Recent market volatility and weaker data from the manufacturing sector raises the question: Is the U.S. economy heading into a recession? Based on our recession model, the chance of a recession is highly unlikely over the next six months.
Predicting the Probability of a Recession: A Probit Model
One useful method to predict the chances of a recession over the near term is to build a Probit model. A Probit model, in the present case, estimates (using a handful of predictors) the probability of a recession for a certain period in the future. Our Probit model predicts the probability of a U.S. recession during the next six months. The model utilizes the LEI, S&P 500 index and Chicago-PMI employment index among others as predictors. Our model has served us well, as it started predicting (in real-time) a significantly higher probability of recession in 2007 (58 percent probability in Q3 2007). In addition, we never joined the “double-dip” camp back in 2010-2012, largely because our Probit model never indicated a recession during that time period was likely. Using the most recent data (through December 2015), our model suggests a low chance of a U.S. recession during the next six months (1.17 percent probability, top graph).
The LEI, one of the key predictors of the model, fell 0.2 percent in December and one would expect a higher probability given the drop in the LEI. The reason behind the low probability is that the average growth rate during Q4 was 0.27 percent. Furthermore, the LEI was positive for October and November (0.5 percent for each month). Typically, if the LEI has a negative growth rate for several consecutive months (negative quarterly average) then the probability tends to move up. For instance, the Q3 average LEI was -0.05 percent and the model subsequently produced a 7.14 percent probability. In addition, other predictors of our model are also experiencing negative/weak growth rates as the S&P 500 index and employment component of Chicago-PMI are negative (middle graph). The LEI is also trending downward, however, it is still positive and consistent with a lower probability of recession. All three predictors were well below zero during the past two recessions and those time periods are consistent with very high recession probabilities (middle graph). Therefore, we are comfortable saying that a recession within the next six months is unlikely.
The Conference Board produces a coincident index (CI) for the U.S. economy and the index consists of four variables: nonfarm employment, industrial production, real personal income less transfer payments and real manufacturing and trade sales. The index, as its name suggests, is a useful measure of the current state of the U.S. economy. The average growth rate of the CI for Q4 is 0.13 percent, which suggests a weak but positive growth rate for the U.S. economy. We decompose the CI using the Hodrick- Prescott (H-P) filter. One major benefit of the H-P filter is that it characterizes a series’ movement relative to the series’ long-run trend. The bottom graph shows the average deviation of the four variables from their respective long-run trends. The average value is below zero for December 2015, which suggests a weaker growth rate compared to the long-run trend. Industrial production is dragging down the average index at present, as the rest of the three variables are either at or above the long-run trend growth.
In sum, our analysis suggests that the chance of a U.S. recession during the next six months is low.
Source: Federal Reserve Bank of Chicago, The Conference Board, Bloomberg LP and Wells Fargo Securities, LLC