As a profession, economists are absolutely hopeless at forecasting recessions. That is true not only in the years before a severe downturn. It happens when the storm is about to break. Back in 2008, the Bank of England failed to predict the biggest postwar slump in the UK’s history even after it had started.
This less than impressive record should act as a cautionary note in the current circumstances when the 10th anniversary of the collapse of Lehman Brothers has generated a thriving cottage industry devoted to predicting when the next crisis will occur. The honest answer is that nobody really knows. Meteorology has improved in the past 40 years: economic forecasting has not. When a weather forecaster says a hurricane is imminent, the public does well to take notice. When an economic forecaster gives a similar warning, the chances are that it is already too late.
This might be about to change. Just as satellite technology has made weather forecasting far more accurate, so machine-learning algorithms could bring economic forecasting into the 21st Century.
Rickard Nyman and Paul Ormerod have compared economic forecasting by humans and machines in both the US and the UK, and come up with some stark conclusions. At the start of 2008 the survey of professional forecasters in the US failed to predict that within a year their country would be in a deep recession. Had US policy makers relied on machine-learning algorithms they would have been much better prepared for the trouble ahead. Even more impressive results using machine learning were obtained for the UK.
There’s more, however. Nyman and Ormerod sift through all the economic and financial variables that might have been responsible for causing the downturn and come up with a conclusion that explodes the myth that overspending governments were to blame.
In truth, the idea that state profligacy caused the Great Recession has never been credible. What really happened was that the expansion of the global marketplace led to cheap goods flooding the west. Inflationary pressure abated and that persuaded central banks to cut interest rates. Financial deregulation meant the only remaining constraint on excessive borrowing – high interest rates – was removed – and so credit was cheap and readily available. The private sector loaded up on debt, which was fine so long as the assets on the other side of the balance sheet were going up in value. When the markets turned, things went pear-shaped very quickly.
Only at that point, did public sector debt become a problem because governments sought to ameliorate the impact of the recession by cutting taxes and increasing spending. The debts incurred by the private sector were, to an extent, nationalised.
Bill White, then chief economist at the Bank for International Settlements was one of the very few who sought to prick the bubble of complacency in the boom years, making the point that periods of low inflation could still end in crisis. His warnings were not heeded.
The BIS remains concerned about debt levels, which is where the work of Nyman and Ormerod is relevant to the state of the world 10 years after Lehman.
Put simply, the cure for the Great Recession was for central banks to slash interest rates and to increase the supply of money by buying bonds from the private sector in the process known as quantitative easing. Debt levels in the private sector fell for a while as households and companies retrenched but have subsequently started rising again. Low interest rates were designed to provide incentives for investors to seek out riskier assets, which is what they have done. Money has flooded into emerging markets, where yields are juicier because the risks are higher. Turkey, where the central bank raised interest rates to 24% last week, is one example of what can happen in a world of footloose capital. Speculative money comes in from abroad. It finances a construction boom and drives up the exchange rate. Eventually, the trade deficit starts to balloon and inflation starts to rise. At that point, the speculators take fright and the exodus of capital triggers a fall in the exchange rate. At that point, the central bank has to raise interest rates to punitive levels to defend the currency and recession becomes inevitable.
The BIS says in its latest annual report that there are already material risks to financial stability. “In some respects, the risks mirror the unbalanced post-crisis recovery and its excessive reliance on monetary policy. Where financial vulnerabilities exist, they have been building up, in their usual gradual and persistent way. More generally, financial markets are overstretched … and we have seen a continuous rise in the global stock of debt, private plus public, in relation to GDP. This has extended a trend that goes back to well before the crisis and that has coincided with a long-term decline in interest rates.
Behind the dry official language, the message is clear. A recovery that is based around high and rising levels of debt is really no recovery at all. The world economy is, in all material respects, the same as it was in the run-up to the 2008 crisis. The necessary reforms to a flawed model have not taken place, which is why the BIS warning should not be ignored.
While it is not possible to say when the next recession will arrive, or where it will start, history is repeating itself. And you don’t need a machine learning-algorithm to know there are dangers involved in that.