November 17, 2005
Inflation under control
Unlike September's troubling inflation statistics, yesterday's release by the Bureau of Labor Statistics of the October consumer price index is more reassuring.
Gasoline prices fell 5% in October, says the BLS, and I expect to see an even bigger drop in this component when the November data get released next month. That improvement helps explain why the change of the CPI over the last 12 months now amounts to a 4.3% annual inflation rate, compared with the 4.7% annual rate that BLS reported last month for the September 2004 to September 2005 change.
What's one to make of these big ups and downs? For reasons I've discussed earlier , , , at least for purposes of a guideline for the Fed and predicting what's coming next, I prefer to look at the median CPI rather than the CPI as calculated by the BLS. Macroblog reports that this measure continues with its incredibly boring performance, recording an inflation rate of 2.3% for the 12 months ending in October, just as it had in September and August and July and June.
If Bernanke is smart (and I already told you that he is), if he does go with an inflation target for the Fed, the goal will be announced in terms of the median CPI rather than the usual CPI. If the Fed announces a target for the regular CPI, which can wander off wildly regardless of what the Fed does, the pundits will never tire of warning us of how far off we're going to be from the target and how dire that might be. But if the Fed had announced that it had been aiming for a target for the median CPI for the last year of 2.3%, and hit it right on the decimal month after month, wouldn't the new kid look like a genius and a half!
Of course, if the Fed announced a target of 2.0%, we'd be missing the target month after month. Moving the economy away from that 2.3% tendency might prove to be like trying to turn a battleship. But if you happen to like 2.3% as a target (and hey, it's a fine number with me), that fact could be very reassuring.
And that may be just the way the bond market is seeing it as well. With the 5-year Treasury yielding 4.43% and the 5-year inflation-indexed 2.03%, it seems the market is figuring on something like 2.4% annual inflation over the next 5 years.
Even if this is a battleship, please, nobody rock the boat.
Posted by James Hamilton at November 17, 2005 03:11 PMdigg this | reddit
Thats the way to sight in a rifle. See where it shoots, and then set the sights to the same place.
Posted by: Bill Ellis at November 17, 2005 05:10 PM
A little off subject, but what do you see as the primary cause of inflation?
The reason I ask is that many of the economists I read give different explanations on the reason inflation occurs.
Also is there any reason why the fed would cease to publish the M3 monetary aggregate?
Posted by: TH at November 17, 2005 06:44 PM
per cover of today's chicago tribune:
median household income
in IL, adjusted to 2004 $s
year 2004 - $46,132
year 1999 - $52,515
what is the impact on inflation?
Posted by: nate at November 17, 2005 07:33 PM
TH, as I described in my post on stagflation, any number of factors contribute to any given year's inflation fluctuations, but very high sustained rates of inflation are invariably due to money growth.
As for abandoning M3, I personally never paid any attention to it and agree with Dave Altig that discontinuing it is no big deal. I suppose that William Polley has a point that continuity of data series is sometimes useful, but one can presumably still get the components of interest.
Posted by: JDH at November 17, 2005 07:34 PM
Nate, I'm drawing a conceptual distinction between inflation (an erosion of the purchasing power of a one dollar bill) and a drop in real income (an erosion of the purchasing power of somebody's earnings). Although there are certainly circumstances where the two things can happen together, there are others in which they do not. My comments above pertain to the issue of measuring and trying to control inflation per se.
Posted by: JDH at November 17, 2005 07:40 PM
I'm unclear on your support for the median CPI. I know in the past you've argued that the energy (and food) components are like statistical outliers, but I think that's a misleading image. The CPI components are not random variables, but rather components of an integrated economic system that we are trying to measure the performance of. The reason you want to effectively throw energy out is because it's volatile and messes up your time series. The reason it's volatile is because demand for it is so inelastic - ie people really badly don't want to do without it, which is an excellent reason not to exclude it from estimates of inflation. Not only that, as we go past peak oil, energy is likely to grow to a larger and larger proportion of the economy, and therefore would weigh more and more heavily in the CPI. Using the median CPI, or other tricks to exclude this, look to me very much like moving the goalposts in order to hide what is happening. Let us instead just deal honestly and frankly with the fact that inflation is a volatile thing.
Posted by: Stuart Staniford at November 18, 2005 01:31 AM
what about a big push during the next year to use substitute, even a tad bit more, public transportation for gas automobiles in big metro areas?
