To know a person, watch what they do, not what they say. Investors shall do the same to the Federal Reserve.
The stock market collapsed two weeks ago when Mr. Bernanke was talking about merely the possibility of "tapering," then bounced up when other Fed officials rushed to soothe the fear. It appears that too much attention was paid to what the Fed has said.
But what did the Fed do? Recent POMO data revealed a significant deviation of the Fed's behavior since June 19th, the very date when the "tapering" bomb was dropped. It may suggest that the Fed in itself is anticipating the "tapering."
The Fed deploys monetary policies by conducting Permanent Open Market Operations, or in short, POMOs. Capital is injected into the financial markets when the Fed buys securities from banks. The securities are moved on to Fed's balance sheet. In exchange, an equal amount of cash is put into banks' pockets. Capital is withdrawn when the Fed sells back.
As a part of its QE operation, the Fed made a net purchase of $248B Treasuries from December 13th, 2012 to June 18th, 2013. As shown in the right chart in Chart 1, besides selling a small amount of 3-year notes, the Fed made purchases on all the longer term durations: 5-year, 7-year, 10-year and 30-year, and the Fed spent cash evenly among them. The Fed didn't seem to care about the differences of durations. At least it was like that before Fed chairman mentioned "tapering."
Chart 1: POMOs of Treasuries, after "tapering" vs. before
But things changed after that. Since June 19th, the Fed started to overweight short-term durations. As shown in Chart 1, the Fed bought almost $10B 5-year notes past two weeks, but purchased only less than $6B 30-year bonds, roughly a 2-to-1 ratio.
Fed Is Shortening Duration
Shortening duration is a textbook strategy bond investors will adopt when they expect rising interest rates. As bond prices and yields move in opposite directions, rising interest rates equal to diminishing mark-to-market value of a bond portfolio. Because yields of long duration bonds tend to rise faster than do short ones, shortening duration - replacing long term bonds by short term ones - helps mitigating exposures to interest rate risks.
But the Fed is not yet another Treasury investor. It is a government agency with a dual mandate to maintain price stability and promote economic growth. Nonetheless the Fed is solely responsible for its balance sheet. Although stimulating the economy is the ultimate goal of THE Fed's Treasury purchases, grossing a handsome profit (or losing less money) is a sweet byproduct.
My theory is, the Fed in itself is anticipating the "tapering." It also knows well that its Treasury purchases artificially held back long-term interest rates and removing the punch bowl will lead to immediate rises of interest rates. Therefore the Fed started to shorten duration to manage the potential loss of its TTHE reasury portfolio.
But could the Fed see other reasons that may also lead to rising interest rates? If so, Fed's shortening of duration may not be connected to the "tapering." Someone argued that economic growth always comes with benign inflation, and the Fed may just foresee a better economy.
Not necessarily. It is reported the Fed's QE3 is capable of absorbing 90% new issues of Treasuries. The Fed can move Treasury yields single-handed. With or without a better economy, the Fed can pin interest rates if it decided not to taper.
Mr. Bernanke repeatedly told that the decision is linked to future economic data. Which direction the economy is heading? The Fed could be as clueless as you and me.
Chart 2. TLT
Another possibility is the Fed is not shortening duration, but merely reacting to price fluctuations. But this is also less likely. Using TLT as a proxy, 30-year bond prices kept dropping in May and got stabilized only in late June. If the Fed was reacting to price fluctuations, it should have sold in May and bought in June. But as shown in Chart 3, the Fed bought the same amount of 30-year bonds in May and June.
Chart 3. POMOs of Treasuries, June vs. May
Yet it could be that the Fed is reacting to new supplies of Treasuries. During the 2-week window between June 19th and July 3rd, the Treasury Department auctioned $35B 5-year notes, but no 30-year bonds. Could it be that the Fed responded by buying more 5-year than 30-year?
In the 2-week window between May 19th and June 3rd, the Treasury Department auctioned $35B 5-year and no 30-year as well, but the Fed bought more 30-year than 5-year, as shown in chart 4. The inconsistency indicates that new supplies of Treasuries may not be a factor in the Fed's purchasing decision.
Chart 4. POMOs of Treasuries, 2 weeks in June vs. 2 weeks in May
Sharp-eyed readers may have noticed that I charted only POMOs of Treasuries, while QE3 includes purchasing of both Treasuries and agency mortgage-backed securities. At a pace of $40B per month, roughly the same size with the $45B per month of Treasuries, the purchasing of agency MBS shouldn't be ignored.
A recent Seeking Alpha article argued that QE3 is all about Treasury purchases. I don't want to repeat the arguments, just want to point out that from the perspective of POMOs, we can reach the same conclusion.
Chart 5. POMOs of Treasuries, before and after 12/13/2012
On December 13, 2012, the Fed announced that on top of the $40B agency MBS purchases, it would add $45B Treasury purchases. As shown in chart 5, in the 6 months before that date, the Fed bought long-term Treasuries but sterilized it by selling short-term ones. The net pace is $3.9B per month. After that date, the Fed stopped selling short-term Treasuries, and the net pace jumped up by 10 fold to $39.6B per month. Before that date, the stock market was confined in a box. After that date, it rallied big time.
Chart 6. Stock market and POMOs of Treasuries
People may still argue that two weeks of data, especially when being compared to six months of data, is less statistically significant. I have no cure for this. Instead, I would suggest readers keep monitoring what the Fed does.
A Simple Program to Monitor Fed
I would like to share my simple program that did the magic to produce all the revealing charts. It is a Python script that downloads POMO data from the Fed website and categorizes the purchases by the durations. Users can supply the start and end dates and the script will do the rest.
I'm sure most of the analysts out there use Excel to process data, though I'm not sure how many of them know how to write program. But writing program is one thing. Reading and tweaking an existing piece is another. I'm confident that most Excel experts will have no difficulty to understand and modify my script. They can keep an eye on the Fed, verify my thought, or come up with better ways to slice and dice POMO data.
The code is shared via the Cloud9 website. It is an on-line Python editor and executor. In this way it saved all the troubles of downloading a programming language package and getting it running on one's computer.
Click on the link and you can view my code. Sign up for free and then chat with me so I can grant you write permission. After that you can edit the code and run it. Everything is online and hassle-free.
I'm not affiliated to Cloud9 but I'd like to promote its technology. It enables sharing and collaboration. I don't know but I envision there can be a community that, equipped with this technology, is able to better understand the market and collectively we can improve the transparency.
Let's get started and comments are welcome.