Are recessions predictable?

By Bob Cunneen, Senior Economist at MLC

“Prediction is very difficult, especially about the future.” Niels Bohr, Danish Physicist, (1885 - 1962)

The ability of economists and investors to predict recessions seems as good as our skill in long term weather forecasting. The IMF suggests that “forecasters either do not have the information or the incentives to forecast recessions” * (1). Essentially, we appear incapable of exactly predicting when the recession’s storm clouds will appear and how intense the deluge of despair will be.

Recessions are typically characterised by weaker consumer spending, lower business investment, and higher unemployment. Recessions also usually feature very sharp falls in corporate profits, which can pull share prices down dramatically. So, whether we are employees, managers or investors, we need to be conscious that a recession is a serious threat to both our income and our wealth.

Recessions in Australia, Asia and Europe are defined as beginning when the level of Real Gross Domestic Product (Real GDP) declines for two consecutive quarters, six months. However, for the United States (US), a recession has a more elaborate definition. For the US, recessions are when economic activity declines in terms of a range of variables. These variables include “Real GDP” as well as “real income, employment, industrial production, and wholesale-retail sales”. A US recession also must “last more than a few months”. 

This US recession definition comes from the National Bureau of Economic Research (NBER).
Indeed, the NBER has its own “Business Cycle Dating Committee which sounds as though it runs Tinder. This “Dating Committee” of economists reviews various data and then declares when a recession has started and ended. However, this does not mean a timely announcement like today’s weather forecast. The “Dating Committee” typically makes their recession call with a significant lag after the recession has begun.

The “Dating Committee’s”key importance is in providing an objective timing measure. This timing also gives us a reference for assessing the predictive ability of any variables. Accordingly, our focus is on the US data as well as the NBER recession timing as this is the most comprehensiveinformation available.

The NBER assessment is that three US recessions have occurred in the past thirty years as seen in Table 1. The first recession from July 1990 to March 1991 can be attributed to the aftermath of the Saving & Loans financial crisis, tax increases as well as the first Gulf War.

The second recession from March 2001 to November 2001 comes after the “Dot.Com” technology wreck but was intensified by the September 2001 terrorist attacks on New York and Washington.

The last US recession from December 2007 to June 2009 follows the housing collapse and mortgage crisis. Importantly this last NBER “Dating Committee” pronouncement that a recession started in December 2007 was only made on December 1st, 2008. This is essentially a year after the event, so the NBER “Dating Committee” appears more like a cobwebbed collection of aged historians than a spritely group of sharp statisticians.

Are there any economic or financial variables that can successfully predict recessions? The Yield Curve is currently the prime suspect given a solid track record. The “Yield Curve” (chart’s red line) is the gap between long term bond yields and short-term interest rates. Currently the US yield curve is barely positive at only 0.15 % if one uses the gap between the US Government ten-year bond yield and the 3-month Treasury Bill. A negative yield curve is the time to start worrying. When bond yields move below shorter-term interest rates, this can signal that financial conditions are stressful and economic activity is set to weaken.

For the last three US recessions starting in 1990, 2001 and 2007, a negative yield curve appeared before US economic activity slumped (chart’s blue line shows the US Real GDP annual change, moves below 0% percent signify a recession). The time lead has varied between 9 months and 16 months for the last three US recessions. So the yield curve gives us an average warning sign 13 month ahead of the recession. This hardly constitutes a precise clock to set our recession timer to. Indeed, the potential causes and catalysts for the next US recession may still be in the making, leaving us guessing what type of actual storm cloud will hit the economy.

Reading the tea leaves

Leading indicators are economic variables whose movements precede the direction of the business cycle. These indicators can serve as the canaries in the coal mine to signal the potential for a looming downturn.

There are various measures that are considered leading indicators. Organisations such as the Conference Board and the OECD even produce leading indexes that combine various economic and financial indicators. Here we have chosen a select composite of US economic indicators such as business surveys, consumer confidence, housing starts, car sales and overtime hours. This composite index of leading indicators gives only a small-time window before a recession hits. The average lead is 4 months before the last three US recessions.

However, there was a critical failure in the last 2007-09 recession where the leading indicators were still positive for 2 months after the start of the recession. So even the leading indicators can prove to be misleading. Also, there was a false recession signal provided by the leading indicators in 1995 after the US central bank aggressively raised interest rates and Mexico had a severe debt crisis.

Another possible signal is credit conditions. Central banks frequently emphasise fundamentals such as credit growth, credit spreads and bank lending standard in trying to assess credit conditions. Tight credit conditions can cause a recession as businesses and consumer struggle to finance on reasonable terms with their lenders. Notably the Federal Reserve of Chicago has recently developed a measure of US credit conditions. This credit measure has lead US recession by between 3 months and 8 months. The chart below takes the average lead of 5 months and suggests that tight credit conditions have a significant explanatory power for US recessions.