V-shape equity recovery: a killer for momentum crossover strategies

Even though the financial markets recently experienced a variety of interesting events, the most surprising one was the fast recovery in equities since they hit their low on March 23rd. Figure 1 shows that the SP500 has pulled back to its 50% Fibo retracement of its yearly high-low range, experiencing one of its fastest rally after plunging by 35%. In addition, we saw that one popular technical indicator, the 50D-200D simple moving average crossover, has been forming a ‘death cross’ in the past, which traditionally indicates a bearish signal to market.

Figure 1

SP500

Source: Eikon Reuters

There are many popular momentum strategies based on MA crossovers (both simple and exponential), but the death cross vs. golden cross is a very known one for all asset classes and is closely followed by many market participants. It is a simple systematic strategy, that sends a positive signal when the short-term moving average (50D) trades above the LT MA (200D), which is also called a ‘Golden Cross Formation’, and sends a bearish when the ST MA trades below the LT MA (‘Death Cross Formation’).

Looking at 25 years of daily data, we compute the performance of the Long/Short strategy on the SP500 and compare it to the performance of the SP500 index. Figure 2 shows that the 50/200 crossover strategy generally performs well in periods of slowly trending market (either bull or bear markets), but experiences severe drawdowns in periods of choppy markets. Figure 2 also compares the performance of the L/S strategy with the buy-and-hold one. For a similar volatility, the L/S strategy enhances our annual return by 1.4% to 8.5% for a Shape ratio of 0.44 (vs. 0.37). In addition, drawdowns are significantly reduced as investors are shorting the SP500 in periods of bear equity markets.

Figure 2

Crossover

Source: Eikon Reuters, RR calculations

Based on these results, it seems very tempting for buy-and-hold investors to change their strategy to the systematic one and therefore avoid severe losses in periods of selloff. However, we think that equities will experience frequent V-shape forms with central banks trying to prevent the stock market from falling by 50%+ in the future, which will be devastating for momentum strategies such as MA crossovers. In other words, it will be the death of the ‘Death Cross’ crossover strategy.

 

Jobless claims and non-farm payrolls

Two popular indicators that investors frequently watch to measure the temperature of the labor market in the US are the jobless claims and the non-farm payrolls. Historically, the market has been focusing more on the monthly NFP prints, but the recent waves of selloffs due to the Covid-19 crisis has raised investors’ interest on the weekly jobless claims figures as the number of Americans filing for unemployment benefits has surged to over 6 million in the past two weeks.

In this post, we try to answer the following two questions that have been making the headlines in recent weeks:

  • Is there a link between jobless claims and non-farm payrolls?
  • As jobless claims are updated every week, does it lead NFP?

 

Jobless claims are a statistic that is reported every week by the US Department of Labor and counts the number of people filing to receive unemployment insurance benefits. There are two categories: initial – with people filing for the first time – and continued, which comprises of unemployed people who have been receiving unemployment benefits for a while. What is surprising this time is the recent rate of change of the initial jobless claims, which surged from multi decades lows of 200+ thousands to over 3.3mil, 6.8mil and then 6.6mil on April 4th. Prior Covid-19, initial jobless claims have generally averaged 360K since the 1960s, from lows of 200K to highs of 700K (figure 1, right frame). The number of continued jobless claims is now greater than the one experienced during the Great Financial Crisis, which reached a peak of 6.6 million in June 2009.

Figure 1

FI1

Source: Eikon Reuters

Non-farm payrolls are also released by the Department of Labor on a monthly basis as part of a comprehensive report on the state of the labor market, but do not include farm workers, private household employees or non-profit organization. It was reported that the US lost 701K jobs in March, which brought the unemployment rate to 4.4%. Figure 2 (left frame) shows a scatter plot of the continued jobless claims with the NFP monthly; we are currently sitting at uncharted territories, and we would expect the next prints of NFP to collapse to much lower levels in the coming months, which would raise the unemployment rate to more than 20 percent. According to the fitted line, a 701K drop in NFP would coincide with continued jobless claims of 4.5 to 5 million.

In figure 2 (right frame), we look at the yearly gains (losses) in NFP overlaid with the continued jobless claims times series. We can notice that to the exception of the Great Financial Crisis, the amount of people filing for unemployment benefits has always been greater than the number of jobs lost in the US economy. This is what we expected as the NFP do not include a little portion of US employees. If you add the number of jobs lost in the agricultural industry in addition to the local government, private household and non-profit employees, you will certainly reconcile the two figures (i.e. number of total jobs lost = continued jobless claims).

Figure 2

FI2

Source: Eikon Reuters

It is difficult to infer a level of unemployment rate from the jobless claims data, but we just know that the level will be elevated in the coming months. Some economists have forecasted a 10% unemployment this summer, but we think it could actually reach 20 percent as the uncertainty around employees’ status will surge to historical highs. US households will start to save more as most of the companies are now very vulnerable to the demand shock, which would in theory be deflationary (at first).

Even though the chart is far from being perfect (figure 3), we like to look at the 3-year returns in stocks (in order to smoothen the volatility in equities) as a leading 1-year leading indicator of unemployment. Sharp and sustain selloffs in equities are usually associated with rising unemployment as equities’ valuations directly reflect the level of consumer sentiment in the market.

