An Intro to the Dreaded #ConvexityVortex
Note: Financially sophisticated readers are probably served best by skipping the words and looking at the charts, you’ll know what I mean.
If you are a homeowner, you probably know that if mortgage rates decline, you can refinance your loan in to a cheaper loan. In financial parlance that is called an option. You have the option but not the obligation to refinance at a new and lower rate or pay your loan early. Although this is the standard in the US home loan market, it isn’t the same world-wide and because of the large size of the country, our culture of home ownership and the average price of these homes, the size of our mortgage loan market is the largest in the world. Most of these mortgages are owned or securitized by Fannie Mae, Freddie Mac or Ginnie Mae into bundles, Mortgage Backed Securities, which are then sold to investors.
At the start of the Global Financial Crisis, the Federal Reserve lowered the target rate at which banks (used-to) lend money to each other to between 0.00% and 0.25% and that began an 8-year cycle of falling mortgage rates, during which most existing home owners had the ability to refinance their loans at new lower rates — and, for the most part, they did! This was great for consumers and home-owners because lower interest rates mean that your money can go further. For example, the decline in mortgage rates from 6.5% in late 2008 to 3.5% today would reduce your monthly payment by 28%. At 6.5%, buying a $330,000 home would require a $2,046 monthly payment, but at 3.5% that payment declines to $1,472.
Those savings went a long way into helping the economy stabilize during and after the 2008–2009 recession, but they did come at a cost. For every party paying interest, there is another one receiving it. In the case of mortgages, investors interest income declined with borrower interest expenses. Lower interest securities have a higher duration, a measure of how sensitive the price of a bond is to a change in the interest rate. The higher the duration, the more the price of a security declines if interest rates rise and the more the price of a security increases if interest rates decline. In the finance world we generally consider higher duration to be a proxy for higher risk. Mortgage Backed Securities are special because the option that I wrote about earlier means that if interest rates decline, the price doesn’t go up much because investors expect that many borrowers will opt to refinance, so they will receive lower interest payments in the future. That property is called negative convexity. What that means, in plain English, is that when mortgage rates decline, the price of existing mortgage securities becomes less sensitive to changes in interest rates or, in some peoples’ opinions, less risky. That means that they are less-likely to hedge against the risk of interest rates rising. Hedging is when you transfer an existing risk (and it’s associated return) in your portfolio to a different party.
Normally, none of this matters too much but, if a lot of people have mortgages at the same interest rate, it can lead to a condition where, when rates are low, nobody needs to or wants to hedge and as interest rates rise, more people want to hedge. That can lead to more people wanting to transfer the risk out of their portfolios than there is people willing to accept that risk and return. Informally, we joke about this phenomenon and call it the “convexity vortex” because it is the point at which too many people looking to protect themselves from falling prices cause prices to go down, causing more people to protect themselves and so on. It’s a lot like a self-fulfilling prophecy or a self-feedback loop.
My concern is that the multiple visits the purple line in the first chart has made to the ~3.5% level visible in the horizontal line has caused most of the mortgages in the US to be clustered at the same interest rate levels, which can trigger the aforementioned convexity vortex. In fact, 85% of US 30-year mortgages from Fannie Mae, Freddie Mac or Ginnie Mae have coupons (the % of total balance you pay every year in interest and principal) in the narrow corridor between present rates and 4.5%.
This means, roughly, that if the average coupon rate for new mortgages were to increase by 0.5%, suddenly a lot of people who did not care to hedge today may want to start hedging, which could start the chain-reaction I described above. This is by no means something that happens often, but we did see it play out a lot like this during the summer of 2013. Of course, this sudden impulse has finite strength and life, after about 1.5% in rate increases the negative convexity effect ceases to have any effect. What are the chances that mortgage rates will rise 0.5% suddenly? Not very high, but not trivial, either. Mortgage rates are affected by 2 things: What coupon rate the market demands for owning these securities (1st chart, top section, white, yellow and red lines) and how much the companies that handle all the details of a loan earn (first chart, bottom section). The first is mostly governed by the level of interest rates and the latter is mostly governed by how many people want loans at any given time and how much it costs to hedge risk of interest rates changing between when you apply for a loan and the closing date.
As you can see above, mortgage rates are currently on the lower side for the present level of interest rates, were that to revert to its regression level, it would mean an increase of 0.25% in mortgage rates. That, combined with a 0.25–0.35% increase in longer-term interest rates could be enough to risk the dreaded convexity vortex.
Some participants have the stance that the potential for such an event is low because the Federal Reserve, as part of its Quantitative Easing (QE) programs, purchased a very large amount of these securities for its Systems Open Market Account (SOMA). However, even after subtracting those purchases, we still have roughly $2 Trillion of securities at risk of experiencing duration increases if interest rates were to rise. The Fed’s ownership didn’t stop the price declines in 2013, when they were still expanding their MBS holdings; I doubt it can today, when it isn’t.
Finally, we have to recognize that, in financial markets, a portion of the value of any security is a function of how much people like or dislike that type of investment. “Buy low, sell high,” most often means “Buy when it’s hated, sell when it’s loved.” And a reality of markets is that people like things more when the price is going up and less when it is going down. This often leads to participants selling only because the price went down. We generally refer to these participants as “momentum” or “trend followers” and they are a significant proportion of market participants. I write this today because over the last few years, not only have we developed the issue described above, but we have also issued very large amounts of US Treasury and Corporate debt. A turn of bonds from “loved” to “hated” or even “feared” could trigger price declines that would shock many investors who believe bonds are inherently “safe.”
While we at New River Investments are, for the most part, pretty upbeat about the economy, we also recognize that it is important to be cognizant that there is some risks that are independent of that and we need to be aware of them before we make any investment decisions.
Geek Note: By my estimates (and I invite you to send me yours), a 25bp increase in primary-secondary spread and an increase of 100bp in interest rates rates would lead to a 100bp increase in mortgage rates and essentially extend most of the stock of MBS, adding the equivalent of ~$350B in 10-year-equivalents of duration to this market. Whether it is direct hedging, or active or passive maturity shortening, the fact is that this market needs to be more willing to aggressively buy bonds and hold duration as prices decline in order to avoid a “jump” event from +50 to +150 (full extension) and given rising inflation and how other asset classes have reacted to changes in trends, I just don’t see that demand being there. That $350B estimate excludes any extension from corporate or municipal bonds due to my inability to compute it without more granular data.
