Stressed asset price correlations
Updated: Apr 2
An analysis of asset price correlations to compare recent relationships in the equity sell off against those of previous equity corrections:
There is a new investment paradigm in interest rate markets as Central Bank Intervention is only effective in supporting asset prices through Balance Sheet expansion - rate correlations have changed and some traditional hedges are less effective.
Swap spreads, long dated interest rate implied volatility, and curves are less correlated to equity compared to previous equity corrections.
The best fixed income correlation remains inflation breakevens, which are positively correlated to equity.
AUDJPY remains one of the best correlations to the S&P. However implied vol now reacts.
The currencies of the highly leveraged SNACS countries are highly correlated to the S&P and to each other.
Oil and Copper are highly positively correlated to equity, whilst Gold's correlation flipped after announced intervention.
In 1Q19 , together with some very smart colleagues (Hat Tip to Omar and Chrissie), I looked at the historical correlations between different asset classes during the 4Q 2018 S&P correction. I then compared the results against the correlations during other equity corrections (dotcom, GFC, peripheral crisis).
This was a very informative study as it showed that some traditional fixed income v equity correlations had broken down in the 4Q18, whilst other new asset correlations were formed that indicated new potential hedges against potential future S&P corrections. To that end, this post looks at whether the conclusions of that study have held up in the 2020 sell off.
A quick summary of the conclusions of the 2019 study into the correlations during the 4Q18 equity correction:
a) Correlations that broke down (and thus no longer provided an equity hedge):
Real Yield correlations were very delayed and of no use as an equity hedge.
Swap spread correlations became positive - normally they were negative.
There was no 2s10s yield curve spread correlation to equity. CMS hedges did not work.
Long dated interest rate implied volatility (vega) correlation collapsed.
Realised FX volatility between Oct and Dec19 hardly exceeded implied levels of Oct19 (except AUDJPY and NZDJPY).
b) Correlations that held:
Inflation breakeven rates remained negatively correlated (which continued to confuse a lot of equity investors who thought inflation would rise if equity fell).
JPY, especially vs commodity currencies, notably vs AUD and NZD.
Oil maintained a strong positive correlation.
c) New correlations
Unlike previous corrections, Gold had a high negative correlation to the S&P.
This to us was a confirmation that the fixed income markets were operating in a new paradigm, whereby central bank intervention would come through balance sheet expansion rather than through more usual interest rate cuts.
Correlations in the 2020 turmoil.
1) Correlations to S&P:
I have taken daily prices over the last 4 months for various financial products (fixed income, fx, commodities). From that data set I have calculated the 1 month correlation of a particular product's price to the S&P for each trading day this year. From that I derive two charts for prices and correlations.
For example, below left is the price chart of the S&P and US 10yr Inflation breakeven rate (left) and below right the 1 month correlation between the two for each date this year. This shows how highly positively correlated inflation breakevens have been to the S&P. Therefore you can conclude that a short inflation position has still been a good hedge to the S&P.
What do the other correlations show (charts at the end of the post)?
Unlike 4Q2018, 2s10s curve was positively correlated to the S&P again (due the large rate cuts which did not occur in 2018).
2yr UST Swap Spreads correlations were again not high.
Whilst there was a small rise in Vega, the correlation remains low (it has now almost returned to multi decade lows).
Unlike during other equity corrections, the USD (DXY Index) had a statistically positive correlation to the S&P, but importantly recently this has turned negative (see thoughts on previous posts).
AUDJPY has been very highly correlated to equity market moves, as has the less liquid NZDJPY.
Like previous correction episodes, oil has had a very high correlation to equity index prices. Furthermore Copper has also had a very high correlation compared to that of 4Q18.
Unlike 4Q18, the correlation of Gold to the S&P has varied. Initially it was negative, allowing Gold to appreciate as equity sold off, but more recently this correlation has flipped (as both have rallied). Whilst this clearly suggests that markets expect both to gain from money printing, the important conclusion is that Gold is now a decent hedge when equity initially sells off.
2) SNACS Correlations
In the 10 years post GFC, 5 countries experienced a huge price appreciation in their real estate markets due in large part to a tremendous increase in their private sector leverage (Householders AND Corporates). These are Sweden, New Zealand, Australia, Canada and South Korea (SNACS).
Some of their currencies had already been weakening in the last 18 months as their asset bubbled had peaked, and in some cases popped, necessitating Central Banks to weaken their monetary policy. In the equity rout of 2020, the SNACS currencies have been very correlated to each other (and to the equity sell off). Below for example is the correlation between AUDUSD and EURSEK. These correlations were not as present in previous equity market corrections.
The above simple correlation studies highlights:
1. The change in fixed income correlations witnessed in 4Q19 were repeated in 2020 and are important in flagging that there is indeed a new and very different investment paradigm due to Central Bank balance sheet expansion. FIxed Income relationships have changed (something that Balanced and Risk Parity funds are coming to terms with).
2. Commodities and thus inflation breakeven shorts remain one of the best equity hedges (rather than long Nominal bonds).
3. AUDJPY remains one of the best correlated assets to equity. However, unlike in 4Q18, implied vol spiked for this FX cross during this equity correction - it is thus no longer such a "cheap" hedge".
4. SNACS currencies can now all be classed as "risk assets".
Correlation charts to which the above commentary refers to: