Timing the recession

Proponents of the recession argument point to the ongoing banking crisis, weakening manufacturing and housing sectors, as well as the inverted yield curve. However, detractors will cite the strong job market and resilient consumer spending. ​

With conflicting signals given by different indicators, it’s important to understand exactly which indicators  have the highest predictive power. Timing is also important: indicators like the job market may only start to fall after recessions have started. ​

Using statistical methods, it is possible to objectively assess the accuracy of common recession indicators and determine their timing.

Our investigations into this yielded the following results:​

The 10 year – 3 month Treasury spread has historically been the most accurate indicator for predicting a recession 12 months in advance. This makes sense, as a yield curve inversion has preceded every US recession since 1970. Currently, this spread stands at a level consistent with an 86% probability of a recession in 12 months, and a 50% probability of a recession in 6 months. ​

However, with the advent of quantitative easing and liability driven investment since 2008, it could be argued that the yield spread does not mean what it did before. Furthermore, 3-month yields have been pushed up recently by sentiment surrounding the debt ceiling standoff in Washington D.C.​

As a result, the outputs of recession probability models, which take yield spreads into account ,should be taken with a pinch of salt. Some models of this type, such as the one by the Conference Board, are currently quoting 12 ahead month probabilities of almost 100%.

As the time horizon is shortened, the predictive power of the yield spread is diminished. Short term we find that the number of building permits issued, the Leading Economic Index (which includes yield spread as a component), and bank tightening standards start to have a higher accuracy. These indicators give a probability of recession between 30% and 50%​.

It is notable that the indicators currently giving the highest probabilities generally fall within the monetary and financial sectors. This makes sense, as the most likely channel through which a recession could potentially manifest is significant contraction of credit, leading to a decline in business investment. In this scenario, employment and consumption, which have remained robust this year, may yet fall at a later point in time. ​

Tao Yu, Quantitative Analyst