Despite the somewhat startling conclusion (at least to me), the implications are pretty straightforwardly arrive at. From Michael Rosenberg, Financial Conditions Watch (Bloomberg, Jan. 27, 2010) (link added 1/29 8am) [not online]:
Fed Funds Rate Outlook — A Taylor Rule Perspective
With U.S. real GDP growth moving back into positive
territory in the second half of 2009 following four consecutive
quarters of negative growth (see Figure 1), the
economic forecasting community appears to be increasingly
optimistic about the U.S. economy’s growth prospects
for 2010-11….
…According to Bloomberg’s latest survey
of 57 economists (as reported on {ECFC}), the U.S.
economy is expected to grow by 2.7% in 2010 and 2.9%
in 2011 (see Figure 2). These projections represent a
significant rebound from the 2.5% decline expected for
2009.
Indeed, judging by expectations of the future course of
U.S. short-term interest rates, the market appears to believe
that the U.S. recovery will prove to be stronger than
a typical post-crisis recovery. Expectations for higher
short-term interest rates are reflected both in the Fed
Funds futures market and in the consensus interest-rate
projections of leading economists (as reported in
Bloomberg’s latest {BYFC} survey). As shown in Figure
4, the futures market is pricing in Fed rate hikes that will
take the Funds rate to around 0.75% by year-end 2010
and to around 1.00% by February 2011. The forecasting
community believes the Fed will be even more aggressive
as they expect the Fed to hike the Fed Funds rate to
1.75% by mid-2011.
While it is certainly the case that the Fed will eventually
have to push its policy rate higher, there is reason to
believe that the policy-rate path predicted in Figure 4 might
be overly aggressive. Indeed, as we demonstrate below,
the market’s projection for Fed rate hikes is not consistent
with the path forecasted by conventional Taylor Rule
models. If we input the Federal Reserve’s forecasts for
core inflation and unemployment into a variety of Taylor
Rule-type models, we actually end up with a zero or negative
Fed Funds rate projected for all of 2010 and, in a
number of cases, for 2011 as well.
The futures (red line) and mean forecast of the survey (blue) are depicted in the figure below.
Figure 9 from Michael Rosenberg, “Fed Funds Rate Outlook — A Taylor Rule Perspective,”Financial Conditions Watch (Bloomberg, Jan. 27, 2010).
In order to directly evaluate the range of possibilities for the Fed Funds rate, assuming the Fed follows a Taylor rule, Rosenberg considers four variants of the Taylor rule using the Fed’s own forecasts of core PCE and unemployment (using Okun’s law to convert the output gap to an unemployment gap.
iTaylor = (r N + π * ) + [ α (π-π* + β θ (U-UN)]
where:
(y-y*) = θ (U-UN)
and iTaylor is the target Fed funds rate according to the Taylor rule, rN is the real natural rate of interest, π is the inflation rate (π * is the target inflation rate), U is the unemployment rate (UN is the natural rate of unemployment). Note that the second equation is Okun’s Law (in levels) θ < 0.
- Benchmark, set rN=2%, π* = 1.5%, U* = 5.0%, α = 1.5, β = 0.5, θ=-2.
- Same as benchmark, but set β = 1.0.
- Same as benchmark, but set U*=6.0%.
- Same as benchmark, but set U*=6.0% and set β = 1.0.
(I’ve changed the notation a little from what’s in the original article.) As summarized by Rosenberg:
Inputting the Federal Reserve’s own forecasts for core
PCE inflation and unemployment into Rules 1-4, none of
these versions of the Taylor Rule would prescribe a policy
rate above zero in 2010. In fact, only Rule 3 would prescribe a Fed Funds rate above zero in 2011, while the
other three rules would prescribe a negative Fed Funds
rate in 2011. Overall, the simulated policy rate paths suggest
that Fed Fund rate hikes are not warranted in 2010,
and a case could be made that no rate hikes should be
forthcoming through at least the first half of 2011 as well.
The specific implied rates are reported in the table below, while the range of implied rates are delineated by the green lines in Figure 9.
Figure 8 from Michael Rosenberg, “Fed Funds Rate Outlook — A Taylor Rule Perspective,”Financial Conditions Watch (Bloomberg, Jan. 27, 2010).
One caveat: even with the variants, this is one specific formulation of the Taylor rule. In particular, it is static in the sense of using only contemporaneous data to generate the implied Fed funds rate, and is not forward looking in the sense of relying on deviations of future forecasted output and inflation. (It also does not incorporate a partial adjustment mechanism in the form of the lagged Fed funds rate.) Finally, the coefficients are imposed and not estimated.
However, I suspect that some of these modifications would not be consequential to the calculation. See for instance the implied values reported by Glenn Rudebusch back in May of last year, using an estimated Taylor rule (once again using contemporaneous values):
Figure 2 from Glenn D. Rudebusch, “The Fed’s Monetary Policy Response to the Current Crisis,” FRBSF Economic Letter 2009-17 (May 22, 2009).
