Published daily by the Lowy Institute

The dark art of economic forecasting

The dark art of economic forecasting
Published 12 Aug 2013 

Since so much international economic discussion revolves around GDP forecasting, it's worth looking at the quirks and pitfalls of this black art. Yogi Berra famously said 'It's tough to make predictions, especially about the future'. This view has been confirmed by more rigorous analysis, both at the Australian Treasury and the Reserve Bank of Australia. How should forecasters respond to the difficulties?

First, they should be explicit in their assumptions about policy: does the forecast assume 'unchanged policy' or, more realistically, that policy will respond to the unfolding circumstances? A forecast assuming 'unchanged policy' isn't all that helpful if we just want to know how things will turn out.

The Steve Keen-Rory Robertson bet on how Sydney housing prices would behave illustrates the point. Keen predicted that prices would fall by 40%. When he lost the bet (and walked from Canberra to the top of Mount Kosciusko), he explained his mis-forecast in terms of the dramatic reduction in interest rates and other policy changes. Of course policy reacts to unfolding events.

In the current China debate, it's often unclear whether a forecast is based on an assumption of unchanged policy or not. If China slows significantly (the unanimous talking point), policy will respond. The interesting issue is: does China have the capacity to change policy and how effective is this likely to be? Without this discussion, it's hard to make sense of a forecast.

The next question is: 'what biases does the forecaster have?'[fold]

Official forecasters find it hard to predict outliers, especially on the down-side. If they predict an unpleasant outcome, people will ask why they don't do something to prevent it. To make things worse, economic forecasts (unlike weather forecasts) can influence the outcome. Official forecasters don't want to 'talk the economy down'.

International agencies such as the IMF have the same problem, but in spades. If an economy is weak, a gloomy prediction from outside experts will be particularly unwelcome.

It's harder to generalise about the intrinsic biases of private-sector forecasters. Much of the daily deluge of economic commentary is made by poorly resourced forecasters who take the easy way out and follow the herd. This means that forecasts bunch together like a school of small fish, changing direction simultaneously as 'risk on' alternates with 'risk off'.

Some braver forecasters seek fame by boldly departing from the pack. Michael Pettis' prediction of 3% for China's annual GDP growth over this decade may be an example. This approach needs to be combined with other techniques to preserve the forecaster's reputation. Either forecast so far into the future that people will forget if you are wrong, or include a proviso that the timing of your predicted disaster is very uncertain, or make frequent revisions so that reality never catches up with your most recent forecast.

For some market players (eg. hedge funds), it can be good business to press the panic button from time to time. They make money out of volatility. An example was the high-profile panic that surrounded the Greek debt crisis in 2010, when the market threatened that if there were haircuts on their Greek debt, they would ensure disastrous contagion for Spanish, Portuguese and Italian debt.

Market forecasters often reflect their own perspective and time horizon, and shouldn't be over-interpreted in assessing wider economic trends. Recent examples have been the panic over quantitative-easing tapering and the China liquidity panic.

In both cases, financial markets warned that these minor events would trigger substantial capital outflows from emerging countries, producing large moves in exchange rates and bond yields. From the perspective of the markets, this was a big deal: a lot of bonuses depended on getting this right. But the impact of this short-term volatility on trend growth is often quite modest.

For example, foreigners hold one-third of Indonesian government debt, making both bond yields and the exchange rate sensitive to the ebbs and flows of foreign money. But the impact of these hot-money flows on the real economy is largely psychological and often ephemeral. The flows which have potential to affect the real economy – foreign direct investment – are not only more stable but small; FDI is just 2% of GDP.

A cautious forecaster mentions risks. But it's unhelpful just to fret about the downside. If things are more likely to turn out worse, why isn't the forecast adjusted downwards? A probability distribution around the central forecast provides discipline.

Economic reporting suffers from the same 24-hour news cycle that trivialises political analysis. Financial markets get over-excited about the latest data release, often a minuscule increment to the overall picture. Underlying growth is a boring story which doesn't change much from day to day, and commentators need a new story every day, even when there is no news.

The world would work better if the 24/7 cycle in both politics and economics were to fade away. But only a cock-eyed optimist would make a forecast like that.

Photo by Flickr user .scribe.



You may also be interested in