Recession Indicator: NY Fed Model

By in
Recession Indicator: NY Fed Model

We are going to dedicate the rest of this week’s Data sections to US recession indicators. Many readers have pinged us on this topic in recent days, often mentioning how much the current economic/investment environment “feels” like prior tipping points. Worries about trade wars, European recession, negative interest rates, and a slower US labor market all feed into this narrative.

Today we want to focus on just one indicator: the New York Fed’s model of recession probabilities. Their graph of this data back to 1960 appears at the end of this section, but here’s how it works and what it says currently:

  • The NY Fed model has 2 inputs: 3-month Treasury bill yields (adjusted to a bond-equivalent basis) and 10-year Treasury bond yields. Higher yields on long-dated paper versus bills means recession odds are low, and the reverse increases the odds of future economic weakness.
  • The model uses these variables to calculate the probability of a US recession in 12 months’ time. In other words, the relative level of interest rates today has predictive power about the state of the US economy in July 2020.
  • One funny thing about this model: it hasn’t reached 50% odds since 1982, even though there have been 3 recessions since then (and a doozy in 2008, of course).
  • To correct for this oddity, most economists/market watchers call 30% the level to watch.
  • The current reading is 32.9%, up from 29.6% last month.

The most important question for our purposes is, of course, if breaking through 30% is a sell signal for US stocks. Here is the data for the last 3 times the Fed model went over 30%:

June 1990: Spot on. The Fed model’s probability breached 30% this month, at 33.2%, up from 27.0% in May 1990. The S&P peaked in July at 369 and closed the year 10.5% lower.

July 2001: Better late than never. Despite the bursting of the dot com bubble, the NY Fed model did not hit 30% odds of a recession until 16 months after the S&P peak of 1553 in March 2000.

If you had waited for that signal, you were already down 21%. On the bright side, if you did sell on the last day of July 2001 you missed a further 25% drawdown over the next 12 months.

August 2007: Close enough. Recession probabilities breached 30% this month at 34.5%, up from 29.5% in July 2007. The S&P 500 peaked in October, so if you sold in August you missed the last 6% of the rally. But, of course, you were on the sidelines as the Financial Crisis unfolded in 2008.

Takeaway: there is certainly enough here to cause concern. The last 3 times recession probabilities hit 30% you were better off lightening up on US stock exposure.

So, where could this indicator be wrong? We can think of 3 points right off the bat:

#1: It ignores oil prices. All three examples here coincide with a doubling of crude prices. The first two were related to geopolitics, the last from oil’s spike to +$120/barrel from financial speculation. As we have highlighted before, the US has not seen a recession since 1970 without oil prices doubling.

Conclusion: a doubling of oil prices in the current environment of high US production seems a remote possibility and at least puts a large asterisk next to the current 33% chance of a recession in 12 months.

#2: The 3-month yield is entirely in the Fed’s control. If they put any weight on this model (and they clearly should), then lower Fed Funds rates will quickly lower the probability of recession.

Conclusion: the FOMC needs to embark on a series of rate cuts that take 3-month yields to 1.80% from their current 2.25% payout. No surprise that Fed Funds Futures have sniffed out this reality: the modal estimate for the December 2019 contract is 1.75% – 2.00% (42% odds). Smack on what gets the recession odds comfortably below 30% as long as the Treasury market buys the story of an easier Fed.

#3: The 10-year note yield reflects fears of Eurozone recession, not a US slowdown. The real reason the Fed’s model sits at +30% recession odds is because the long end of the curve has come down by 80 basis points since January. The short end is only 20 bp lower. That 10-year Treasuries yield essentially 2.0% rather than 3.0% is largely due to the fact that German 10-years crossed the “zero rate Rubicon” in May and yield a close-to record low of negative 0.35% today.

Conclusion: Treasuries do not trade in a vacuum, and one must decide if a European recession (forecast by negative sovereign rates across the region) will spill over to the US. Also, one tangential thought: German Bunds do not show inversion (1-years negative 0.8%, 10-years -0.35%); does that mean the odds of recession there are extremely low? We doubt it…

Summing up: while we have focused on just one indicator here, this discussion writ large is exactly why we expect US equity market volatility to increase in Q3. The NY Fed’s recession odds have a good enough track record to make them actionable, but… The caveats we highlighted – oil prices, Fed policy, and global rates – are reasonable counterweights to the argument that a 2020 US recession is now a fait accompli.

Source:
NY Fed model description and FAQ: https://www.newyorkfed.org/research/capital_markets/ycfaq.html#Q11