long-term migrant and counted in the statistics. By this definition, net migration accounted for around two thirds of growth in the adult population over the OBR's March 2026 medium-term forecast. As part of our response to recommendations in our most recent external review, this article sets out our process for forecasting net migration, including our use of a stay-rate model and of the ONS National Population Projections (NPPs).²

The OBR has previously drawn on the ONS NPPs as our main source for assumptions on medium-term net migration.³ These are typically published only every two years, which makes them more suitable for forecasting when net migration outturns and policy are relatively stable. However, net migration has been exceptionally volatile in recent years. It increased significantly following the pandemic – substantially driven by increases in humanitarian visas (particularly from Hong Kong and Ukraine); international students; and dependants of those coming to work or study. The ONS estimates that net migration reached a record high of 891,000 in 2022, compared with an average of around 240,000 a year between 2012 and 2019 (Chart 1). It fell to an estimated 171,000 in 2025, driven by the tightening of visa rules in recent years – which lowered immigration – and a rise in emigration – as a portion of those in the previous surge in immigration subsequently left the country.

Given the scale and pace of recent changes in net migration, and the lag between updates to the NPPs, the OBR has developed an in-house migration model to help incorporate the latest data and policy changes in a timely manner. We use this model to inform the forecast for net migration which, like all other variables, will ultimately reflect the collective judgement of the Budget Responsibility Committee (BRC).

The OBR’s forecasts for net migration are informed by a stay rate-based model, following a ‘bottom-up’ approach. The model requires assumptions about future immigration flows, and the proportion of immigrants who remain in the UK a given number of years after their arrival, called a ‘stay rate’. These estimated stay rates provide information on both how many migrants are likely to leave and when they are likely to leave. The model uses these stay rates to forecast future emigration flows.

To forecast immigration, migrants who arrive with a visa are grouped into a small number of broad categories – including those for work, for study and dependants. For each category, immigration is then forecast using recent trends in visa applications and grants, adjusted where appropriate to reflect the share of visas holders who are expected to become long-term migrants and the impact of policy changes.⁴ This means we assume the volume and composition of immigration remain broadly unchanged across the forecast in the absence of stated policy changes, as shown in Chart 3.

The model forecasts future emigration flows using estimated stay rates for each visa category over the next ten years. These are set out in Chart 2. We assume that nearly half of those arriving on study visas return to their country of origin after six years. That proportion is closer to 30 per cent for workers, 15 per cent for those on dependent and family visas, and 10 per cent for those on humanitarian visas. These are estimated using Migrant Journey data which records migrants’ visa and leave status at the end of every year since they initially entered the UK. We use an average of historical stay rates from 2004 onwards as a starting point but place greater weight on more recent cohorts in the first five years after arrival. This reflects evidence that workers and students arriving under the post-Brexit migration system have been less likely to leave the UK. We also adjust stay rates beyond the first five years after arrival to reflect our judgement that the higher stay rates observed among recent cohorts are likely to persist. Stay rate assumptions are a key input into the model, but they are also highly uncertain. Stay rates depend on economic and social developments in the UK and abroad, alongside the characteristics of migrants – which could change significantly in response to migration policy.

To forecast the net migration of individuals without visas, such as British nationals and EU nationals who arrived under the pre-Brexit regime, the model also requires assumptions about future net migration flows. In the March 2026 forecast, we assumed that net migration of British nationals would return to around -85,000, i.e., a net outflow, consistent with the average from 2012 to 2024, the longest period supported by ONS data. Recent years have seen negative net migration of EU nationals. Given that EU nationals can no longer arrive without a visa and that there is a fixed number of people with pre-settled or settled status, we assume net migration of this group gradually increases to zero by 2028.

The model combines the projected immigration and emigration from both visa and non-visa categories to produce a forecast for net overall migration over the medium term (Chart 3).

A stay rate-based model is most accurate for forecasting net migration over the near term because uncertainty increases the further the forecast moves away from the latest data. This reflects the structure of the model, which requires separate forecasts for immigration and emigration. In the near term, current government policy is known, timely visa data provide a reliable indicator of immigration, and stay rates are applied to outturn immigration data. Further into the forecast period, stay rates are increasingly applied to forecast immigration which can compound differences. The net migration forecast over the longer term is subject to greater uncertainty because it will be heavily influenced by future government policy, whereas our forecasts are conditioned on stated policy at the time of producing the forecast.

At longer horizons, where uncertainty is substantially greater, the OBR has typically aligned its assumptions with one of the ONS NPP variants as the BRC have judged these to be a reasonable projection for migration. The ONS projections draw on advice from their migration expert advisory panel and are widely used across government. They also provide projections by single year of age, which are useful for forecasting participation rates using a cohort-based approach. However, recent development of our cohort-model infrastructure has reduced the importance of aligning with the ONS projections.

As the net migration forecast ultimately reflects the collective judgement of the BRC, we may diverge from these ONS projections in the medium term if judged appropriate to do so. This could be due to policy or data changes since the last projections. Past examples of this include:

• In the Autumn 2025 forecast, we expected a reduction in future work visas due to the Immigration White Paper measures enacted in July 2025, including tighter eligibility requirements for skilled worker visas and the end of the care visa route. However, the latest evidence at the time suggested that stay rates under the new migration system had risen by more than we had assumed in our March 2025 forecast. These higher stay rates resulted in lower emigration over the medium term which offset the effect of lower immigration, meaning our net migration forecast was unchanged. Such offsetting changes are unlikely to occur in all circumstances.

• In the March 2026 forecast, we used the migration model to inform the forecast over the whole five-year period following an ONS change to the methodology for estimating migration flows of British nationals.

Overall, our approach is to produce a judgement-based forecast for migration which is informed by the information contained in the ONS projections, the outputs of the stay-rate model explained here, alongside analysis of the impact of any stated policy changes and other factors that affect migration. We also hold discussions with external migration experts to inform our judgements. In all cases, we transparently explain the changes we make to our migration forecast in each Economic and fiscal outlook, along with their economic and fiscal consequences.

Acknowledgements

The author would like to thank OBR staff for their valuable contributions to this article.

Downloads

Forecasting net migration (article PDF)