Thursday, October 7, 2010

Not Quite Deflation Yet

A few months ago, Paul Krugman did something interesting.  I can't find the link right now, but basically he noted that in considering rapid changes in inflation, the normal year-over-year percentage changes can be slow to recognize turning points.  He therefore looked at month-over-month changes, and then used a trend line to overcome the noisiness of the resulting data.

I wanted to update what that data looked like.  So I took the BLS CPI-U ex food and energy (to avoid the most volatile components, with a view to seeing more clearly the underlying trends in money-vs-goods-and-services).  I used the seasonally adjusted index, and computed monthly changes (annualized).  The result, together with a six month moving average to cut the worst of the noise, is here:

That graph shows the long view (from 1960 on), and you can see the long steady decline in inflation since the era of loose-money-and-oil shocks in the late 1960s and 1970s.  Just looking at the data since the beginning of 2008 (ie pretty much the beginning of the great recession):

it's clear that the overall trend has been very steeply downward.  However, there was a little relief over the summer of this year (the data go through July).  Not quite deflation yet.


Kweksma said...

Does it realy matter if we are at deflation now, or if it takes anotrher quarter to get there. The economic situation is :

+ banks get sh*tloads of money for free,
+ they sit on it, improving their balance sheets
+ consumers pay of debt and stop using their credit cards (if they can still get one)
+ invetory is way up, so businessspendig will drop next quarter
+ Oil is racing up to 90$/b eating into discretionary spending (cause of the uptick ?)
= there is no money floating around to buy goods, so shops will lower prices to get people to buy

I can't see the deflation NOT going to arrive

Mike Aucott said...

Doesn't excluding food and energy constrain the analysis considerably? They may be volatile, but also may be subject to long-term and significant trends despite that volatility.

Stuart Staniford said...


Fair question. My personal view is that it very much depends on what one is trying to do. If one is trying to compare prices/incomes/etc over long periods, I think absolutely they need to be included as the measurement should be across all goods and services. However, I'm persuaded by the economist's arguments that for the specific question of whether the economy is heading to deflation or not, the volatile components are not helpful in determining that, since they tend to bounce around and obscure the underlying monetary phenomena.