United Kingdom: 9 years in a row, borrowing above the Treasury’s forecast
United Kingdom: public borrowing, 2001–2009
2001–2009: every year both next-year forecasts (Budget and autumn) missed the same way, by £24.7bn on average.
More
HM Treasury until 2010, then the independent Office for Budget Responsibility, publish an economic and fiscal forecast with each Budget and autumn statement. Here: the next-year forecasts against the outcome.
By target year: blue, the last forecast for that year; orange, the outcome. The shading is the error; all these years it stays on one side.
- November 2000 forecast for fiscal 2001/02
£−5bn → £5.6bn
- March 2001 forecast for fiscal 2001/02
£−5bn → £5.6bn
- November 2001 forecast for fiscal 2002/03
£12bn → £35bn
- April 2002 forecast for fiscal 2003/04
£13bn → £43.5bn
- November 2002 forecast for fiscal 2003/04
£24bn → £43.5bn
- April 2003 forecast for fiscal 2004/05
£24bn → £51.7bn
- December 2003 forecast for fiscal 2004/05
£31bn → £51.7bn
- March 2004 forecast for fiscal 2004/05
£33bn → £51.7bn
- December 2004 forecast for fiscal 2005/06
£33bn → £45.4bn
- March 2005 forecast for fiscal 2005/06
£32bn → £45.4bn
- December 2005 forecast for fiscal 2006/07
£34bn → £40.9bn
- March 2006 forecast for fiscal 2006/07
£36bn → £40.9bn
- December 2006 forecast for fiscal 2007/08
£31bn → £45.1bn
- March 2007 forecast for fiscal 2007/08
£34bn → £45.1bn
- October 2007 forecast for fiscal 2008/09
£36bn → £116bn
- March 2008 forecast for fiscal 2008/09
£43bn → £116bn
- November 2008 forecast for fiscal 2009/10
£118bn → £160bn
A year counts when every forecast of it made a year earlier missed the same way by more than 0.05; two editions for one year are one check. Years with a known outcome: 52. Ten selected long runs are shown. Error signs can be dependent; countries, variables and windows were searched. The probability of such a run arising by chance has not been estimated here. Outcome: OBR outturns (March 2026) or ONS.
Fiscal forecasts assume announced policy; new laws, crises and statistical revisions (new accounting rules) count as “error”. A run of errors with the same sign can reflect shared assumptions, shocks or data revisions; this selection does not separate them or rule out chance.
A question for this picture
What explains the run of errors: assumptions, shared shocks or data revisions?