Wednesday, April 11, 2007

Too many trades?

In response to a recent comment, I made the off hand remark that the average bond mutual fund has 200% turnover. I looked it up in Morningstar's database, and in fact, the average turnover is 203%.

Regular reader Vivek asked me about the common claim that stock managers trade too much. Given that transaction costs are undoubtedly higher in the bond market compared to the stock market, and given that the average stock fund's turnover is only 84% (according to Morningstar), does that mean that bond mangers trade too much also?

First on the transaction cost issue. There are two basic types of transaction costs: overt costs (commissions) and market impact costs. The first is obvious in the stock market, but not so much in the bond market. A bond buyer rarely knows what commission their broker is being paid. This is because the commission is embedded in the price. But even in the most liquid
markets, where the bid/ask is only a tick, that's equivalent to a stock commission of 3 cents/share on a $100 stock. I know institutional buyers can do better than that in most cases. So while the Treasury and TBA mortgage market might have the same transaction costs as stocks, corporates, agencies, and municipals certainly don't.

In terms of market impact, that's a tougher question. I don't trade stocks professionally so I'm loathe to make a ill-informed opinion about that market. 99 times out of 100, a medium sized bond manager (say $1 to $10 billion) doesn't do trades large enough to influence the market at all. The exception might be in a off-the-run corporate where the trader is dead set on a specific bond. In order to find the issue, a brokerage might have to entice a current holder of the bond with a particularly strong bid. Anyway, that's a rare circumstance, and a smart trader can mitigate this kind of problem.

But trading is very different in bonds versus stocks. Consider the most fundamental reason why one invests in a bond: the income generation. Consider also that with any fixed rate, bullet bond, the income generation is known at time of purchase.

This makes it very easy to isolate the differences in risk between two bonds. For example, let's say that I own a Bear Stearns bond maturing in January 2017. Let's say that Lehman Brothers also has a January 2017 bond. Both are fixed rate, non-callable issues. Both have identical credit
ratings. Let's say that I can sell my Bear Stearns bond at a yield of 5.70%, and can buy the Lehman Brothers bond at 5.80%.

I've increased yield by 10bps, and the only change in risk is now I'm exposed to Lehman's credit as opposed to Bear's. Let's say that I think the companies are essentially the same risk, which isn't much of a stretch. So I've done this trade and picked up the extra 10bps.

Is this example realistic? Yes. The 10bps differential might be a bit on the high side for an obvious trade like this one, but stuff like this is available all the time in the bond market. Why does it happen? Typically because there happens to be more supply of one name versus another (such as it was just issued), or that a particular dealer is short one name and really needs to buy it.

Other examples might include trading in the same name but altering maturity, trading in the same name but going from fixed to floating, trading in the same name but going from senior to subordinated, etc. People do similar trades outside of the corporate market as well. For
example, I did a series of trades involving one-time callable agencies back in 2004 and 2005. I had the view that the short-end of the curve was going to rise and I could therefore sell agency option risk and gain extra income.

A one-time callable is a bond which can only be called on one single date. Once that date passes, there is no more option risk. With all of my bonds, that date passed without the bond being called, because as I correctly forecasted, short rates just kept rising and all of the calls were out of the money.

Anyway, once the call expired, the value of the bond has increased by the value of the option (now worthless). But maybe I still had the same view on the short end of the curve, so I went ahead and sold my old bonds and bought new callable bonds. Let's say that the new callable bond I bought yielded an extra 50bps versus the bond I sold. If yields rise, the call expires worthless and I get to keep the 50bps in yield plus the bond will appreciate once it becomes a bullet.

So what's my point with all this. Its that when trading one bond for another, the relative value is more certain than when trading one stock for another. That is because the fundamental reason for owning a bond can be summed up in a mathematical formula -- yield. The price appreciation on a given stock is far less certain. So if I sell a Bear Stearns bond and buy a Lehman bond and pick up 10bps in yield, I know exactly what I gained in the trade: 10bps. If I did the same with the stocks, it all depends on which one appreciates more than the other. That's far, far, far less certain.

So that's one reason why turnover is logically higher in bond portfolios versus stock portfolios. In fact, bond managers who don't take advantage of these kinds of opportunities when appropriate are probably giving away performance.

There are two other facts of bond market life that inflate the turnover statistic. First, bonds create cash flow. While it would be possible to literally have zero turnover in a stock portfolio for several years, bonds mature, pay sinking funds, get called, MBS prepay, etc. So a bond manager
has to do reinvest cash even if s/he is trying to keep turnover as low as possible.

Second, most managers use Treasury bonds as a means of locking in spreads. This is how it works. Let's say that I own 10-year Rescap bonds, and I no longer like the credit and want to sell. So I collect bids from dealers in spread vs. the 10-year Treasury. When I go to actually sell the bond to the winning dealer, I ask the dealer to also sell me the same amount of the
10-year. Now in selling the Rescap issue, I didn't really want to decrease my overall credit allocation. So now I want to spread out that cash into various corporate names I already own. I go out into the market and find the best offers on each of the names I want, and as I execute those buy trades, I simultaneously sell the 10-year to the executing dealer. Why do managers
do this? Because it prevents a sudden move in the overall Treasury market from impacting the success of the trade. If I were to sell the Rescap bonds at 10AM, and there were to be a huge rally at noon, then without buying the Treasury bond in the interim, I might be forced to buy my other corporates at higher prices. In effect, this would be like getting poor execution on
the Rescap bonds. As you can see, this practice results in what was 2 trades becoming 4.

I hope this explains why bond managers trade so much, and why it might not be such a bad thing.


Vivek said...

That was an excellent explanation. Coming from the quantitative research side, I'm completely divorced from practical trading realities. I am the worse because of it.

Your post gives me some ideas...

Anonymous said...

Wow. Double wow. Great, great post. I learned a lot. And thank you very, very much for hyperlinking jargon.

Great blog!