Friday, September 6, 2013

Data, Indexes and GIGO

Here is comment I put up at Asymptosis that stands alone and is worth consideration here.


Inflation can't be measured precisely since there is no observable price level. The price level is an index and an index is an arbitrary figure that could be arrived at through different paths and rationales.

What is important is the rate of change of a constructed price level, so if the index is figured the same way each period, then a rate of change for that index can be computed and used to measure actual changes in purchasing power in contrast to the apparent rate that volatility of some goods suggests. While the price level constructed is an arbitrary figure, it's action is not as long as measurement of the variables over time is reliably accurate.

However, data collection in the case of economic data is not the result of direct observation as in the natural sciences either. Economics is a social science and the data is much looser. Some important economic data is anecdotal rather than observational.

Inflation is a bogus measure when applied beyond the limits of the data, which is most of the time historically and even today in countries without adequate institutional arrangements for data collection and processing. Inflation rates extending back centuries in historical studies are usually presumed to be true. On what basis?

The US is a leader in the field of economic data, and to suggest that its agencies get the data wrong or misconstruct it (Sumner), or manipulate the data for political purposes (Jack Welch) is to suggest that the data upon which macro analysis is founded is garbage. Ergo, macro analysis is GIGO. The question is, How true is that? It's a question I have been pondering for some time. It seems to me that a lot a macro analysis may be GIGO, such as Reinhart & Rogoff turned out to be on critical analysis.

Then there's Robert Eisner's work on national accounting, which also suggests that data is misconstructed and misinterpreted based on the institutional construction and interpretation of national accounting, e.g, in comparison with firm accounting. As a result the reported fiscal balance may not represent the actual fiscal stance, and so politicians are misguided in relying on it to formulate economic policy.

Is there a pernicious tendency to take reported figures for constructs like price index and fiscal balance as exact when they are only estimates or best guesses? Enquiring minds would like to know.

Same goes for medical studies for medications and procedures, as anyone who has had the occasion to question one's physician on specific recommendation and knows what questions to ask comes to realize. Are consumers generally too trusting of physicians recommendations? My conclusion from experience is yes.

There's a lot of flying by the seat of the pants that gets swept under the rug of conscious awareness in the presumption of a degree of exactitude that is non-existent.


Brian Romanchuk said...

I am not sure that the problem is necessarily inflation indices, rather how they are used. Over time, a price index like CPI should give useful information.

I think the underlying problem is the assumption that we can treat all goods in the economy as a single good, using the CPI as the price of that "good". The mainstream approach of having a single (consumer) good in their models gives rise to some serious analytical biases.

One classic problem we have seen - how to treat house prices? By ignoring them in inflation targeting, bubbles are allowed to blow up.

I probably will do a longer explanation on my blog within a few days.

Tom Hickey said...

Great. Post a link here when you do.

Brian Romanchuk said...
This comment has been removed by the author.
Brian Romanchuk said...

(Fixed the link html.)

I've written my comment (on

I cover some of the analytical blind spots created by a single-good economy. In addition to problems with aggregates like the CPI, there are big problems in understanding investment bubbles. If there is only one good, it is impossible to allocate investment incorrectly.

article link

Tom Hickey said...

Thanks, Brian. Promoted to a post.