Today's furious stock market short squeeze notwithstanding, the US 10Y Treasury and the yield curve has barely budged, and the result is a yield curve that remains deeply inverted.
This, of course, is a problem because as we have noted on various occasions in the past, a yield curve inversion has always preceded a downturn and/or a market crash. The one variable is the lead time between inversion and the actual recession, which has varied significantly, making it an unreliable indicator in timing recessions. Additionally, a look back at the historical record reveals that there are two usually is a sustained inversion and multiple cases spells of inversion prior to the past 5 recessions, as the following BofA table shows:
Here is what the pre-recessionary empirical data reveal: the longest streak lasted 120 days for the 3mo/10yr curve and 208 days for the 2yr/10yr curve and there have usually been 4 inversions for each curve measure prior to a recession, on average, according to a recent Bank of America analysis. Also the total number of days the curve is inverted is quite high: on average, the 3mo/10yr curve inverted 181 days and the 2yr/10yr curve inverted 247 days prior to a recession.
So where are we now?
According to Bank of America, when compared to history, the current inversion episode is still far from getting a strong signal from the bond market that a downturn is around the corner: So far, the 3mo/10yr curve has been inverted for a total of 13days (through June 4th) while the 2yr/10yr curve has yet to invert through the current cycle. The latest flattening of the yield curve is likely a reflection of early market concerns around a slowing economy and the latest round of US-China trade tensions.
But a bigger question that needs to be answered between these surprisingly close correlations, is in what direction does the arrow of causality point: i.e., whether curve inversion is the cause of the weakening data, and subsequent recession, or merely a symptom.
Or, as BofA explains, can the yield curve help predict the evolution of economic data? That is, can the inversion of the curve help forecast the deterioration of economic conditions? If so, every forecasting model should include the yield curve.
To answer this question, BofA runs a Granger causality test. What it finds is that the yield curve generally does not have information that would be beneficial in forecasting these economic indicators. Granger causality tests based on the 3mo/10yr curve show that most of the causality runs the other way. That is, the economic data tend to explain the oves in the slope of the yield curve.
For the 3mo/10yr yield curve, with the exception of industrial production, initial claims and existing home sales, economic data explains the moves in the curve:
Changes in industrial production and initial claims show inconclusive results as the relationship between the data and yield curve shows mutual causation (i.e. causality runs both ways). Only existing home sales shows that the curve has some explanatory value in forecasting the data.
Similarly for the 2yr/10yr curve, the results show most of the Granger causality either runs from economic data to the curve or shows mutual causation. The 2yr/10yr yield curve only shows one-way Granger causality to auto sales.
In conclusion, BofA's results imply that the move in the yield curve is more likely to be a reflection of the moves in the economic data rather than the driver of it. This suggests that the inversion of the yield curve is likely to be a symptom rather than the cause of weakening data. Even so, and using historicl precedent as an example, unless economic data picks up notably in the next few months, and certain in less than 170 or so days, then whether it is cause or effect will be irrelevant, as the amount of economic imbalances that will have built up will be so great, that only a painful reset would be able to restart the global economy.