Financial Mathematics Text

Thursday, June 5, 2014

Relationship between VIX and Credit Spreads

This is motivated by a post from David Merkel of the Aleph Blog, called Buy Stocks When Credit Spreads are High, Sell When They are Low.

Merkel suggests that "credit spreads and implied volatility are cousins. When there is complacency, both are low. When there is panic, both are high."

I was curious about this relationship. The linear regression looks fairly ugly but grouping both implied volatility (measured by VIX) and credit spreads (measured by Baa over treasuries) into quintiles offers some interesting results. The daily data begins in 1990 and goes through May 2014. I removed dates which did not have values for both VIX and Baa spreads.

Here are the average values for the quintiles:

So do low (high) Baa spreads go with low (high) VIX?

There are 25 total boxes. Each row and each column adds up to 20%.

If there was no correlation between the variables, you'd expect to see around 4% in each box. I made the boxes yellow for the 3-5% range (green for greater than 5% and red for less than 3%). Essentially there should be no red or green boxes so there's obviously some correlation going on there.

It appears to be about as Merkel described with low VIX having a high When VIX is low (Q1), 73% of time Baa's are in the bottom (Q1) or second from bottom (Q2) quintile. On the other end, 57% of the VIX Q5 are in Baa Q5 and totals to just over 3/4 in both Q4 and Q5.

I was also curious what subsequent returns were. Here are the averages (arithmetic averages which has limitations). These are subsequent 1Y returns as determined by Wilshire 5000 total return index.

I'm not convinced any conclusions can be drawn from this. Some of the boxes don't even have enough observations to warrant much from them. (Consider VIX Q1, Baa Q5 which is only 0.05% of the observations.) I've yet to find any solid relationship between things like VIX and subsequent stock returns.

I only posted it because there's an academic tendency (and I do this too to some extent) to only report interesting relationships, results, etc. Sometimes doing an experiment and not finding something out is just as important.

Merkel suggests proceeding with caution in the low territories. My guess (assuming there's anything insightful from the above table), that VIX and credit spreads start going up right around the same time the market starts declining. They reach their high point when the market bottoms out (which is precisely when you want to buy.) That might explain why returns are worst in Q3-Q4 while Q5 has decent returns. But I'm not convinced one can extrapolate all of that from the above chart.

No comments:

Post a Comment

Some common OpenID URLs (no change to URL required):