Home > CPI > A Welcome Retreat, But Is It Enough?

A Welcome Retreat, But Is It Enough?

The highest 10-year yields since mid-June were the reward for the news that President Obama had executed a remarkable pirouette and agreed to extend the duration of the Bush-era tax rates. 3.17% is not exactly an asphyxiating rate of nominal interest, but compared to the 2.39% rate that prevailed in early October or the ~2.55% rate that was the standard when Bernanke first broached the subject of QE2 back in August, this seems almost dizzying.

Stocks initially rallied on the notion that the deal will help stimulate the economy, although equities hadn’t exactly been selling off on the alternative – other investors seem to have noticed that, and the market ended flat. Volume was much heavier than it had been over the last few days, actually more than double Monday’s tepid volume, at 1.57bln shares. An unchanged market isn’t awe-inspiring, but the internals were encouraging (at least for technical types). I personally think that the growth estimate baked into current index levels is very generous, but there is no question that such an estimate is more plausible today than yesterday – at least, in the short run.

The about-face from Obama is interesting. Whereas the Administration previously had been insisting that some Americans get higher tax rates and that the tax code get even more progressive, it finally agreed to allow tax cuts for all on the condition that there be … even more tax cuts. In addition to the two-year extension in the marginal tax rates, there were increases in various credits such as the earned-income tax credit and a small reduction in payroll taxes.

Assuming that the President still has the charisma to get his own party to agree to passage, and assuming that the Republicans don’t also do an about-face now and insist on higher taxes, the agreement effectively removes the possibility that gridlock will cause a large rise in taxes and crush consumption. It is a clear positive for near-term growth. And, to the extent that the Federal Reserve was taking the oars in order to pull the economy through that potential Q1 swamp, it makes QE2 less necessary (since it was mostly political cover anyway, and a risky form of it at that). So this is all good news. We may as well enjoy it, because after Friday’s Employment report good news seems to be at a premium. The chart below shows the S&P index against the Citigroup Economic Surprise Index for the USD. You can see that the improvement in the indices from September occurred in concert with a gradual improvement in the tenor of the data, but the report on Friday really damaged the underlying sense of consistent positive surprises in the data.

Friday's economic surprise removed one underlying support to the rally.

As I noted yesterday, there is really not much in the way of significant data this week, so the Citi Surprise index isn’t going to improve very much for a little while. I don’t know whether this will be significant: I am just pointing it out.


I have previously on this site presented the output from my model(s) of core inflation, which indicated a trough in year-on-year core inflation in September. (I have two marginally different models which I usually run with two marginally different sets of parameters in order to get a sense of the sensitivity of the model and a range of potential estimates.) The actual trough, because of base effects, was probably October’s figure; we will find out next week. However, for more than a year now I have always presented the models’ forecasts with great big, flashing-letter warnings because the models treat “core inflation” as a homogeneous slurry and do not differentiate between the various subcomponents of core inflation.

Ordinarily, that is not a fatal assumption because to the extent that there are “special effects” that perturb one part of core inflation, they tend to be random (and mean-reverting) and typically not large. But over the last couple of years, as I have noted repeatedly, core inflation has really been composed of two heterogeneous parts: housing, which is descending from a bubble and can be expected to exhibit depressed price pressures regardless of the underlying dynamics of the economy, and core-ex-housing. These are roughly equal parts, and so a model that predicts the sum of two basically different variables should rightly be used with caution.

I finally got around to bifurcating my model and re-estimating the coefficients with respect to forecasting ex-housing core inflation. That is, using essentially one of the same models as I was using before I can now produce a forecast of ex-housing inflation. I can combine that with a very simple model of housing inflation (such as one based on the inventory of existing homes, which I illustrated here) and solve the problem that the model was trying to predict something that is inherently not “normal” right now.

The encouraging outcome is that the resulting forecast series is very similar to the forecasts made with the naïve model that ignores the dynamics in the housing market. The chart below shows a whole riot of lines: The thin solid (green and orange) lines are the old models; the purple dashed line is the “new” model using the same explanatory variables but predicting housing and ex-housing separately and then combining the results. The thick red dashed line shows the average of the three models, and the heavy black line shows the actual results.

The new model (purple line) suggests that the biggest drag from housing is mostly past.

One quick technical observation: don’t read too much into the little spike running up into January followed by a setback. That spike occurs because I am using set lags rather than distributed lags, and it just happens that some important lags all lined up with interesting moves in the variables at different times. The true forecast path is better thought of as a smoother line connecting the last solid-black point and the last dashed-red point. Someday I may refine the methodology but this serves my purposes for now. The point estimate right now stands at 1.5% core CPI for calendar 2011. The risks lie predominantly on the upside; the main risk is that a sudden shift in the monetary dynamic abruptly flushes a large quantity of bank reserves into M2 where it can do real damage (see my comment here to get a sense of how big a shift that would be).

The inflation market currently is pricing 1.4%-1.5% inflation for 2011 (it depends a little on how you look at the seasonal pattern in December 2011 versus 2010), so it seems to be approximately fairly priced for a change. Inflation for 2012 is priced at 1.7%, however. That not only seems a little low to me, but monetary policy errors would probably have a much larger effect on 2012 inflation than on 2011 inflation (at least some of which is already “baked into the cake”). This makes Jul-13 TIPS, at -0.5% real yield, appear to be an attractive alternative to nominal Treasuries at 0.685% for the same maturity. Neither one is going to rock your world performance-wise, but if I were an institutional money manager I would be optimistic that inflation will exceed 1.3% per year for the next couple of years.

Categories: CPI
  1. Lee
    December 8, 2010 at 11:28 am

    With housing being the biggest expense, my brain can’t seem to get why you want to exclude it. And with labor losing to outsourcing, the real estate market still a problem, r.e. taxes going up, government jobs decreasing, etc., my feeling is that everyone will be deflation spiraling. Is there an easy, obvious factor you have in mind that I’m missing?

  2. December 8, 2010 at 12:12 pm

    We exclude factors that do not help us forecast inflation. Housing is not responding, and should not respond, to the factors that influence inflation: money supply being the most important one of them. We know that housing inflation for the next year or so will be around zero or less, regardless of what happens to money supply. So if we want to measure the efficacy of policy, we need to look at things that might actually respond to policy!

    Taxes going up, government jobs, etc…are all fiscal variables. They’ve never been shown convincingly to affect inflation. Yes, that’s right, Employment is not directly related to inflation – the supposition that it does is based on a confusion of real and nominal quantities. Employment supply and demand affects REAL wages.

    In fact, the only variables that consistently test out as clearly and closely related to inflation, if you do the research correctly, are money supply and leverage (the latter of which is related to money velocity). My model has only a couple more factors that mostly provide alternative ways that money is transmitted with different lags (e.g., the fx value of the dollar) rather than representing new factors.

  3. Lee
    December 9, 2010 at 9:49 am

    Thanks for the detailed answer. I see that my thoughts weren’t about policy but were more cosmic. To me policy-maker hands are tied by the quality of most new jobs and by future layoffs via toxic assets. It feels like there’d be a limit on how far bubbles can get from that, since we seem to be popping them at this level.

  1. No trackbacks yet.

Leave a Reply

%d bloggers like this: