For many years I have been looking at and and commenting on the importance of economic data. Every month I produce my own report on the BLS Jobs Report and share Rick Davis's GDP analysis.
There are two things I am annoyed by:
1) media accounts regarding the implications of economic data are something worse than lame. They are a disservice to any investor if this is all they act on. The problems here are two-fold: 1) media such as nightly news is structured so as to try to report something in less than 30 seconds. Trying to explain something in 30 seconds often leads to an oversimplification and false impression. I get interviewed by media every few months. I understand that the reporter is looking for a few catchy lines. I tell the reporter this: first I am going to take 5 minutes to explain this and then I will give a 30 second synopsis with, hopefully, a couple of catchy line and then you can then decide what to do.
2) apart from media reporting much economic data is , in my view, measured with too little accuracy to be useful. This is very serious. What we should be striving for is the best possible measurement limited only by the fact that economics is a social science measuring people's behaviors and opinions whcih, unlike the physical sciences, have few constants.
A perfect example is GDP. GDP is measured largely with questionnaires. If a company does not return the questionnaire BEA uses the data from the previous one supplied. They could be using GDP data for companies which no longer exist. If we did this in the mortgage business we would be arrested.
Rick Davis elucidates another issue:
"That, in turn, harkens back to the competency issue on several counts:
1) They clearly don't have any "real-time" grasp on inventories. That particular line item has the largest statistical deviation from first to second estimate for any quarter. And it is the most likely line item to change in annual revisions.
2) Their inventory numbers are subject to distortions caused by pricing changes. For many items the data they actually collect is the raw quantity of goods stored (e.g., barrels of oil).
The inventory line item, however, is the valuation of those goods. Hence, fluctuating commodity prices can creating changing "inventory levels" from one time period to the next even if the physical quantities of the goods stored does not change. BTW, the "imports" line item suffers from the same source of period-to-period distortion (although in that case somebody is actually paying the higher/lower commodity prices).
3) The whole "inventory line item" is in essence a plugged number that the BEA had to invent to reconcile the differences between production and consumption. Remember, we are talking GD-"P". And the BEA has traditionally found it easier to measure production than consumption (fewer entities to survey). But everyone wants to know the PCE numbers -- i.e., how healthy is the appetite of the American consumer. In theory, inventory change is the difference production and consumption. But somewhere along the line they actually started to survey inventories.
And then they discovered that the simple equations (consumption = production +/- inventory change) never balance. They should simply report production and consumption separately and be done with it. The current GDP "spreadsheet" is a hopeless mess of apples and oranges, all distorted by deflators (and economists) run amuck.
Not only are they poorly measuring the inventory side but they are not using the best source for the consumer side. To me it is obvious that the most accurate metric of consumer spending is the daily data from MasterCard and Visa. Yes, people still write checks and pay cash but even seasonal variations in the percentage of stuff purchased with cash or check should not vary much. Only recently has BEA started accessing MasterCard and Visa to get raw data.
BEA has the problem of being too enamored of their ability to present data going back decades. The implication. I think, is that this makes them less willing to change there methods.
The worst part here is that BEA's metrics appear to be most inaccurate just when the economy is in flux and we need accurate data the most.
Rick gives a capsule of the history of 1stQ2008 GDP here.
BEA's Changing View of First Quarter 2008 GDP
| Reported Growth Rate | Report Date | Months Lag |
|---|---|---|
| +0.6% | April 30, 2008 | 1 |
| +1.0% | June 26, 2008 | 3 |
| -0.7% | July 31, 2009 | 16 |
| -1.8% | July 29, 2011 | 40 |
| -2.7% | July 31, 2013 | 64 |
Yes it is unusual for BEA to miss by this much but this was the time when we needed accurate data the most and we didn't get it.
Since MasterCard sells their data to Wall Street investment firms for about $1.5 million a year it may be the case that they have accurate data and the public has cruddy data from BEA. Occupy protestors take note.
I assume that the Federal Reserve has its own data and makes monetary policy decisions based on that data but I think that the public, the community of economists, and the markets should have access to the best possible economic data and I don't think that we do. Some market participants have better data than others is not the way markets are supposed to work best. We should be demading that the government provide better data to all.
There are folks trying to improve this. See, for example, The Society for Economic Measurement http://www.centerforfinancialstability.org/index.php This is an international group dedicate to (in their words)
"... promote research on economic measurement, using advanced tools from economic theory, econometrics, aggregation theory, experimental economics, mathematics, and statistics. The long run objectives are to meet the data standards established for the physical sciences, subject to the inherent limitations of a social science."
Comments