Industry Comment

The problem in planning for the future is that all predictions are likely to be wrong, says Stephen Nicol.

‘Events, dear boy, events’. The words of Harold MacMillan may be ringing in the ears of forecasters as life’s tendency to spring surprises on us can confound even the most carefully laid plans.

The so-called science of forecasting has certainly been undermined by recent events. Society relies on long-term forecasts for a whole range of decisions - planning for transport, public services, future house-building needs, and employment and retail space. Large amounts of public and private sector funding are committed and huge areas of countryside swallowed up on the basis of such predictions.

As an economist and observer of economic development over the last 25 years I’ve seen the successes and failures in forecasting, and its use and abuse.

The key problem is that forecasts are inherently unreliable. At the heart of all forecasting models is the assumption that the past provides a reasonable basis to predict the future. This is a fundamental error.

Firstly, it depends on the period of the past you choose. Anyone who has worked in the Treasury knows the truth is based on the right end points. For instance UK growth was 3 per cent from 1991 to 2007 - the trough of the last recession to the most recent peak - but from peak to peak (1989 to 2007) was 2.5 per cent, and peak to trough (1979 to 1991) was 2 per cent.

Secondly, there is a great danger in taking a period when the patterns cannot be replicated in future. Over the last 15 years retail spending has grown at unsustainable levels as we splurged on credit. As we are finding to our cost, we cannot simply extrapolate that pattern into the future.

This matters as key decisions about new retail space are based on assessments of capacity. If too much new space is developed, then there is a risk of harming trade at existing centres. In my view the planning system has almost certainly allowed too much retail capacity. Spending growth will be much more modest than forecasts would have had us believe.

Similarly, population forecasts are placing too much weight on recent trends in migration. In 1996, the Office of National Statistics was predicting that between 2006 and 2021, England’s population would grow by around 2 million. Five years later the growth forecast for the same period had increased to 2.7 million. The most recent 2006-based projections suggest growth will be a whopping 6 million.

Why is this? Essentially the higher recent predictions were largely driven by in-migration from the new EU states as a result of the UK’s economic boom. But such rapid economic growth and so sustained net in-migration was not sustainable. The consequences are that we are planning for too many houses in an already crowded island.

Finally, when making forecasts for over 10 years we run the danger of assuming that the economic growth model that has worked in the recent past will work in the future. This does not take account of “structural breaks” - where something creates an unexpected change in the economy, such as the oil prices in the 1970s.

To my mind we are now facing one of the biggest structural breaks ever in the form of climate change. The ‘business as usual’ model of economic growth looks unstable beyond a five to ten-year horizon.

The UK has had more than its share of investment decisions based on unrealistic forecasts. Kielder Water – one of Europe’s largest man-made lakes - was planned in the late 1960s to cater for substantial growth in heavy industries, which never materialised. The North East has been left with a plentiful supply of water and a useful tourism asset, but with hindsight should the project have gone ahead?

The three-step solution

So what is to be done? It’s easy to identify the problems, but much harder to provide solutions. There are three ways we can be smarter about the use of forecasts:

1. We must acknowledge that forecasts have a large degree of uncertainty. The Bank of England should be commended for its now famous fan graphs projecting possible inflation and economic growth. This ought to be mandatory for all longer range forecasting.

2. Even if clear probabilities cannot be attached to different forecasts at least there should be plausible high and low or pessimistic and optimistic scenarios around any central case.

3. Policies need to be seriously stress tested against scenarios, especially the more pessimistic ones given our inherent tendency to be over optimistic. What would be the impact of a large new retail development on neighbouring town centres if retail growth was lower than expected? What are the perils of allocating too many houses to an area? Could it lead to urban cramming or green belt incursion?

We must remember that all forecasts are likely to be wrong. It’s a question of how wrong and what are the consequences of making decisions based on the wrong forecasts.

Stephen Nicol is managing director of Regeneris Consulting