Figure 3

FI3

Source: Eikon Reuters

Mind The Rise In The Term Premium!

Introducing the Term Premium

Through the use of economic models, academic research has decomposed the observable long-term bond yields (i.e. US 10Y Treasury bond yield) into the expected path of the real interest rate (r*) and the additional term premium, which is thought as the extra return that investors demand to compensate them for the risk associated with a long-term bond. Using the dominant measure developed by the NY Fed (Adrian et al., 2013), we overlay it with a set of macro and financial variables and look at the pros and cons of a rise in the term premium in the coming months.

Figure 1 shows that the evolution of the US 10Y yield along with the expected r* and the term premium. While we can notice that part of the fall on the 10Y was driven by the decrease in the expected r* from 3.15% to 1.80%, the elevated volatility in the short run was mainly coming from the moves on the Term Premium (TP). The TP hit a historical low of -1.47% on March 9th and is still standing at extreme low level of -1.1%. Many investors had expected the term premium to start rising the US in 2018 and in the first half of 2019, but it surprised most of them by constantly reaching new lows.

Figure 1

fig1Source: NY Fed

Term Premium: a counter-cyclical variable

One important characteristic of the term premium is that it is a counter cyclical variable that tends to rise when the uncertainty around unemployment (or the business cycle) and inflation expectations starts to increase.

Figure 2 (left frame) shows the striking relationship between the unemployment rate and the term premium since 1961. Periods of rising unemployment have been generally associated with a sharp increase in the term premium. Now that we expect the jobless rate to skyrocket following the dismal prints of NFPs and initial claims in recent weeks, could we see a response in the term premium as well?

Figure 2 (right frame) shows another interesting relationship between the US 2Y10Y yield curve and the term premium. We know that the inversion of the yield curve is usually marked by a sharp steepening effect within the next 12 to 24 months as the economy enters a recession. This is referred as a ‘bear steepener’ as the long end of the curve starts rising due to a surge in the term premium. Will the Fed’s emergency measures and QE purchases be enough to deprive the yield curve and term premium from rising significantly in the coming months?

Figure 2

fig2Source: Eikon Reuters

Term Premium vs. inflation expectations

The 2-trillion USD increase in the Fed’s balance sheet as a response of Covid-19 has brought its holding of securities to a new all-time high of $5.85tr. The Fed has recently been buying $625bn of securities each week, which corresponds to an annual pace of $32.5tr and is $25bn more than the entire QE2 run between November 2010 and June 2011. The balance sheet of the Fed is now expected to hit 8 to 9 trillion USD by the end of the year in hopes that it will bring back confidence in the market. Hence, it is fair to raise the following question: will we experience rising inflation in the medium term?

As the term premium is very sensitive to the uncertainty around inflation expectations, it shows an interesting co-movement with the 12-month volatility of the Fed’s balance sheet assets. When interest rates reach the zero bound, central banks run aggressive asset-purchase programs in order to decrease the shadow rate below the neutral rate of interest rate (r*) and stimulate demand and inflation. Figure 3 (left frame) shows that previous periods of rising 12M vol in Fed assets were associated with a short-term increase in the term premium.

Investors could argue that inflation expectations have been falling if we look at the market-based measures – the USD 5Y5Y inflation swap. Figure 3 (right frame) shows that the 5Y5Y inflation swap is currently trading at a historical low of 1.75%, down from nearly 3% in January 2014. However, we previously saw that inflation swaps have been very sensitive to equity and oil prices in the past cycle; in theory, an oil shock should not impact inflation expectations as better monetary policy readjustments from central banks will offset that shock. Hence, these products represent more the demand for inflation hedges (which decreases when energy prices fall), but do not tell us anything about long-term inflation expectations.

Figure 3

fig3Source: Eikon Reuters

Term premium and free-floating bonds

Certainly, moves on the term premium also depends strongly on the amount of free-floating securities in the market. As central banks keep increasing their balance sheet through the purchase of securities such as government bonds, the amount of free-floating bonds have dramatically been reduced in the past cycle. For instance, it was estimated that large asset-purchase programs in the Euro area have decreased the free float of German government bonds from approximately 40% in 2015 to 3% in early 2020 (figure 4, left frame). Figure 4 (right frame) shows how the relationship between the 10Y Bund term premium and the free float flattened in the past few years; low free float is associated with a flat term premium.

In the US, the amount of free float is much higher due to the large quantity of marketable debt securities (USD 16tr) held by the non residents; non-resident holders (NHR) hold nearly 40% of the US debt. Hence, even though the Fed’s aggressive purchases will reduce the free float in the medium term, there is still the risk of a sudden rise in the term premium in the short run as the economy enters a recession.

Figure 4

fig4

Source: Danske, ECB

To conclude, the risk of higher long-term interest rates in the US is still there in the coming months; even though we do expect long-term rates to eventually go to zero, there is still a high probability to see a little short-term surge in the 10Y -plunge in US Treasury prices (TLT), which would significantly steepen the yield curve coinciding with the drastic rise in unemployment.