Economic Research THE VIEW 15 May 2019 Photo by Pixabay Photo Jonah from Pettrich THE SHAPE OF THE U.S. YIELD CURVE: A CAUTIONARY TALE 03 Beware of the yield curve 07 Is this signal going to persist? The View by Economic Research In March 2019, the spread between the long and short EXECUTIVE end of the U.S. yield curve entered into negative terri- tory, sparking concerns about an economic downturn ahead. For the past 60 years, each U.S. economic SUMMARY downturn has been preceded by a yield curve inver- sion. While some may argue that this time it is differ- ent, our calculations show that implied recession prob- abilities range between 43-60%, hinting towards a very likely economic correction in the near future. The tim- ing is, however, unknown. Our long-term yields proprietary model shows that the current flat shape of the yield curve is likely to stay. Additionally, the current anchoring of the short end of the curve by the Federal Reserve at 2.5%, plus the im- plicit stickiness of the long end of the curve, fairly val- Jordi Basco Carrera ued at 2.5%, hints at the possibility of further yield Capital Markets Research at Allianz SE curve inversion. The warning signs derived from the +49 (0) 89380017896 current and future shape of the yield curve suggest [email protected] that a U.S. economic downturn is probable. Investors need to look towards a downside risk-managing in- vestment strategy. 43-60% ESTIMATED RECESSION PROBABILITY IMPLIED BY THE U.S. YIELD CURVE 2 15 May 2019 BEWARE OF THE YIELD CURVE The U.S. economy has just matched the look. For the past 60 years (earliest long-term and short-term U.S. Treasury longest cycle ever recorded in modern data point), an inversion of the yield yields (term spread) stood at only times (Figure 1). In this context, market curve heralded each economic down- 13bps and, most importantly, the yield participants are increasingly watching turn in the U.S. (Figure 2). curve inverted for six days between for a turning point, focusing on the in- March 22 and March 28 2019 (Figure Through most of the current period of version of the U.S. yield curve (i.e. when 3). expansion, the yield curve has been on long-term treasury yields are lower a flattening spree. As of the end of than short-term treasury yields) as an April 2019, the difference between indicator of the future economic out- Figure 1: United States economic cycles Figure 2: U.S. 10Y-1Y term spread Figure 3: U.S. yield curve % 400 3,5 Recession Great Depression (1929-1933) 50 10Y - 1Y Recession of 1937-1938 80 300 3 Recession of 1945 34 Recession of 1949 45 200 2,5 Recession of 1953 39 100 Recession of 1958 24 2 Recession of 1960-1961 106 0 bps Beginning of Expansion of Beginning Recession of 1969-1970 36 1,5 Recession of 1973-1975 58 -100 Recession of 1980 12 1 Recession of 1981-1982 92 -200 31.12.2016 31.12.2017 Early 1990s Recession (1990-1991) 120 0,5 31.12.2018 25.03.2019 Early 2000s Recession (2001) 73 -300 30.04.2019 Great Recession (2007-2009) 120 -400 0 0 20 40 60 80 100 120 140 1960 1970 1980 1990 2000 2010 2020 1M 3M 6M 1Y 2Y 3Y 5Y 7Y 10Y 30Y Months Sources: National Bureau of Economic Research Sources: Bloomberg, Allianz Research Source: Bloomberg 3 The View by Economic Research Historically, pundits have always found To assess the reliability of the signal, Note that we could have used the 10- reasons to downplay the significance we computed four logistic regressions year treasury yield minus 1-year treas- of yield curve inversions (Table 1). This based on four yield curve metrics as ury yield as shown in Figure 2 as most time is no different: An argument per- proxy variables for the term structure metrics are yielding similar results. Also sists, suggesting that the relationship of the U.S. yield curve. We found that: note that recession probability sig- between the slope of the yield curve 1. When using the 10-year treasury naled by the three term spreads is con- and the business cycle may have now yield minus 2-year treasury yield, sistent with former turning point levels4. changed due to unique current circum- the probability of a recession at This becomes more visible when one stances, including the unusually low current term spread is 43%; uses the Near Term Forward Spread risk premiums holding down interest 2. When using the 10-year treasury metric alone (Figure 5). Last, though all rates. However, this paper empirically yield minus 3-month treasury yield, indicators confirm the likelihood of a shows that one should take the short- the probability of a recession at recession, timing it with precision lived yield curve inversion as a warning current term spread is 50%; (lapse between inversion and contrac- signal. An inverted yield curve can re- 3. When using the Near Term For- tion for e.g.) is not statistically signifi- flect lower future economic expecta- ward Spread2 (18-month forward cant. tions and hint at a palpable conjunc- spot yield minus 3-month Treasury tion of both reduced economic health yield)3, the probability of a reces- and future growth. Secondly, it can sion is 58%; and also be a symptom of the Fed driving 4. When using the Fed Funds rate up shorter-term rates too high, which, deviation from equilibrium, the in turn, could lead to a contraction in probability of a recession is 60%. both lending and the economy. Table 1: Yield curve inversions: a summary Period Argument Result The U.S. economy suffered a brief recession from July 1990 to The yield curve is artificially inverted because com- March 1991 because of rising interest rates and an oil price 1989 mercial banks are buying longer-term Treasuries shock caused by Iraq’s Invasion of Kuwait in 1990 and the subse- instead of lending to real estate. quent first Gulf war. The yield curve is artificially inverted because the The collapse of the dotcom bubble and the September 11 terror- 2000 Clinton administration is running a budget surplus ist attacks caused a mild recession from March 2001 to Novem- and no longer issuing long-maturity Treasuries. ber 2001. The yield curve is artificially inverted because a The deflation housing bubble and subprime mortgage meltdown 2006 global savings glut is keeping longer-term bond caused a global financial crisis and the worst recession since the yields pinned down. Great Depression, lasting from December 2007 to June 2009. The yield curve is artificially inverted because of the Fed’s market-distorting quantitative easing pro- 2019 Unknown. gram, a jump in Treasury bill issuance and low glob- al interest rates. Source: Financial Times 1 1 Financial Times: Has the yield curve predicted the next U.S. downturn? Robin Wigglesworth and Joe Rennison, Apr. 3rd, 2019 2 E. Engstrom and S. Sharpe. "(Don't Fear) The Yield Curve," FEDS Notes. Washington: Board of Governors of the Federal Reserve System, June 28, 2018, https://doi.org/10.17016/2380-7172.2212. 3 The near-term forward spread does appear to be a pretty good gauge of market expectations regarding monetary policy. Consequently, assuming market participants have some foresight, it is not all that surprising that negative readings for the near- term spread tend to precede (and thus can be used statistically to forecast) recessions 4 In contrast to the NY Fed indicator, the computed indicators show a much higher recession probability. Apart from minor methodological differences it is worth mentioning that the NY Fed indicator tends to peak at 35-45% levels hinting that a 40% recession probability for the Fed indicator may prove enough to predict future recessions. https://www.newyorkfed.org/research/capital_markets/ycfaq.html 4 15 May 2019 Figure 4: U.S. term spreads Figure 5: Recession probabilities based on Term Spreads 400 100% Recession Recession 10Y - 3M 90% 10Y - 3M 10Y - 2Y 10Y - 2Y 300 80% 70% 200 60% 50% bps 100 40% 30% 0 20% 10% -100 0% 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 400 100% Recession Recession Near Term FWD Spread 90% Near Term FWD Spread 300 80% 70% 200 60% 50% bps 100 40% 30% 0 20% 10% -100 0% 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 Sources: Bloomberg, Allianz Research Sources: Bloomberg, Allianz Research Figure 5 shows that the Near-Term For- ican peso crisis. Moreover, at that point, economic conditions (late-late cycle ward Spread generated two false sig- the length of the U.S. expansion was stage and most economic indicators nals (first in 1995, then in 1998). Look- only four years long. In the case of the point towards slower U.S. activity) are ing at those two short-lived inversion 1998 inversion, a similar story holds. different from those prevailing at the periods, one can argue that the nature The (near) inversion was due to the ex- time of these two erroneous signals. of those misleading signals was not ogenous shock provoked by the Rus- only different, but completely unrelated sian financial crisis. While Brexit and to U.S. domestic matters. In 1995, the the U.S.-China trade tensions might be U.S. economy was confronted with the compared to the 1995 or 1998 exoge- exogenous shock provoked by the Mex- nous shocks, we consider that current 5 The View by Economic Research Figure 6: Fed Funds deviation from equilibrium 10% Federal Funds rate deviation from equilibrium Recession 8% UST 1-10 year spread 6% 4% 2% 0% -2% -4% -6% 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 Sources: Bloomberg, Allianz Research Figure 7: Recession probabilities based on Fed Funds “fair value” 100% Recession Fed Funds deviation from fair value 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 Sources: Bloomberg, Allianz Research As for Fed Funds deviation (Figures 6 & rates too high (i.e.