This graph is courtesy of my colleague, Kevin Kliesen.
If you squint your eyes near the end of the sample, you'll see that Operation Twist appeared, on impact, to move short and long inflation expectations in opposite directions. The effect did not last long, however. The march downward continues--for now, at least.
Update: October 7, 2011
At the request of one of my readers, I had my RA (Constanza Liborio) plot a measure of inflation forecast errors over the last five years.
The inflation rate data is based on headline CPI. We use quarterly inflation at annual rates, and then subtract off the rate of inflation expected in a quarter, 2 and 5 years prior. The results are plotted here:
If you squint your eyes near the end of the sample, you'll see that Operation Twist appeared, on impact, to move short and long inflation expectations in opposite directions. The effect did not last long, however. The march downward continues--for now, at least.
Update: October 7, 2011
At the request of one of my readers, I had my RA (Constanza Liborio) plot a measure of inflation forecast errors over the last five years.
The inflation rate data is based on headline CPI. We use quarterly inflation at annual rates, and then subtract off the rate of inflation expected in a quarter, 2 and 5 years prior. The results are plotted here:
It seems that the error term on inflation expectations (expectation - actual) would be key data. For example, the expected annual inflation for the next two years (blue line) in August of 2009 was in the neighbourhood (proper English) of 0.5%. But the actual annual inflation rate on the CPI-U during that period was 2.46% (according to CPI data from the BLS). So that's a positive error of nearly 2%.
ReplyDeleteI understand you are saying the expectations have come down, but I think my question is this: how has the error term behaved over the past 4 years or so? Are market participants making larger errors or smaller errors? I don't think it's easy to predict inflation right now, and therefore I wouldn't put much stock in inflation expectations.
Prof J: good question; let me look into it.
ReplyDeleteDavid,
ReplyDeleteStephen Williamson thinks Operation Twist is irrelevant and neither had, nor will have, an effect on output, absolute prices, or relative prices.
http://newmonetarism.blogspot.com/2011/09/post-fomc-what-does-fed-think-it-is-up.html
What do you think?
"how has the error term behaved over the past 4 years or so?"
ReplyDeletethere are some speeches by foreign central banks on this very topic.
@JP Koning: I am inclined to side with Steve's view on this matter. It is difficult for me to see how altering the maturity structure of the Fed's bond portfolio (in the magnitude proposed) will have any significant impact on anything we should care about. I think that such portfolio rearrangements could have an impact in principle (say, on the basis of some "preferred habitat" theory), but I have a hard time believing that the quantitative effect can be large. I dunno. What do you think?
ReplyDelete@Anonymous: it would have been useful to provide links to those speeches.
ReplyDeleteI am agnostic about any effects on output.
ReplyDeleteBut 2 year bond prices fell and 10 year + bond prices exploded upwards within the first 180 seconds of the announcement. So Twist must have had some influence on relative prices in bond markets. Without Twist, the yield curve would be steeper.
JP, I'm willing to entertain the notion that the policy announcement had some effect on the yield curve. I wonder, however, how much it really "matters" in the present environment. There are dark forces operating out there that cannot, in my view, be remedied by twist operations.
ReplyDeleteJP Koning
ReplyDeleteHere is a graph of the Treasury term structure at the end of September for the past 5 years: http://1.bp.blogspot.com/-qRBkgBknhoQ/To3SkysqzTI/AAAAAAAAAB8/2_0d0lPKQY4/s1600/termstructure.JPG
The yield on 10-year and 2-year bonds are right now about where they are on the graph for September 2011. I don't see a huge effect from the policy announcement. Moreover, the steepness hasn't changed that much in the past three years, even though rates are down somewhat.
I don't see what real effect this portfolio shuffling could possibly have.
I won't argue with either of you about how much Twist really matters, you're probably right.
ReplyDeleteJust curious if you take the extreme neutral side of the argument as Steve does (i.e. no change whatsoever). I have troubles understanding his arguments about the effects (or lack thereof) of QE and Twist on markets, and since you and him have similar training, thought you might be able to clarify.
JP,
ReplyDeleteI'm not a macro-trained person like David and Steve; I'm a corporate finance person. But I agree with them that it won't matter much from a business investment perspective (which may be different from their perspective).
If we just look at this from a marginal effect point of view, the change in the interest rate is from around 2% to a little less than 2%. The effect on a corporation's hurdle rate is essentially zero. And corporations don't finance long-term investments with short-term financing, so the increase in short-term rates is irrelevant for them. It might bump up commercial paper rates a touch, but again that will be negligible.
WACC is also a long-run hurdle rate, more connected to a corporation's underlying risks than to bond market manipulations. In my view it's not possible for the 1950s dance moves to counteract the real troubles facing corporate investment.
But that's just me; I'm told one can't always argue in a micro-to-macro direction.
Sounds right, Prof J.
ReplyDeleteJP, I don't think Steve's view differs very much from John Cochrane's or Paul Krugman's on this matter. Just think of the two types of assets as constituting very close substitutes in the wealth portfolios of individuals. If this is so (and I think it is, apart from minor exceptions), then the composition of these assets is largely irrelevant.
David,
ReplyDeleteThanks to you and your RA for checking on the inflation errors.
It appears the 5-year errors are close to zero, but 2-year errors have definitely gotten bigger. So I'm half right. That's good enough for a Friday.
More importantly, it indicates that it is difficult to forecast inflation with any real accuracy in times of great uncertainty, as we have now. We may yet see big inflation, or it might be quite low. Who can say?
Related to David's point, have you (or your RA) done the exercise of comparing the forecasting errors of a random walk model with the one made by the analysts? Just to be curious... Thanks!
ReplyDeleteYour colleague, Kevin Kliesen, has given us a great insight about the inflation expectation. I want to learn to prepare such diagram for the presentation. Can you please tell me which software is the best for depicting the inflation points?
ReplyDeleteThanks,
Apoorva
HCBL Bank - Tathastu