It’s flattering to see your research eventually make it into the mainstream … more than five years after you first published research on the topic.

Back in November 2005, in that month’s issue of Options For Investors, I noted that there was an amazingly synchronous relationship between interest rates and financial asset volatility.  The key, however, is that you have to time shift the rates about 2 to 2-1/2 years to get everything lined up.  That is, once interest rates start rising, volatility starts to rise about 24-30 months later.  At the time, I speculated that the very low interest rates promoted a “leverage-is-good-and-more-leverage-is-better” mentality.  Here’s what I wrote more than half a decade ago:

Take a look at this first graph.  It shows two indicators during the past twenty years.  If you look close, you’ll see something shocking.  With the sole exception of 1987, these two lines tend to trend together.  When one line goes up, so does the other.  When one line goes down, so does the other.  While their values may be different, their shape is remarkably similar.  For instance, look at the action in late-1988.  Both lines bottom out and begin to rise.  Then in late-1991, both lines peak and begin to fall.  The decline lasts all the way until 1996 when both begin to rise again.  They both top out around 1998 and remain at a relatively high level before beginning their descent in 2003.  You can see similar correlation from 1985 till the beginning of 1987.

So what are those two practically “conjoined” lines?  Volatility (the bold line) and interest rates.

The volatility that we are measuring is the one-year historical volatility of the S&P 500 index.  The interest rate is the three-month T-bill rate.  The thing about the T-bill rate is that it is shifted to right by 2-1/2 years.  In other words, the rate shown today is really the rate 2-1/2 years ago.  That is, the T-bill rate plotted on January 2, 2005 in the graph is really the T-bill rate on July 2, 2002.

What this relationship means is that if rates begin to rise, about 2-1/2 years later volatility should start to rise.  If rates begin to fall, about 2-1/2 years later volatility should start to fall.  The trend in interest rates predicts the trend in volatility 2-1/2 years ahead of time.

The key question is, how could this occur?  Is this truly a relationship that has an explainable, fundamental reason for being?  Or is it simply a coincidence?

My suspicion is that there may be a reason.  One likely reason is the increasing use of derivatives.  Typically, derivatives involve leverage, and that means someone is borrowing money.  As the cost of money becomes cheaper, the use of derivatives increases.  Because off-the-floor institutional traders using options and other derivatives tend to be volatility sellers more than volatility buyers, their increasing use of derivatives tends to depress volatility.  As the cost of money increases, the use of derivatives becomes costlier, and therefore, their use diminishes.  As that use dwindles, the “selling pressure” on volatility lessens, making it easier for volatility to rise…

The original article was picked up by several derivative-related publications, and even got the attention of the folks at The Economist.

What’s interesting is that the forecast was pretty much a blueprint of what happened.  For instance, the infamous Goldman-sold Abacus deal involved a purportedly sophisticated institutional investor (ACA Management) going short volatility via a derivative (a synthetic CDO based on a package of mortgage-backed securities), while Paulson and Co. made billions going long volatility.  Essentially Paulson bought a put whose strike price was based on the percentage of the subprime mortgages that defaulted.  ACA sold a put, looking for the income generated by collecting the premium Paulson paid.

That type of transaction was repeated over and over, and not just on Wall Street.  It was the global equivalent of shampoo: wash, rinse and REPEAT.  Low rates create the environment that allows the leverage to rise to unsustainable levels.  When the rates rise, volatility follows with a lag of a couple of years.

In 2005, I performed my analysis back into the 1980s, but no further.  I updated the research periodically, most notably in August 2008 where I noted that the rise in interest rates in 2005 and 2006 were indeed correlated with the increase in stock market volatility.

Here is the same graph (same data set, but with different dates and using a different program) updated to today.

Here’s the critical thing to take away: Note that the trends don’t go against one another.  For instance, while the lagged interest rates may have declined in 2004 and 2005, volatility only dropped a little.  Importantly, volatility did not go against the rate trend. In 2007 and 2008, the lagged rate trend was up.  Volatility did not decline any further, and eventually it started its ascent higher.

In other words, the trend in interest rates didn’t necessarily tell you what was going to happen, but it did tell you what would not happen.  Declining rates pretty much ruled out a rise in volatility!  And rising rates ruled out a further decline in volatility.

Knowing what volatility is unlikely to do is incredibly valuable information to a derivative trader, which is what you are if you trade options.

Here’s the thing [and I have a somewhat selfish reason for having not disseminated this to others prior to this post]:  What if this relationship between rates and volatility didn’t stand up to the test of time? You’ll note that in each of the articles linked to at the bottom, the time period studied is limited to the last 20 years … at most.

Because the correlation between lagged rates and volatility gave us such a terrific roadmap to 2007 and 2008, and because the volatility in October 2008 was so similar to the volatility behavior in October 1929 and October 1931, I decided to analyze the relationship as far back as I had data.

To do this, I divided the time periods up by decade as far back as 1900 and then ran correlation analyses.  It was then that I discovered that the relationship between volatility and 2-year lagged interest rates was a miserable FAIL.

In the 1930s, 50s and 60s, the relationship was closer to a three-year lag.  In the 1920s, it was close to a one-month lag.  And in the 1940s, there was hardly any relationship whatsoever between interest rates and financial market volatility.

In the 1970s, it was a 10-month lag.  Below is a chart showing the relationship during that decade.  You can see that 10-month lagged rates and volatility line up almost perfectly.

So what accounts for the variation in time delays?  Here’s my guess:  It could be that the time-delay variable is dependent on the amount of leverage in the system.  For instance, if leverage is low, then the time in which the rate change impacts volatility is shortened, but the impact is reduced.  If leverage is high, then the change in rates takes longer to impact volatility, but the impact will be bigger.  On the other hand, the fact that there is almost zero correlation in the 1940s hints that any correlation between rates and volatility over the past 25 years is just a coincidence.

Obviously, this warrants further research, which I plan on doing this summer.

– Don

Links:

U.S. Stock Volatility to Drop, Yield Curve Shows: Chart of Day

VIX Headed Much Lower in 2010 

Predicting Volatility With Interest Rates