Why do we Need Forecasts?
Because the only constant in life is change, and while change is certain, the form of change isn’t. We thus require information to enable a more informed view of the future. We can gain a more informed view via in-house resources, although in reality, companies lack funds, and suffer bias when they take this route. Better, therefore, to consult the experts!
But what type of forecast do you need? This depends upon the use – long term, short-term, whatever. Short-term housing forecasts will have less of a reliance on demographics, and more a reliance on income ratios, expenditure etc, and the overall state of the economy.
Types of Forecast
Econometric models
Use a vast array of variables to determine projections. They suffer from specification error (inability to establish correct variables), a certain rigidity (relying on the past), and cost problems. Most models require constant inputting; in essence they entail at least one full time job running the model. They are also entirely logical in their approach – but as we know, society often acts in illogical ways, best defined by the quote – the difference between truth and fiction is that fiction has to make sense.
Simple regression based forecasts combined with human intuition:
These simply rely on past relationship to determine future pattern, in other words straight forward extrapolation. But the equation can be tweaked when human judgement allows or suggests. This is less complicated than the above/model approach, but again, suffers from relying on past relationships, etc.
Panel based forecasts:
Similar to the Delphi approach, where a panel of experts is surveyed, and the results compiled. The panel type forecast is common in construction. They have their merits, and are cheap to run. However, it is very people orientated, and strength of character in a meeting can sway a forecast. It can also result in mud – if you combine lots of colours, one gets a mud colour!
So what is the Best Approach?
Making accurate forecasts begins with selecting reliable and/or relevant data. Construction orders data are relevant, but less reliable than they used to be. House price data are relevant, but less so than used to be the case.
Deal with the figures in your own mind – do they all make sense? What data is of importance to me? Remember, you are trying to be a bit better at understanding and predicting the market, not providing a universal economic model.
Track the macro economy: but be specific, there is a mass of data, it would take all day every day just to sift through it all. If you have too many data sets, it will become too complicated, and you will have to move over to a model which itself can be cumbersome.
Keep your eye on the big picture, and on what is important today – in the ever evolving complexity of human affairs, some factors will come to the fore and others will recede – your job is to become good at judging what is important now, and what will be important tomorrow.
From there build your model, or should I say method. Challenge the results every year – where were we wrong, and why? Then take stock and move on to the next forecast.
Forecasting Construction
We start with the macro economy – viewing investment, consumption, profits, global growth, interest rates, public expenditure etc. But we now need to look more at new factors, notably the role of debt in the economy – since 2000 debt has been a key driver of GDP growth.
Housing: typically this is determined by incomes, prices, interest rates, supply factors, planning, demographics and politics. Now politics has taken over, in the form of Help to Buy, while incomes are less significant. Demographics are clearly important, although they count for little if those who need a house cannot afford one, and the Government is unwilling to fund large-scale public provision.
Infrastructure: this sector is the one least liable to be modelled. One should monitor Government spending plans, corporate utility plans, and political change. Watch out for delays, political double talk (policy by announcement – projects that will never take place but sound good), and incoming Government promises that look likely to fail.
Public Non-housing: monitor Government spending plans, Local authority finances, whilst always keeping an eye out for an end to austerity – the most significant issue since 2010.
Industrial: monitor manufacturing output, exports, profit margins, inward investment, global trade, and structural change in the marketplace i.e. overseas sourcing.
Commercial: monitor rents, availability, development pipelines, employment, costs, structural change (outsourcing, downsizing, home-work,), planning. Consider the impact of IT – what is of interest on this note is how, despite huge advances in IT, its impact on office work has been minimal, if that were not the case, rush hour would not exist!
So, what do Forecasters Offer?
A glimpse of certainty in an uncertain world, or a vision of tomorrow today. Perhaps more importantly, they offer an understanding of how current market forces are operating, and how they are likely to move in the future.
The marketing manager must thus ask: does this forecasting product offer concepts that allow me to understand the current market situation, and possibilities for the future, thus allowing me to move quicker and smarter than the competition?
Importantly, what matters is how forecasts are used. Most offer a single point estimate, but we have to be aware, you cannot always tell both by how much and when a market will move.
So, perhaps the best approach then is for the forecaster to provide alternatives forecasts, or scenarios. There is a problem here, as users require certainty in their forecasts, they often like to be told. They want a single headline forecast. Possibilities require thought and judgement, users may ask: why should I assess possibilities and scenarios when that is what I am paying the forecaster for?
However, users can overcome this by taking an array of forecasts, that is of those views ranging from pessimism to optimism, and taking a stance from then (i.e. take an average).
Pitfalls to watch out for:
Information overload – paralysis by analysis, we have more information than ever, yet our capacity to absorb that has barely increased. One edition of the FT contains more information than the average 18th century person would have absorbed in a lifetime. Look for the overall trend and do not get too absorbed in the detail.
The process is more important than the output – when economists become carried away with methodology, at the expense of the larger picture, they are in trouble, as are their clients!
Right & Wrong: better to be roughly right than precisely wrong.
Finally, the most important ingredient in making forecasts and taking a future based view is the quality of the mind.
Martin Hewes, Managing Director of Hewes Associates is one of the industry’s foremost forecasters, he publishes Construction Outlook which provides three years of forecast data for Housing, Infrastructure, Public non-Housing, Industrial, Commercial, Repair & Maintenance.
