One of the most successful investment strategies since the turn of the century has been the risk-parity “long-long” of combined equity, credit and duration derivatives. In a simple form this trade takes continuous joint equal mark-to-market exposure in equity or credit and duration risk. A simple passive portfolio in the G3 would have outmatched most macro hedge funds since 2000, with a Sharpe ratio well above one and not a single annual drawdown. There have been three apparent contributors to this success: undiversifiable risk premia, implicit subsidies paid by central banks, and great diversification benefits from negative return correlations. These forces remain largely in place, but setback risks bear careful watching: excessive leverage in duration exposure, exhaustion of downside scope for yields, attempts of monetary policy normalization, and the possibility of a fundamental shift in macroeconomic policy regimes.
The below is based on research of SRSV Ltd. and Macrosynergy Partners Ltd. Only the below more general results are for the public domain and should be viewed as basis for expert discussion, not as financial or trading advice.
The post ties in with the SRSV summary on implicit market subsidies.
What is a long-long trade?
The long-long trade is a position that takes simultaneous exposure to equity (or credit) and duration risk, whereby the weight of the two legs of the trade is equal in terms of expected volatility. The latter is often called risk parity. For example, a risk parity equity-duration long-long trade takes long exposure in a local equity index future and a local-currency government bond future in such proportion that the mark-to-market fluctuations of both positions are expected to be equal.
Why and how to set up a long-long trade
The rationale for taking long-long exposure is simple: over the long run both equity and bond markets pay risk and other premia from issuers to investors. Moreover, in a low inflation environment, the dominant return drivers are real economic and financial system shocks, which affect returns on equity and duration exposure in the opposite direction. A negative economic shock drives equity prices lower and (high-grade) bond prices higher because the latter stand to benefit from a shift in expected monetary policy from tightening to easing. As a result, returns on equity and duration exposure have mostly negatively correlated in low inflation regimes, producing strong diversification benefits and increasing the ratio of implicit long-long premium to price volatility.
Liquid long-long risk parity positions typically focus on large economies, such as the U.S., the euro area and Japan (G3 henceforth). Long-long positions in smaller economies and emerging markets would have different nature, due to the strong influence of the exchange rate and credit spreads.
We select a representative set of long-long trades in the G3 with the following positions. The tickers in brackets are used for subsequent charts.
- The U.S. credit-duration long-long (USC) takes risk parity positions in (inversely spread-weighted) CDX investment grade and high yield indices and the 10-year USD IRS receiver.
- The U.S. equity-duration long-long (USE) takes risk parity positions in the S&P500 future and the 10-year USD IRS receiver.
- The euro area credit-duration long-long (EUC) takes risk parity positions in (inversely spread-weighted) iTraxx main and crossover indices and the 10-year EUR IRS receiver.
- The euro area equity-duration long-long (EUE) takes risk parity positions in the Euro Stoxx 50 future and the 10-year EUR IRS receiver.
- The Japan equity-duration long-long (JPE) takes risk parity positions in the Nikkei 225 future and 10-year JPY IRS receiver.
The credit exposure is based on spread parity positions in the main investment-grade and high-yield indices. This means that the notional of the composite position is predominantly investment grade (85% on average), but the risk contribution of the two credit segments is equal.
The ratio of notional IRS positions to equity/credit positions is quite different across markets. In the credit markets, the ratio of IRS to CDS positions has been 0.4-0.5, reflecting the narrow spreads and shorter duration of the dominant investment-grade segment in the index. In the equity markets, the ratio of IRS to index futures positions has on average been 2.3 in the U.S. and 4.3% in the euro area. Both ratios have been stationary over the past 20 years. By contrast, the IRS-equity position ratio in Japan has been rising over time, due to long-term rates compression and subsequent yield curve control and has by now reached a range of 10-11.
The diversification benefit of long-longs has been significant and consistent. As the below chart panel illustrates, since 2000 the correlation between equity or credit return and IRS receiver returns has been negative between 84% (U.S. equity duration) and 94% (euro area credit duration) of the time, based on 36-day lookback windows.
The average 36-day lookback correlation coefficients have been -22% to –30% on a daily basis. There have been recurrent short episodes of positive correlation but they typically did not last more than 2 months. A notable exception was the 2013 “taper tantrum” and its aftermath. On that occasion, the Federal Reserve’s announcement of a gradual reduction in asset purchases rattled treasury, equity and credit markets alike. In particular, a joint hit on duration and credit exposure arose from the nature of reduced asset purchases, which had supported both treasury and mortgage-backed security (i.e. credit) markets.
The performance of a long-long risk parity portfolio
The performance of a simple stable and equally-weighted long-long position in the G3 since 2000 has been remarkably strong and consistent by conventional standards. The below graph shows the performance of the overall G3 portfolio for an annualized long-term volatility of one and without compounding. The long-term Sharpe ratio has been 1.4 and the long-term Sortino ratio close to 2. Correlation with the S&P500 has been 45%.
Returns have displayed fat tails relative to the normal distributions and modest positive skewness. The highest positive monthly return was 1.1 annual standard deviations. The largest monthly drawdown was 0.6 annualized standard deviations. The deepest peak-to-trough drawdown would have been 1.4 annualized standard deviations, which is not too bad given that the trade neither applied stop losses nor used volatility control. Most remarkably, this naïve strategy would not have produced a single negative annual performance since 2000.
Across all markets, the long-long position would have posted a strong performance. The euro area credit-duration risk parity would have produced the highest absolute and risk-adjusted return. Also, in each market, the risk-parity long-long would have outperformed equivalent positions in credit, equity or IRS receivers alone. It also appears that the more credit and equity markets underperform, the more IRS receiver positions outperform.
