Ahead of the referendum in June this year, various institutions estimated the impact on the economy of a decision to leave the EU, relative to the outlook if we remained a member.a These studies identified a number of channels through which this impact might be felt. Among them:
- capital deepening: firms will be less likely to invest during periods of heightened uncertainty. Lower business investment reduces growth in the stock of capital and therefore the amount of capital available for each unit of labour (worker or hour worked). That would reduce labour productivity growth and therefore potential output growth;
- net migration: stricter controls and/or a reduction in the attractiveness of the UK as a destination could reduce net inward migration. That would reduce growth in potential output via lower population growth. Depending on the age- and skill-characteristics of the prevented or deterred migrants, it could also positively or negatively affect the employment rate or productivity growth;
- openness: a less open economy – in terms of trade and/or foreign direct investment –could reduce labour productivity via ‘total factor productivity’ (the amount of output an economy can produce from a given level of labour and capital inputs). For example, less trade may slow the process of specialisation in the activities where firms are most productive. It could also reduce the extent to which firms are able to adopt or adapt techniques and processes from overseas trading partners;
- research and development: as with business investment, greater uncertainty could reduce investment in R&D. That would affect potential output growth via total factor productivity, potentially with a longer lag than slower capital deepening; and
- adjustment costs: during the transition to new trading arrangements with the EU, some firms may need to focus resources on creating new products or entering new markets. That could reduce productivity temporarily as firms use some resources for adjustment rather than producing output. This process could involve a greater turnover of firms closing and new firms being set up. As new firms tend to be less productive initially, this could have a temporary compositional effect on productivity.
These channels reflect the relatively direct consequences of leaving the EU on flows of goods, services, investment and people across the UK’s borders. But some studies have also looked at how policy settings in the UK could change after leaving the EU, including trade agreements with non-EU countries or the removal of regulations associated with EU membership. If such policy changes generate new productive opportunities, they could increase productivity along similar channels. That said, there is no guarantee that policy would move in this direction.
The weight placed on these different channels is largely a matter of judgement. Most studies assume some effect via capital deepening, although the scale varies. Only the OECD study attempted to quantify a migration effect. There is a degree of consensus that leaving the EU will reduce openness, although the scale depends on what trading arrangements are assumed to prevail afterwards. There is much less agreement on whether that will affect productivity – this channel was an important element of the Treasury’s analysis, but NIESR chose not to include it. The effect from other channels is typically estimated to be quite small.
There are, of course, huge uncertainties associated with any estimates of the effect of leaving the EU, since it is not something that has happened before. The sources of uncertainty include what will ultimately replace EU rules in terms of trade, investment and migration, as well as any knock-on effects to regulatory or other policies. The latter are less relevant to our analysis, as we are required to forecast on the basis of current policy rather than to predict how governments might choose to exploit the opportunity to change policies in the future.
For any future policy setting, there is then uncertainty about the extent to which the economy might be affected along the channels identified here or in other ways. And, for a 5-year forecast like ours, there is additional uncertainty about the timeframe over which any effects will take place and how much of it will occur within the forecast horizon. Finally, because we cannot observe potential output directly, or know with confidence what would have happened otherwise, the true scale of any effects will never be known with certainty.