It is almost cliché to state that asset correlations have increased following the financial crisis of 2008, and that maneuvers such as zero interest rate policy may be driving this. My question is whether the markets have been fundamentally altered in the aftermath of the financial collapse. I did not test multiple different time frames. I simply selected in advance the relatively benign year 2006 and compared to 2012. Also, I used exchanged traded funds rather than individual equities (except for the business development company, Ares Capital (ARCC), since there are not mutual funds consisting solely of those).
This is an exploratory, idea-generating analysis intended to be a starting point rather than a destination. The assets under consideration, along with their stock ticker, and brief description of the asset class:
|Vanguard Total Stock Market ETF||US Broad Equity Index Fund|
|ARCC||US Business Development Company (possible private equity / junk bond proxy)|
|Vanguard Small Cap ETF||VB||US Small Cap Index Fund|
|iShares DJ US REIT ETF||IYR||US Real Estate Investment Trust Index Fund|
|EFA||International Equity Index Fund|
|TLT||Long-Term Bond Index|
Next, I performed a correlation matrix of one year duration at vanguard.com, for 2006:
Then I ran the same analysis for 2012:
Now I will compare 2006 to 2012, for VTI (US Broad Equity index):
|Vanguard T StMk Idx;ETF||Vanguard T StMk Idx;ETF|
|Vanguard T StMk Idx;ETF (VTI)||1||1|
|PowerShares DB Cm Idx (DBC)||0.061||0.646|
|Ares Capital Corp||0.506||0.818|
|Vanguard Sm-Cp Idx;ETF (VB)||0.957||0.965|
|iShares:MSCI EAFE Idx (EFA)||0.866||0.908|
|iShares:Dow US Rl Est (IYR)||0.698||0.843|
|iShares:Barc 20+ Trs Bd (TLT)||-0.082||-0.76|
|T. Rowe Price International Bond Fund (RPIBX)||0.22||0.56|
|SPDR Gold (GLD)||0.081||0.407|
Observations of other asset classes to US equities:
- Commodity fund correlation to US Equities has increased dramatically, from 0.06 to 0.642.
- Likewise, Gold ETF has shown a similar increase from 0.08 to 0.40.
- The BDC, ARCC, appears significantly more highly correlated now, compared to 2006.
- US Small cap correlations remain very similarly high.
- US REIT correlations have increased from 0.69 to 0.84.
- International equities remain similarly highly correlated.
- The single biggest change in correlations is the US Long term Bond fund going from a non-correlated -0.08 to an anti-correlated -0.76.
- International bonds, however, have become MORE correlated to US Equities, in contrast to US Treasuries.
The financial crisis of 2008 made correlation analysis worthless during a time of illiquidity. Such conditions can and probably will happen again at some point in the future. Simply adding one or two additional holdings to your portfolio will not isolate or protect you from the consequences of such market conditions.
I have shown how a tool such as a correlation matrix can be used to independently test a hypothesis, such as the commonly-held notion that asset classes are now more highly correlated than prior to the crisis. My summary conclusion is that among non-US Treasury ETFs that I tested, yes, correlations were higher in 2012 compared to 2006. Even an international bond fund demonstrated a higher correlation to US equities over this period. Among the limited selection of mutual funds that I tested, only the US Treasury bond ETF, TLT, showed a decreased correlation to US equities decreased over this time frame. I believe this confirms the often-heard claim about "risk-on" and "risk-off" trades and suggests that asset classes are moving more in unison. While I speculate that this is due to Fed interest rate policy, that is an untestable hypothesis.
The difficulty in applying these findings to a portfolio is that if US Treasuries are one of the few or only things that have not shown an increased correlation to US equities, there is a problem. With correlations now close to -1, all holding US Treasuries (via TLT) really do is reduce the beta of an equity-heavy portfolio. In other words, the anti-correlated movements of TLT merely offset the movement of stocks, and reduces the gains as well as the losses. That isn't true diversification. It is hard to find assets with correlations close to zero relative to equities. The correlation closest to zero was with GLD, but at 0.40 is still directionally related to equities.
Thanks for reading and have a fantastic 2013!