Monday, November 8, 2010
Below is the unedited version of my column for this week. You can read the final version at the Daily News website, but since I have been editing my columns myself on the Daily News media webeditor since March, you won't see much of a difference between the two. The title is inspired from one of the most famous B movies of all time.
Coming to more serious matters, as I promised when I posted the column to Facebook, I will talk a bit about my econometrics methodology: First, I like VARs for this kind of exercise because a priori, I don't have any idea on what the relationship between core inflation, food inflation, inflation expectations and the exchange rate is. In such cases where you don't have a structural model, VARs are extremely useful. It is more like gumbo soup; put everything in, and then the model will make its magic and spit out the results. Actually, it is a just a tiny bit more complicated than that: You also need to specify the lags, but there are straightforward methodologies for doing that.
Anyway, to expand on my results: The effect of a one standard deviation shock to food prices results in a maximum impact of 0.4-0.5% on core inflation, with the effect maxing out in 4-5 months. The effect somewhat lessens to 0.2% in the next 2-3 months and stabilizes thereafter. As for the transmission from food prices to inflation expectations: A 1% mom rise (I converted standard deviation for this one, as I was also doing recursive regressions and wanted to compare my results- I got similar results at the end) in seasonally adjusted food prices increases 12-month ahead inflation expectations the next month by 0.09%. The effect is nearly twice as much for end-year expectations and negligible for 24-month ahead expectations. Finally, the relationship between inflation expectations and core inflation comes out to be almost a perfect quadratic: Again using a VAR, a standard deviation shock to inflation expectations causes a 0.4-0.5% increase on core inflation, with the effect maxing out in 10-12 months.
On a final note, CBT released the latest inflation expectations this week:
As you can see, 12-month ahead expectations have creeped up, whereas 24-month ahead expectations actually fell. More or less what I was describing above. No wonder they say "a picture is better than a thousand words".
And now that the promised addendum is over, on to the column:
Last week’s October inflation, at 1.8 percent monthly, came in at much higher than expected.
Tomatoes stole the show, as their prices increased a whopping 112.2 percent over the previous month. Unsurprisingly, tomato sauce prices rose 19.8 percent. Your friendly neighborhood economist, whose duties during his seasonal moonlighting in the family business include doing the purchases, was disappointed that he couldn’t get even with the grocer - they had to lock up just as tomato prices were finally coming down at the end of the month.
While tomato’s rocket launch was supposedly due to infection of tomato plants by the tomato moth, food prices were solely behind the higher-than-expected October inflation turnout. Food inflation was 4.5 percent, contributing 1.2 percent to the headline figure. In fact, 13 out the 20 items with the highest monthly price increases were food products.
Unfortunately, high food inflation is not a recent phenomenon. Food prices have risen 13 percent in the past 3 months, and yearly food inflation is now running at 17 percent, almost twice the headline figure.
As a result, the headline figure and measures of core inflation, which exclude items responsive to temporary shocks but not to monetary policy (and are therefore key price metrics for central banks), have been diverging for a while. For example, the Central Bank of Turkey’s, or CBT’s, preferred measure of core inflation, which excludes alcohol & tobacco, energy and gold in addition to food, is at an all-time low of 2.5 percent.
Then, the natural question to ask is whether headline inflation will eventually revert to core inflation. A recent CBT working paper argues that will be the case, but a casual look at the data does not reveal such a relationship since the new price indices began in 2003.
On the contrary, just the opposite might happen: Food or other non-core prices such as energy could be feeding in to core inflation in two ways. First, they could directly lead to increases in core prices to the extent that they are used as inputs in other goods. Second, they could lead to higher inflation by causing an increase in inflation expectations.
A simple econometric exercise reveals that the first effect, while slow to build up, is nevertheless there. Food inflation seems to pass through into core inflation fully in four to five months, and while the impact is not that large, it does not dissipate for up to a year.
On the other hand, food prices feed in to inflation expectations rather quickly. The effect is fully there after a month, reflecting the adaptive nature of inflation expectations in Turkey. As for the transmission from inflation expectations to core inflation, the overall impact is similar to the first effect, although its build-up is slower but more persistent.
One might argue that the lira’s recent appreciation might tame expectations. But the effect of the currency comes out as asymmetric in my analysis: While depreciations push expectations higher, appreciations don’t pull them down.
One might also argue that food inflation is a global phenomenon, but comparing Turkey and the EU reveals that one-half to two-thirds of Turkish food inflation is due to domestic factors. Turkish food inflation is also much more volatile than the EU’s and has recently been driven mainly by fruit, vegetables and meat. These results suggest that a global correction in food prices may not spill over to Turkey by much.
I am sure the CBT has gone through all this analysis, but downplaying headline inflation, while at the same time focusing more and more on core inflation, is the only way it can create a breather as the economy heads towards a challenging inflationary outlook.