There are two main approaches to these Brexit loss estimates, and they both have significant challenges, which is why they shouldn’t really be taken seriously. They’re illustrative (at best), rather than instructive.
The first approach is to construct what you might call a doppelgänger economy, by aggregating together a number of different countries which, together, have a similar economic structure to the UK. You then compare the growth rate of this doppelgänger economy to the UK and see if the UK growth rate has diminished relative to that benchmark post-Brexit. Obviously the issue here is that you are effectively attributing any UK shortfall to Brexit, which seems an unrealistic assumption when you think how COVID and then the war in Ukraine will impact different economies. Germany’s recent struggles following the loss of Russian gas show how misleading this can be.
The alternative method is to take a production function approach, and make assumptions around how Brexit might diminish UK productivity growth and/or the terms of trade (price of UK exports relative to the price of imports). This approach is more commonplace, but again the issue here is that you need to make some very bold assumptions around variables which are extremely difficult to forecast, and precisely isolating any Brexit impact isn’t really possible if the broader environment remains volatile.
The NIESR paper for instance assumes that Brexit will reduce the UK’s terms of trade by 1% over the long-term, but the surge in goods and energy price inflation which followed COVID and then Ukraine reduced the UK’s terms of trade by as much as 5% on some estimates. Likewise the OBR assume that Brexit will reduce productivity growth by a further, cumulative 2.4% over the long run. But this 2.4% figure is actually far smaller than most of the forecast errors surrounding the OBR’s five-year ahead productivity growth estimates, and so again it seems improbable that a precise figure can be attributed to Brexit.
Basically the more you look into these estimates, the less significance you attach to them.