Posted by: nate at November 18, 2005 05:42 AM
thanks for the post on real incomes vs. inflation. I did not mean to confuse things and imply that changes in real income, or real income, were the same as changes in inflation, or inflation. I do think it is possible that these things might be correlated, but I am not for sure. For example, if real wages are down, this may put less pressure on inflation.
The correlation of real income and inflation is a hypothesis. I could be wrong. I have not tested it with historic or current data.
Posted by: nate at November 18, 2005 05:45 AM
Nate, JDH, and of course others:
check this out for an excellent discussion of the difference between "inflation" and "cost of living." the link is also in Mike Bryan's post on Macroblog yesterday.
Posted by: Peter Summers at November 18, 2005 07:25 AM
While we are at it, wouldn't a trimmed mean inflation rate better than the median inflation? A separate but related issue is that currently, trimmed mean inflation is not as tame as median inflation (still tamer than the overall inflation). Dallas fed publishes a trimmed PCE inflation on their web site.
Posted by: pat at November 18, 2005 08:09 AM
Stuart, You seem to want to define "inflation" as "a rise in the average price level", which is the way the word is commonly used today, but not, I think, the most appropriate way. "Inflation" is a metaphor, and the implication is that prices go up not because of anything of real economic significance but because there is an increase in the supply of money -- like blowing up a balloon, the rubber is still the same; it just has more air in it. I would suggest that when the average price level goes up because of something real (like an oil shortage), that is not inflation, just rising prices.
To see my point, consider an economy that uses gold as currency. If there is a big gold discovery, the price of everything will go up. That is a classic example of "inflation" as the metaphor was originally intended. But suppose instead that there is a blight on the crops and the price of food goes way up. Still, the average price level will go up, because gold has become less valuable relative to everything else (food being a significant part of everything else). But this, I would argue, is not inflation. It is more like a deflation of the value of gold relative to everything else.
If we were to face a situation of persistent increases in energy prices with stable core prices, that, I suggest, would not be inflation. Nothing gets blown up like a balloon. Rather, one commodity (energy) becomes scarcer relative to everything else, including, as it happens, another commodity called money.
Posted by: knzn at November 18, 2005 10:59 AM
Knzn, that makes sense, but it raises the question, why do we, the public, care about the inflation rate? There may be separate reasons for the Fed to care, but for us in the public I think it is so we can judge better how the standard of living is changing. The figure of merit is something like average household income, adjusted for inflation. This tells us how many goods the average household can buy each year. If it is increasing, we are collectively getting wealthier; if it is decreasing, we are getting poorer.
For this purpose, I think Stuart is right and we need to treat in the same way both forms of price increases that you describe. Whether prices are rising due to an increase in the money supply or an increase in costs, either way we need to adjust people's nominal incomes in order to see whether they are able to buy more or fewer goods.
Posted by: Hal at November 18, 2005 02:18 PM
Hal's exactly right that the appropriate measure depends on the question, and the question I'm focusing on here is how the Fed should assess how it's been doing. The Fed can't produce oil, even when we need it. Their responsibility is to create money at an appropriate rate.
But I've also emphasized that if you want to form an assessment about what's coming up as opposed to summarize what's just happened, again the median CPI can prove more useful than the usual CPI. The change in the CPI as usually measured between September and October 2005 implied 2.4% inflation at an annual rate. If you'd wanted to forecast this based on what we'd seen in September, you'd have come far closer (in fact, virtually nailed it perfectly) if you used the Sept. 2004 to Sept. 2005 change in the median CPI of 2.3%, and you would have missed it dreadfully if you used the Sept. 2004 to Sept. 2005 change in the usual CPI of 4.7%. I don't want to make too much out of one month's numbers-- indeed, my basic point is that they can be pretty wild. But in this case, it does give another illustration of why, even if your interest is in the usual CPI, you might want to be paying attention to the median CPI, if you're looking forward rather than backward.
After all, the inflation-indexed bonds that I refer to are based on the usual CPI, not the median CPI.
Posted by: JDH at November 18, 2005 03:06 PM
Good boy. You looked at the imputed inflation risk from the bond market. BTW, thta 2.03 is a bit high over traditional theory. Wonder if that indicates some decreased picture of safety of the Treasury or if that is a sign that market thinks the metric itself is not quite accurate. Theory could be wrong too...just to be MECE.
Posted by: TCO at November 18, 2005 05:20 PM