Have you ever wondered how accurate our Labour Market Information forecasts are? We believe that the data we give you on past and present industry and employment trends is second to none, but how do we fare as an economic forecaster?
As we are always at pains to stress, our forecasts are just that: forecasts. Just like weather predictions, economic forecasts will be based on both past and current trends, and so there are at least two issues that prevent accurate forecasting:
1. Not possessing the quality of omniscience, we cannot possibly foresee the unforeseeable. A natural disaster, or an unexpected announcement by a company that it is closing down or even starting up a plant in a region, for example, will skew the forecasts in ways that could never have been predicted.
2. Whilst past and present trends are likely to be a good indicator of where things are going in the short term, obviously the longer range the forecast the less likely it is to be wholly accurate.
Those caveats aside, we want to give you good value from our data, which means we want our forecasts to be as accurate as possible. With this in mind, we recently conducted a study of past predictions we have made, checking our methodology and accuracy against the actual figures, using the Workforce Job Series as a benchmark.
So how did we do? Actually pretty well. Using our current projection methodology, we calculated the absolute error in two, three and five-year forecasts starting from 1998, 2003 and 2008, for 1 digit industries. We did this for Britain, England, Scotland, Wales, and also for all nine regions of England, with the following results:
Percentile % error at percentile after 2-Years % error at percentile after 3-Years % error at percentile after 5-Years
10th 1.2% 1.4% 1.4%
25th 3% 3.7% 3.7%
50th 5.9% 7.5% 9.8%
75th 11% 13.8% 18%
90th 19.7% 22.2% 29.2%
As predicted (pardon the pun), the shorter-term forecasts were much more accurate than the longer range projections, but even in the longer range forecasts there is still a reasonably high degree of accuracy. Some highlights of this data are:
- 25% of yearly predictions are within 3% of the actual future values after two years
- 50% of yearly predictions are within 10% of the actual future values after five years
- 90% of yearly predictions are within 22% of the actual future values after three years.
Although we are always looking for ways to make our forecasts achieve an even higher degree of accuracy, we hope that these figures will give you reassurance that the projected data you see in Analyst can be relied on to give you a good idea of the direction that various industries and occupations are heading in.
For more information regarding our data and methodology, contact Andy Durman at firstname.lastname@example.org