Posts filed under “Data Analysis”

Inflation Data: Accuracy versus Precision (as if either matters)

One of the more interesting issues of any modelling or statistical work is whether they have the attributes of Accuracy or Precision.

Accuracy is defined as the "degree of conformity of a measured/calculated quantity to its actual value." I often refer to this as how closely a measure represents  reality.

Precision is defined as "the degree to which further measurements or calculations will show the same or similar results." A precise measure will be reproducible or repeatable.

Ideally, any economic measurement should aim for both qualities: they will be
accurate in that they reflect what reality looks like, and they will be precise, with measurements close to and tightly
clustered around "a known value."

Which brings us to the Bureau of Labor Statistics: they are contemplating a change in the consumer-price index (July data released tomorrow) which may have a big impact on how the latest inflation data gets interpreted. BLS is  considering publishing the index (and subindexes) "to three decimal places instead of one . . . [to] greatly reduce the frequency with which rounding produces a misleading inflation rate."

Let’s take a moment to consider the humor of this statement:

On the one hand, it is an admirable goal on the part of the econ wonks who do the heavy lifting and produce the initial data for CPI and PPI. I have spoken with many of the economists and statisticians at BLS, Commerce Department, and DoL. I have yet to meet anyone there who wasn’t intelligent,  hardworking  and diligent in the pursuit of their craft.

The problem comes about after they produce their numbers. It is not the precision or accuracy of the data; rather, it is the bias of the model and the subsequent torture of that data where the inaccuracies and imprecisions are generated.

Accuracy and precision have become almost
irrelevant, as it ultimately gets fed into economic machinary that
is so biased towards showing a minimal level of inflation — regardless of what is happening in the real world.

Example: Yesterday’s WSJ had a front page article on commodity prices:

"Shortages and high prices for raw materials are fueling a new and
unusual wave of acquisitions and deals. Steelmakers are buying iron-ore
mines, airplane manufacturers are striking long-term deals for
titanium, and the world’s second-largest tire maker is cultivating
rubber trees.

This return to a type of vertical integration that has been out of
favor for decades signals a new phase of industry consolidation. Having
bulked up by acquiring rivals, manufacturers are turning their
deal-making prowess to raw materials providers in hopes of ensuring
adequate supplies and controlling costs."

And yet, inflation is said to remain tame "in the core rate." An entire wave of M&A and vertical integration in several industries is underway due to soaring commodity prices, and yet, we continue to hear from officials how contained inflation is. Once you consider the absurdity of that, you will realize why 3 decimal accuracy is a noble effort in a wasted cause.

Precise data fed in, inflation ex-inflation out; it is a classic example of torturing the data long enough to get it to confess to whatever crime you want.

For example, the BLS surveys prices on a monthly basis about 80,000 items in 200 categories. The results of that survey then gets converted to series of index numbers. That’s how the  CPI and PPI index and core get created.

The WSJ noted that:

"Because of rounding, these inflation figures can be misleading — in two ways. One is that the percentage change is rounded. For example, an increase of 0.249% would be rounded down to 0.2%, while an increase of 0.251% would be rounded up to 0.3%. The difference between 0.2% and 0.3% seems large, but without rounding the difference is trivial. Economists, however, can adjust for that problem by calculating the percentage changes in the indexes themselves.

The second, more serious way the figures can be misleading results from the fact the BLS rounds the indexes as well before publishing them. Suppose the index for one month is 198.945, and then rounded down to 198.9, and the index for the next month is 199.355, and then rounded up to 199.4. The change in the rounded numbers is 0.251%, which rounds up to 0.3%, but the change in the unrounded numbers is only 0.206%, which rounds down to 0.2%.

The difference between 0.2% and 0.3% can have a huge impact on the market."

Accuracy_and_precision_1This precision difference is analogous to sterilizing the scalpel for an incompetant surgeon. No matter how clean his surgical instruments are, if he kills the patient, it is irrelevant. As long as the pateint doesn’t die from an infection caused by dirty scalpels, the person who preps the tools has done their job.

That is similar to the situation the BLS statistical analysts find themselves in. They want to make sure that the inflation statistics — nonsensical though they might be — are neither imprecise nor inaccurate due to anything they did.

The net of this is that, assuming this change gets incorporated in the BLS data surveying methodology, we will have the luxury of more precise measurements to be ignored by policy makers.

Inflation? What Inflation?

After all, if the focus is remains obsessed on the core rate of inflation, if the Fed and other policy makers ignore long term underlying inflationary trends — if they only consider those items not going up in pricehow much does it really matter if the inflation data is accurate to one two or three decimal places?


The problem lies not with the dirty scalpel, but the incompetent surgeons . . .




Key Inflation Index May Get Greater Precision
WSJ, August 14, 2006; Page A2

Accuracy and precision

A Hot Commodities Market Spurs Buying Spree by Manufacturers
WSJ, August 14, 2006; Page A1

Category: Commodities, Data Analysis, Federal Reserve, Inflation

The S-Word

Category: Data Analysis, Economy, Inflation

New Column up at Real Money (08/10/06)

Category: Data Analysis, Federal Reserve, Inflation, Investing, Markets, Psychology

To the Pink Sheets! Tradesports’ Bad Call

Category: Data Analysis, Markets, Psychology

The Sensitive Sectors

Category: Apprenticed Investor, Data Analysis, Economy, Media, Technical Analysis

Where are the bodies? In line at Unemployment

Category: Data Analysis, Economy, Employment

The July Effect

Category: Data Analysis, Markets, Technical Analysis, Trading

What Does Poor IPO Pricing Indicate About Sentiment?

Category: Data Analysis, Markets, Psychology

There They Go Again: NRF Redux

Category: Consumer Spending, Data Analysis, Economy, Retail

1. Earnings 2. Reaction 3. Guidance

Category: Corporate Management, Data Analysis, Earnings, Economy, Investing, Markets