Over the years, I have criticized a variety of official data points as misleading: Consumer Price Index [CPI] for woefully understating price increases, Non Farm Payrolls [NFP] due to the Birth/Death Adjustment, Core Inflation for omitting anything going up in price (aka inflation ex inflation), The Unemployment Rate due to the shrinking labor pool, and GDP due to the excesses of the deflator.
All of these data series have something in common: They all have had gradual changes in their methodologies over time. These incremental improvements in modeling, data gathering and analysis have slowly altered the various econometric models that are used to produce the official numbers: NFP, GDP, CPI, etc.
My main criticism has been that many of these changes have been for the worse. If we define any model as a mathematical attempt to portray reality, than anything that takes us away from reality is a negative. Any modeling change whose output generates further variance from the ideal construct worsens the veracity of the model.
Stated differently, do the changes improve the official data, or do they create an artificial world that does not resemble the one we live in?
Over these many years of critique and criticism, I have noticed two interesting factors:
1) The tendency for many Wall Street and academic economists to 'circle the wagons' around their chosen profession (i.e., blindly defend standard precepts). Hey, if you are going to go through the trouble of getting a doctorate in something, you probably don't want to hear how flawed many of its basic assumptions and/or methodologies are.
2) The false counter-arguments, strawmen, and phony debates. There is no faster way to admit the weakness of your argument than to claim that credible criticism is merely a tinfoil hat of conspiracy theorists. (A variation of "When the Law and Facts go against you, call the other lawyer a jerk" approach).
Good debate, on the other hand, makes your analysis better -- it sends you back to the data, forces you to look at things in different ways, and sharpens your arguments. Indeed, good arguments should help lead to their own defeat in the marketplace of ideas, as they bring about even better critiques. (See David Altig's spirited defense of the Deflator here).
Which leads us to taking one last look at the revised Q2 GDP data, showing the economy expanded at an annual rate of 3.3%.
My argument, repeated ad nauseum -- see "Previously" below -- was that the number misrepresented what was occurring in the real world economy, primarily due to the high price of imported oil oddly inflating GDP. This certainly doesn't feel like a 3.3% GDP, but rather than go on gut feel, I wondered what other periods of expansion looked like in terms of economic data. If I am going to trash the deflator, I want a more quantitative basis for doing so.
Our question: Does this 3.3% GDP resemble in most economic data points other, similar economic expansions? Or, is this GDP data, as we have argued, merely the result of a modeling flaw?
With the help of Mike Panzner (Financial Armegeddon), we looked at other periods of time when GDP was similar to the Q2 3.3% -- we used any quarter where GDP was between 3.0 - 3.5% as our range. Going back to 1959 (that's all the data available) there were 12 quarters (6.1% of the total) where GDP was greater than 3.0% and less than 3.5%. We then looked at the median Unemployment Rate, NFP (trailing 12 month change), ISM Manufacturing, CPI, PPI, Industrial Production, New Housing Starts, and Consumer Confidence.
What were the results? Consider the following: As you can see in the table below, much of the contemporary data is quite simply incongruous with an annual growth rate of 3.3% GDP.
In the past, a 3.3% GDP produced significant growth throughout the economy. Millions of job gains over the prior year versus less than 50,000; robust Industrial production versus essentially flat; Low inflation against high; Expanding ISM versus contracting; Producer prices were stable versus extremely elevated; Average home starts were 50% higher. And consumer confidence was more than double where it is today.
Comparing Current Economic Conditions With the Past: Prior Expansions of 3.0-3.5% GDP
|Economic Data Point||Q2||Latest value||Median prior 12 months|
Chained 2000 Dollars QoQ SAAR
Unemployment Rate |
Total in Labor Force SA
Nonfarm Payrolls |
Total Net Change SA From Year Ago
ISM Manufacturing |
Urban Consumers YoY NSA
Finished Goods Total YoY NSA
Industrial Production |
YoY 2002=100 SA
New Home Starts |
Privately Owned Housing Units Started Total SAAR
Consumer Confidence |
Conference Board SA 1985=100*
Q1 1959 - Q1 2008 = 197 quarters There were 12 quarters (6.1% of the total) over that span where GDP was greater than 3.0% and less than 3.5%. Note: where data is generated monthly, the value given is for the last month of the relevant quarter (*Only 10 quarters of data; series began Q1 1967).
Yes, high Oil prices contributed to the revised Q2 GDP data. No, this is not a "Goldilocks economy." The bottom line is that if this is a legitimate 3+% GDP, it is one of the worst economies ever to generate that data point since the US began recording its economic history.
Regardless of how we manage to generate a 3.3% GDP number, the table above makes it pretty clear: This is not your father's 3.3% economy.
Is GDP (via BEA) Measuring Growth or Inflation? (August 2008)
What Conspiracy? (June 2008)
Deflator Inflator ! (September 2008)
GDP Deflator versus CPI (August 2008)
GDP: Lowest Inflation Rate in 5 Years (August 2008)