Misuse of Data Part Deux: Redux

Fixing the Economists Article of the Week

by Philip Pilkington

Okay, so there’s been a bit of a misunderstanding regarding my previous post on Catalán’s misuse of data. Some of this is likely squirming but some may be due to my own presentation. So, let’s run through this in order to be crystal clear.

Catalán’s original argument was that the PPI index measuring the price of capital equipment could be used to show that said prices of capital equipment are flexible enough to respond to flows of profits. This, according to Catalán, would then show that fixed-price theory when applied to capital goods was “highly debatable” — his words.

My point was that the measures that Catalán had used were really just proxies for inflation. I went on to say that if he had deflated his measures by the CPI then he would have seen far smaller changes in the price of capital equipment (whether this has anything to do with fix versus flexi-price we will consider shortly). Catalán and his friend then said that it was stupid deflating one price index by another price index. But is it?

Consider an economy in which there has been a very poor harvest of apples and said apples have risen in price by 45%. Now consider that this same economy has a general increase in prices of 10% due to as yet unknown reasons. Do we then attribute the whole 45% of the rise in prices of apples to the poor harvest? Of course not. Some of it must have been due to the general rise in prices. Provided that we assume that apples only make up a very small part of the general price index a good proxy measure for how much of the price rise in apples was due to the poor harvest can be gotten by deflating the apple price by the general price index.

Again, what Catalán was seeing in his graph was largely the general drift of inflation; not some particular price increase for capital equipment. This is where the big question comes in: does the existence of inflation disprove the idea of fix-price markets for capital equipment? Well, sure if you understand that fix-price markets mean “markets where prices don’t fluctuate and inflation doesn’t exist”. We will give Catalán the benefit of the doubt in that he didn’t make this amazingly stupid mistake.

No, a graph that basically shows the effects of inflation says nothing about whether markets are fix-price or flexi-price.  Nothing.  Nada.  Zip. And that is why it was glaringly obvious to me that Catalán had misused the data from the moment I cast eyes on his graphs. I thought this would be straightened out fairly quickly by simply pointing out that his graphs were proxy measures for general inflation and had nothing to do with whether or not there existed fixed or flexi-price markets for capital goods. But alas, mistakes made die hard.

Let’s go one further:  even if Catalán had deflated his graphs would it have told him anything about whether the markets were fix-price? Of course not. Unless we interpreted the far smaller changes in the price of capital equipment that would result from said deflation as proof that fix-price markets don’t exist. But again this would entail interpreting fix-price markets to mean “markets in which prices don’t fluctuate”.  Again, we will give Catalán the benefit of the doubt in that this is not what he meant.

Update: I just noticed a comment on the previous post from Catalán’s friend that might be worth responding to in this post directly. Dkuehn wrote:

[Your argument regarding deflating the price of capital equipment by the CPI is] like deflating the Dow by the NYSE and concluding that financial markets are largely placid.

Is this really a fit analogy? I have already laid out a theoretical argument (the apple argument above) why this is not necessarily the case. But one would wonder if dkuehn would make the same case if, say, market analysts used a measure for inflation-adjusted oil prices or any number of other inflation-adjusted measures. After all, these are just price indices being deflated by other price indices (which probably contain the original price indices as a component). Remember, oil price measures are themselves price indices and according to dkuehn:

Thinking about real and nominal differences in price level changes is interesting, but you don’t get there by adjusting one price index by another price index!

He then goes on to talking about how we should instead adjust a single price by the price index to get an “interesting” result. But what is a “single price”? Is he talking about going to the supermarket and looking at the price on a single television and then deflating this?  After all, the manner in which television prices are compiled in the CPI is by aggregating many prices and then weighting them; that is, compiling them into an index and then putting this into the broader index. It is not clear what “interesting” results such ultra-micro level price deflating would produce — but then I suspect that dkuehn got confused as to what the words “price index” mean.

The deeper you dig into bad data usage the more profound the misunderstandings revealed become.