The macro trade du jour from, say 2010 to 2016 was risk parity. Be long risk, short USD, and hedge with long duration treasuries. It was a product of the central bank regime of the time and it worked well.
The Trump administration changed things by making fiscal policy through taxes and spending the major driver of economic policy. Further, Trump appointed a Fed chairman with the express intention of “normalizing” monetary policy in effect making it less like 2009-2016 and more like year prior to the Great Recession. The effect is Trump is Making Cash Great Again, keeping equities great, and more inflation to come.
The Trump trade is an opening line with as many variations as the Sicilian defense, but it consists of three components:
Long US equities
Long USD
Short US duration
This is a shift from the popular risk parity trade of the last decade:
Long US equities
Short USD
Long US duration
Just as in opening lines in Chess change, so do fads, trends, trades in finance. Why trades lose their popularity is an interesting story, but there is a deeper value because these changes highlights what has changed in expectations about the world… and what could reverse them.
Seriously, if you can get 300 basis points in the money market with a pinch of credit risk, what is the point in 300 basis points for a 10-year treasury with all its term risk? Or if that is too hot, why not just stay with 3-month bills and lose a little in carry in exchange for far less risk?
But there is another issue at work: the negative long-run return correlations between bonds and equities broke down in 2018. Risk parity crucially depends on this correlation. When it breaks down, the treasury ballast used to smooth portfolio values given equity volatility stops working and you get big losses on all legs of the trade.
Let’s start by look at a closely related issue: a rising interest rate environment inducing a change in tolerance to term risk.
Term risk transfers volatility from debtors to creditors. Speaking with rigor, term premium is a measure of the uncertainty around the two variables of real GDP and inflation. The Fisher equation verifies this by defining nominal bond yield as real yield plus expected inflation. While this is valid over the long term, QE presented a serious distortion.
Practically speaking, term premium is the unknown around what we think we know. Cash flows far in the future are worth less today. Default losses become less important in present value terms the further out they occur. In effect, valuation becomes harder the further out we go in term.
There are five components to the term premium, and they are all in flux:
Fed asset purchases: expect the curve to steepen as the Fed shrinks its balance sheet simply because they can sell SOMA maturities as they see fit to keep the curve from inverting.
Central bank reserve balances: countries with large current account surpluses invest in safe and liquid assets. Purchase flows can dry up based on declining oil or commodity prices, reduced export margins, and other effects like currency targeting.
Safe-haven flows: insurance buying for return of capital, not return on capital. Inflation expectations will kill they flows.
Volatility of rates. Rates vol is likely only going up from here… someday… but no point in betting on when a decent move will occur. Betting on this is just a cheap widow-maker.
Correlation with equities: QE made bond and equity prices move together. A reversal of QE should reflect a lower correlation to equities, other drivers equal.
Note that equity-stock correlation driver affects both term premium and simultaneously drives the underperformance in risk parity trades.
A recent study by Marcello Pericoli provides an analytical framework that defines the determinants of the bond-stock correlation. His approach is to first define the expected nominal yield on bonds, then the expected nominal yield on equities, then combine them in a covariance function. All that is needed beyond that point is to plug input values into a correlation formula.
I rearrange his equation for nominal n-period bond returns B to instead define the term premium in the first equation below, and define the equity risk premium in the second equation below be rearranging his equation for nominal stock returns S:
The reason for my rearrangement is the identities above highlight common drivers of bond and equity returns:
Rising real rates raise premia for both bonds and equities via the higher discounting of future cashflows
Inflation shocks raise premia for both bond and equities
Growth shocks raise premia for stocks and lower premia for equities
One question that I had after reading the paper is what makes for a shock. So, I asked him. Here is our conversation:
Jeff McGinn <jeff.mcginn@fourpawnscapital.com>
Tue, Feb 5, 2:56 PM (23 hours ago)
to Marcello.Pericoli@bancaditalia.it
Dr. Pericoli,
Buongiorno. I enjoyed reading your recent working paper "MACROECONOMIC DETERMINANTS OF THE CORRELATION BETWEEN STOCKS AND BONDS". I do have a question about some of the equations in your paper, replicated below for convenience. Namely, how are "shocks" defined in your affine model? I am thinking that this does not imply any particularly "violent" change of a Dirac delta type. Rather, it just means that any shift in market expectations sufficient for possible regime change, correct?
Cordiali saluti,
Jeff
Marcello.Pericoli
5:37 AM (8 hours ago)
Dear Jeff,
I appreciate that you enjoyed reading my paper.
Yes. The second interpretation is my baseline assumption: any shift in market expectations is sufficient to change the regime.
Best regards,
Marcello
The takeaway is that the equations lead to a rigorous definition of the bond equity correlation, but not necessarily a predictive relationship. Any move in expectations, large or small, can lead to a bifurcation, break in stationarity, or a change point under the right conditions.
That said, together these equations obtain an explicit measure of the correlation between equity and bond nominal returns, shown below.
The first term on the right-hand side of the equation implies that higher uncertainty about the real interest rate increases the positive correlation of equity and bonds.
The second term implies that unexpected inflation increases the negative correlation of equities and bonds.
The third, fourth, and fifth terms capture the interaction effects between interest rates, inflation expectations, and growth and show their effects are ambiguous.
The cash-flow channel relates to whether or not inflation is demand driver or supply driven. As an illustration, think of the impact of oil prices declining due to lower customer demand or declining because of increased production available to customers. Without knowing the nature of what shapes inflation expectations, the effect of a shock can’t be determined.
The discount factor channel relates to the difference between nominal and real discount factors, which could have significant effect on the correlation of equities and bond if inflation becomes high.
The portfolio rebalancing channel relates to the relationship between dividend and real interest rates. It measures the relative performance or equities versus bonds on a real yield basis (multiples not included) and is positive during expansions and negative during recessions.
Summary
The macro trade du jour from, say 2010 to 2016 was risk parity. Be long risk, short USD, and hedge with long duration treasuries. It was a product of the central bank regime of the time and it worked well.
The Trump administration changed things by making fiscal policy through taxes and spending the major driver of economic policy. Further, Trump appointed a Fed chairman with the express intention of “normalizing” monetary policy in effect making it less like 2009-2016 and more like year prior to the Great Recession. The effect is Trump is Making Cash Great Again, keeping equities great, and more inflation to come.
It is intuitive to wonder whether the correlations between money market instruments and long-term bonds would make cash a bad stabilizer to the Trump trade. Be assured this is not a problem.
First, long-term rates decoupled from money market rates in the late 1980s and have never looked back. Long term rates priced outside of the Fed’s control, and they were effectively ignored by policy rules. Now this did change with QE. QE was intended to lower long-term rates because high long-term rates hurt growth among other things like add reserves to the banking system. However, QE is going away.
Also, companies in the US issue far more long-term debt in capital markets than short-term debt in money markets. Most homebuyers borrow at the 30-year mortgage rate. Capital allocation discounts long-term profits based on the long-term cost of funds. Thus, growth shocks minimally impact money markets relative to long term rates. Since there is little term risk, and inflation shocks do not matter either. Money markets are resistant to these shocks and act as an uncorrelated source of return to equities.
The problem was that returns were abysmal. Times change, and trades change with them.
Pericoli, Marcello (2018), “Macroeconomics determinants of the correlation between stocks and bonds”, Banca d’Italia, Working papers, No 1198 – November 2018