Government Statistics Are A Really Poor Basis For Policy, Because They’re Really Poor Statistics.
“The correct answer, I am afraid, is that we have virtually no economic use for national accounts, partly because we cannot be in control of our economy and partly because our economy has a dynamism which outpaces such accounts.” Those were the words of Sir John James Cowperthwaite, Financial Secretary of Hong Kong from 1961 to 1971.
Cowperthwaite famously harbored a dislike for government statistics given an unrelenting aversion to government intervention in the economy. The production of economic statistics would perhaps cause politicians to “do something” in response to measures indicating economic trouble, so Cowperthwaite avoided them altogether. Considering how desperately poor Hong Kong once was, it’s not unreasonable to suggest that Cowperthwaite was on to something. An economy is not a blob, or an organism that can be massaged or nursed, rather it’s just people. People have needs and they produce in order to fulfill them by exchanging their production with others. Where people are broadly free economically, their production has a tendency to be substantial.
All of this came to mind while reading a recent column by the Washington Post’s Catherine Rampell. Smart and witty as she plainly is, there’s an odd reverence for experts, and in particular experts in government, that informs much of what Rampell writes. Because they’re at the Fed, central bankers are wise in eyes of Rampell despite their track record of being routinely incorrect about the economy. This isn’t a knock on Fed officials as much as it’s a statement of the obvious. Anyone with a really good feel for the economy’s direction wouldn’t toil at the Fed for well below 1 percenter wages. They’d earn millions and billions as investors, and would do this despite still being wrong much of the time. As top macro traders would surely relay to Rampell, they’re incorrect nearly as often as they’re correct. Often stupendously so.
Rampell’s reverence for government and its experts extends to the statistics they produce. As her April 16th column emphatically stated, “[O]ne basis of a democracy – not to mention a healthy economy – is good official statistics so that people can make informed decisions.” Except that what Rampell presumes about statistics isn’t true. And the previous statement has nothing to do with conspiracy theories about government officials having it in for Republicans, or affection for Democrats. Instead, it has everything to do with Rampell seemingly being unaware of what she doesn’t know.
To see why, consider Gross Domestic Product, a production of the Department of Commerce’s Bureau of Economic Analysis. Seen by some as worthy, it’s this very statistic that has a president whom she despises ranting about so-called “deficits” in trade. But we have these “deficits” precisely because the U.S. is a magnet for both investment and imports. The world’s greatest investors are more than eager to purchase American equities, debt and land, but “export” of those certificates of ownership doesn’t count in trade statistics, while our import of shoes, socks and t-shirts does. This accounting abstraction is the source of “trade deficits” that bring down GDP, and this same abstraction that is a happy sign of prosperity has our 45th president threatening trade wars. Yet Rampell wonders why some on the right look askance at government statistics?
Sometimes the dislike of government statistics is rooted in a simple aversion to waste. Consider the monthly unemployment report produced by the Bureau of Labor Statistics (BLS). Notable here is that the BLS employs 2,500 people at an annual cost of $640 million to produce its report. Meanwhile ADP, an S&P 500 payroll company, produces a monthly jobless analysis that nearly mimics that of the BLS, all at no cost to the taxpayer. Better yet, the ADP measure is released one or two days ahead of the BLS’s. And since Rampell is of the view that our economic health is at least somewhat an effect of quality information, does she really think the BLS’s much smaller sample size data is more thorough than ADP’s? Think about this for a second in consideration of how vast is ADP’s data; it once again being the most prominent payroll company in the U.S.
And if ADP doesn’t convince Rampell that private industry could and does more accurately and inexpensively produce economic statistics, consider the monthly retail sales report produced by the U.S. Bureau of the Census. While consumption is always and everywhere an effect of production that has taken place first, the economics profession broadly puts the cart before the horse. It focuses on consumption as an economic indicator, even though our wants are endless and can only be fulfilled through production first. But that’s a digression. Let’s consider the retail report.
It’s a creation of 5,000 surveys sent out to businesses each month, of which generally only 2,500 come back to the federal workers compiling the numbers. Comical here is that so dated is the survey that more than a few of them reach the Census Bureau by fax machine! Rampell might contrast this blast to the past with the monthly retail report produced by Visa. Figure that the credit card company is operating in the 21st century, and better yet its consumer data springs from 400 million accounts. Rampell rails against rejection of government data as the stuff of mouthbreathers, but isn’t her protest a wee bit overdone? For her to pretend that the Census Bureau’s retail stats are anywhere close to as accurate as those of Visa would bring new meaning to willful ignorance.
So while Rampell surely has a point when she critiques President Trump’s aversion to facts and numbers, along with those in his employ, she might take a step back herself. And examine her own biases. Smart as she is, there’s much she doesn’t know. Or hasn’t considered. The low quality of government statistics that she oddly venerates reveals the latter in spades.
John Tamny is editor of RealClearMarkets, and Director of the Center for Economic Freedom at FreedomWorks.