As long as the Treasury can sell Treasuries, the Fed will be able to do whatever they like with the Taylor Rule. Pick a number, any number, they all work when the Treasury can sell bonds.
But still good news on the auction front. They are selling like hotcakes. 5s and 7s just flew out the door in copious quantities. Way above average direct bids, which people always think are CBs.
So we are safe for now 🙂
I’m surprised you are surprised at the conclusion. I would be very surprised by a rate increase in 2010.
I would not be surprised at a rate increase in 2010.
Firstly the fed might increase interest rates just so they are away from the zero bound, even though their “rule” suggests otherwise. Secondly if the fed has to worry about the vast reserves the banks are holding from being lent out and increasing inflation then they might have to pay a higher rate on reserves which would push the fed funds rate up. And finally as Bernanke mentioned in a recent speech the Taylor rule is just meant as a guideline; there is no consensus on what variables should go in there or what the value of the coefficients should be …etc.
The big caveat here is that the Taylor rule’s predictions have a tendency to overstate the fed fund rates target after recession periods (because of the Greenspan put). So, the big issue is more the forthcoming Bernanke put(he can’t stop justifying the low interest rate policy after the dot com bubble, will it be different this time ?). Thus, shall we really be surprised but nonetheless conclude that there is indeed a strong probability of no hikes in 2010 ?
I’m not that sure though since I believe that the New normal would mean that the “future” NAIRU will be something very different from the past two decades- 6% is definitely too small, 7-8% is more likely. I wonder what are the implications on the models results. But, clearly 0.75% is more the 95% upper bound. More predictions here:
http://www.market-melange.com/2009/12/22/melange-predictions-for-2010-fed-funds-rates-and-more/
Good post.
While the run through the Taylor Rule analysis is fun and all, the author got off to a really bad start. He insists on implying expectations of growth from expectations for the funds rate. The same survey which produces a median estimate for the funds rate also produces median estimates for real GDP growth. For 2010, that estimate is 2.70%. For 2011, its 29%. Both are below the long-run average for real GDP growth. Contrary to what the author implies from fed funds forecasts, actual measured expectations among the forecasters in question is for below-average growth. These forecasters expect a hike in the funds rate in the face of sub-par growth.
That is a reasonable starting point for its own analysis, and I don’t think it is all that hard to get started. Whatever ones own beliefs are about the potential for inflation – or asset bubbles – these guys seem to take seriously that, given a funds rate near zero, there is a risk of inflation or inappropriate asset price appreciation. That, or they just think the Fed is overwhelmingly biased toward inappropriately hiking rates. Certainly, the author’s premise that forecasters expect a faster than normal rebound doesn’t agree with available fact.
Thanks Menzie. In this case econometrics works well because you are forecasting what someone else who uses econometrics will do. I suspect that the use of the Taylor rule is a good forecasting tool because the FED is probably some variant of the Taylor rule to make its decisions.
The only caveat is that once the Taylor rule indicates an increase in interest rates it is doubtful that the FED would raise rates in the face of a continued contraction.
Never forget that GDP is more an indicator of government sending than it is of economic recovery so increases in GDP in the face of unprecedented deficits should not be a surprise.
The 4Q GDP is out! First approximation, anyway. Goldman wins the forcasting award, nailing it almost exactly.
“Official” numbers:
Topline growth at 5.7 percent, but only 2.2 of that is final demand, the rest being an inventory bounce.
Take that and stick it in your Taylor Rule. But wait, there is nowhere to put it. Besides, what’s the Fed going to do, try and raise the cost of credit to the Treasury so the Treasury borrows less? Haha. Monetary policy at it’s finest.
Link to Rosenberg article added to post.
Sorry I put very little faith in the taylor rule because it does not account for wealth/income inequality and changes in retirement.
Plus, I don’t believe that it works very well in either extreme of debt, which is usually too much.
Someone charted an analysis of the Taylor rule in the past decade or so, and the Fed Funds rate. They match well – even when the Fed held rates down a long time in 2002-2003. We could use the equation rather than the FOMC. Maybe that earlier analysis was presented here, and if so, it would be interesting to review in light of this new analysis.
Mike Laird
Actually, JDH and others charted it over the last decade and found that it called for much higher rates than the Fed set, especially during the 1% years.
But there are at least a dozen varieties around depended on what estimates for “output gap” and inflation are used, not to mention the empirical coefficients, so I’m sure someone could curve fit it to match Fed policy. (I used to have some cool curve fitting software that we used to generate equations from engineering test data. Just use the economy instead.) We could call the resultant equation the “Greenspan Rule” so it would confuse people less. (John Taylor should copyright that thing)