Explaining the performance of the long-long -trade
The apparent main forces behind the positive long-term performance of risk parity long-longs are risk premia and implicit subsidies: Risk premia arise from the undiversifiable portion of the market risk of a security price. Since equity, credit, and government bonds are issued to the market by institutions that (partly) have objectives other than mark-to-market risk-return optimization, the financial investor community as a whole must be long these risks and unable to diversify them away.
Implicit subsidies are paid over and above conventional risk premia through transactions that have motives other than conventional risk-return optimization (view summary post). Implicit subsidies may be paid by a range of market participants. In particular, over the past 20 years, G3 central banks have supported the market by providing highly accommodative refinancing conditions and intervening in securities markets verbally and directly in the form of asset purchases. Since central banks have become more responsive to market distress, their policies have been interpreted as a ‘central bank put’, i.e. some measure of free insurance for investments that are exposed to directional and systemic market risk.
A crude proxy of premia and subsidies is the risk parity carry. The general cross-asset definition of carry is return for unchanged prices (view post here). Here we set credit carry equal to CDS index spread, set equity index futures carry equal to the average of index dividend and earnings yield minus the real short-term interest rate (view post here), and set IRS carry equal to the sum of yields spread and rolldown return (view post here). The risk parity carry is the inverse volatility-weighted sum of the carry on the credit/equity leg and the IRS leg.
Since 2000 the risk parity carry has been positive across all five markets almost all the time, except for a few weeks in the U.S. equity/IRS markets in the early 2000s. Moreover, the link between premia and risk parity returns is supported by a highly significant positive correlation between the size of the risk parity carry and subsequent long-long risk parity returns, across all markets.
The importance of implicit policy subsidies is supported by the correlation between economic slack and subsequent long-long returns. Economic slack in a broad sense should give rise to accommodative monetary policy. The below proxy for economic slack is a composite z-score based on five constituents,  the shortfall of estimated current GDP growth relative to potential growth,  the latest seasonally-adjusted unemployment rate relative to a 5-year moving average,  the shortfall of private credit growth relative to expected long-term nominal GDP growth,  the shortfall of wage growth relative to expected long-term GDP growth, and  the shortfall of annual core inflation relative to the inflation target.
There is strong evidence for a positive correlation between economic slack and long-long returns for the U.S. and euro area markets. There has been no correlation in the Japanese markets, probably because the Japanese cycle was out-of-sync with the other large countries in the early 2000s, while its markets were strongly affected by U.S. and euro area developments.
What can go wrong?
In principle, the basis for ongoing premia and subsidies on long-long risk parity positions has remained in place:
- Subdued inflation and high financial leverage have reinforced central banks’ bias towards accommodative monetary conditions and interventions in times of market distress.
- Low inflation levels and variance mean that real economic and financial system shocks continue to dominate the monetary policy outlook. This predominance of real economic shocks, as opposed to inflation shocks, is the main factor behind negative equity bond correlation, which in turn secures the diversification benefit of long-long risk parity positions.
- Positive long-long risk carry (as a proxy for premia and subsidy) has prevailed to this day, even if it is on the low side by historical standards. On average, the risk parity carry for the five markets considered stood at 9% on July 9, 2019, compared with a 13.4% average since 2000.
However, there are also severe setback risks for long-long risk parity performance. Put positively, the long-term performance of risk parity portfolios can very likely be improved upon and protect against devastating losses by monitoring the below key risk factors:
- Leverage risk: As bond and swap yields are being compressed and central banks are tightening their grip on rates markets price volatility of duration positions can collapse relative to the price volatility of credit and equity positions. As a result, IRS leverage in the long-long position may surge. This has already happened in Japan, where the central bank has introduced full yield curve control in 2016 (view post here). In Japan, equivalent notional for a USD10 million equity position in a long-long is now about USD110 million notional 10-year IRS exposure. This greatly increases position tail risk. The DV01 of a 10-year IRS position is roughly USD1,000, which means that a mere 50bps increase in 10-year swap yields would produce a mark-to-market loss of roughly USD5.5 million on the JPY 10-year IRS leg of the long-long position. It would require a 55% increase in the value of the equity position to offset this. The bottom line is that leverage should be a negative contributor to positioning signals.
- Boundary risk: While there is no clear zero bound for nominal yields, deeply negative swap yields are unlikely unless there is a significant change to monetary regimes (view post here). This means as long-term swap yields get close to zero or even go negative the upside on IRS receiver returns is increasingly constrained. As a result, receiver positions will no longer protect the long-long against large declines in equity and credit prices. The proximity of bond and swap yield levels to lower boundaries should be a negative contributor to positioning signals.
- Tantrum risk: Shifts in market expectations towards monetary policy tightening, over and above those that arise from strong markets themselves, hit both the equity/credit and duration, producing drawdowns and increased volatility in risk parity positions. This is why any policy-induced tightening of financial conditions is dangerous for the long-long. This danger was prominently illustrated by the 2013 ‘taper tantrum’. Hence, indications of persistent economic strength and rising inflation should be negative contributors to positioning signals.
- Fiscal risk: It is possible that at some stage G3 policy regimes will change so as to allow some form of helicopter money, i.e. monetization of fiscal spending (view post here). This may be a fairly effective way to spur reflation. However, in this case, the predominant response to negative economic shocks would be higher inflation expectations. Even before inflation itself increases, the correlation between equity and swap returns could flip from negative to positive, destroying the diversification benefit of the long-long. Higher inflation expectations and actual inflation levels would reinforce this positive correlation, as was the case in the 1970-90s (view post here). Hence, a sustained shift to central bank-funded fiscal expansion would invalidate a key part of the rationale for long-long risk parity positions.