Actually, the OBR have explained it. In quite some detail. In their 246 page report, which is available on their website.
But in essence, it's this :-
- any economic forecast is an inexact science
- you design a model, collect the most accurate data you can, and feed it into the model,
- you then constantly review and revise both the model and data collection methods to improve accuracy.
Over several years, a variety of changes have occurred. The ONS changed how they categorise housing association receipts, the OECD studies of the comparison of international forecasting showed most countries underestimated GDP growth, HMRC changed NIC contracting out rules, and so on.
So, given that all economies are so vastly complicated than any economic forecast is only a fairly crude tool, forecasters are constantly reviewing models to take into account changes to how data is collected (by third parties) and how well existing models used historic data to accurately predict future events which, with the passage of time, have become past events about which actual outcome data is now available for comparison with forecasts. Where actual outcomes suggest model outcomes are wrong, and wrong consistently, you look for the source of that problem, and address it, either by improving data collection, of the model analysis.
It's really a bit like weather forecasting. The more data you have, and the more accurately its collected, the better the analysis can be, mainly because it allows you to fine tune your models to correct mistakes, or inaccuracies, in previous models. As a result, both economic and weather forecasting are considerably better than they used to be, but both are still inexact sciences.
Given that we have difficulty predicting weather with any certainty more than a few days away, we have to understand what economic forecasting several years away really is, which is the best we can do of a very, very difficult task. But providing model changes are made based on an improving accuracy and breadth of data, and understanding of their impact, and are designed to improve the accuracy, plaudibility and transparency of the models, then sometimes, changes to existing models fed with recent data will result in a re-analysis of near future outcomes better than the previous model suggested.
And that's what happened here. Improvements in the model since the July forecasts changed the impact of some VAT classification changes and NIC Class 1 contracting out contributions, over forthcoming years but with the biggest impact in about 3 or 4 years time. Those are two of the biggest changes, and the OBR explain them in box 4.2 of the report I mentioned (page 105, if I remember correctly) with actual data in surrounding charts and tables, and the other changes explained in similar boxes, charts, tables and voluminous text throughout that (extremely dry) report.
For the cynics that think "how convenient for Osborne", well maybe. But while he's seen various drafts of this OBR report over several weeks, the actual final signed-off report wasn't concluded until November 20th, and handed to Osborne Nov 21st. So he had the final, confirmed forecasts for a total of 4 days when he have his statement and spending review.
