by Business Editor David Murphy

Economic forecasts can have a major impact on policy decisions

Later this month the Department of Finance will update its forecasts for the Irish economy.

Economic forecasts matter; yet they are frequently wrong and fail to predict important developments.

Projections for economic growth, inflation and unemployment are used as the basis for assumptions upon which national budgets are based.

But they can also form the rationale for wider policy decisions, such as investment in long-term capital projects.

Perhaps the most crucial part of a forecast is its outlook for economic growth. For a government, that determines how much tax revenue it can expect to collect, how many people will be unemployed and the cost of social welfare.

If the forecast is too optimistic the state runs the risk of a deficit, being obliged to borrow money and pay it back with interest. That development alone can have a significant negative impact.

But it is worth considering where forecasts for Ireland went wrong and why.

In an academic study I completed last year, I examined predictions for the Irish economy between 2002 and 2011.

The thesis looked at the track-record of the Central Bank, the OECD, the Department of Finance, ESRI and European Commission. It compared their forecasts for unemployment, inflation and economic growth with figures from the Central Statistics Office.

There are various ways to measure forecasts.

The study used two methods, a mean square error (MSE) technique and a root mean square error (RMSE). The former penalises errors in forecasts to a greater degree.

The theory behind this is that significant errors in economic forecasts are more damaging than smaller ones, particularly when policy makers rely on the projections for what will happen in an economy.

Errors in forecasts from 2002 to 2011

The chart above examines forecasts which looked 12 months ahead. If a prediction was absolutely accurate, the mean square error would read zero. The chart shows forecasters missed the crash.

To put this in context, the Department of Finance expected the economy to expand by 3% in 2008 in terms of GDP, but in fact it contracted by 2.1%, according to the CSO.

It is also clear that, during the economic crash, forecasts for unemployment and inflation were more accurate than for GDP.

It is not surprising that errors for unemployment were lower than for GDP as a large fall in economic activity is usually followed by a rise in the number of people out of work.

Despite the fact that unemployment lags economic growth, the rapid surge in jobless numbers in 2008 and 2009 clearly caught forecasters unawares.

Which components of GDP were the most difficult to predict

GDP is made up of five components: private consumption, government spending, investment, exports and imports. The chart above illustrates that investment proved the most challenging of the five to predict accurately.

When the crash began in 2008 the forecasters underestimated the collapse in investment. The errors for investment escalated in 2008, 2009 and 2010.

Lessons for policy makers

The most important issue for policymakers is that if projections can be wide of the mark, how can any government which relies on forecasts say with confidence how much tax revenue it will collect, or how much it may have to spend on social welfare when people lose jobs?

Ireland was not the only country let down by misleading forecasts.

While debate and research will continue into how to improve economic models, until better forecasting methods are developed, accepted internationally and used widely, policy makers are left with projections that can be unreliable.

That means governments should consider running public finances to allow for the fact that budgetary expectations may not be met.

In Ireland’s case, immediately before the crash the Budget had no buffer to allow for significant erroneous projections.

As a result, when the State’s tax base crumbled it instantly plunged into a fiscal crisis. If the country had run a surplus it would have been better insulated from the effects of a downturn.

The €62 billion recapitalisation of the banks and the funding of the gap between spending and revenue left no room for a stimulus.

Many were surprised when the projections for Ireland were so inaccurate. Perhaps there is a lesson here for forecasters.

Greater emphasis on the fact that a prediction is simply one of a number of possible outcomes might drive home the message that politicians and the civil service should not read too much into forecasts.

UCD Professor Morgan Kelly and a few others correctly predicted the scale of the crash, but their warnings were dismissed by the authorities in 2007 including the Central Bank.

Speaking to the Oireachtas Committee on Finance in June 2010, Central Bank Governor Patrick Honohan commented that Professor Kelly was a “well-known clever guy who should have been paid more attention to.”

Governor Honohan pointed out that while contrarians are nearly always wrong, this one was right. Forecasters can be susceptible to herd behaviour.

Another lesson is that in times of uncertainty close attention should be paid to arguments running counter to the consensus of economic forecasters.

Covered banks’ customer loans 2000 – 2008 (annual reports & Eurostat, cited in Nyberg report on banks)

There was publicly-available information about the financial sector which showed there had been an enormous expansion of Ireland’s banks.

The chart above shows the growth in loans to customers from Irish banks in comparison to the size of the economy. This credit boom fuelled the construction industry.

The failure of forecasts to capture this information was one of the reasons why the projections did not predict the economic collapse. This lapse highlights the need for models which reflect a flow of funds to aid forecasters to assess the impact of the financial sector on the economy.

But more work is needed in this area before forecasters will be happy to use such models.

It is dangerous for governments to rely entirely on projections.

Forecasts differ from many other types of economic analysis because they depend, to a certain extent, on the judgment of individuals.

And as Ireland’s recent experience lays bare, that is not reliable either.

David Murphy will be speaking at the Cantillon conference on Economy and Entrepreneurship about forecasts on Friday April 12 in Tralee, Co Kerry.