« "Controversies about the Rise of American Inequality: A Survey" | Main | How Will You Use Your Wish? »

Sunday, May 11, 2008

Numbers Racket?

The argument in this piece is that over time, a series of "opportunistic" changes in the statistics describing the economy have led to a biased and overly rosy view of the health of our economic system. You can read it for yourself, there's a link to the article below along with a brief passage, but my take is a bit different.

There are answers to each of the claims in the article - none are new - but let me take a more general approach. The economists in the agencies that prepare our national economic statistics are dedicated individuals whose only goal is to make the statistics as accurate as possible. They are not political hacks willing to distort the picture of the economy to help the administration. Far from it. Their goal is to provide the most accurate statistics possible and there are always improvements they would like to see made. But due to factors such as data compatibility and the cost of redefining and recalculating a measurement, statistics are not generally adjusted until it is believed the improvement is large enough to justify the change (though sometimes the statistics can be adjusted backward or spliced together in a way that preserves data compatibility).

However, even when it is clear that the measurements could be improved, and the improvements are substantial, the political process may prevent definitional changes. When are we most likely to see desired changes implemented? At some points in time, the changes may work against the administration in power, in others the changes may be helpful, and it is unlikely that the administration in power would approve of a chance unless it was helpful.

This means that changes in the definitions of statistics will, looking back, generally appear to have been put into place to help the administration in power, i.e. it will appear that they were manipulated for political purposes since the changes almost always make the economy look better. But that's not the motivation - the reason for the change is to improve the measurements - and the politics surrounding such changes dictate that they will occur most often when the changes are politically favorable.

This does impart a bias, but it is not a bias in the measurements themselves, it is a bias in the types of improvements that can be made. All of the changes improve the measurements, they don't distort the picture of the economy they make it more accurate, at least that's the intent, but the changes do not generally occur unless they are politically favorable. (Let me acknowledge that there could be some bias toward a better than true picture if this is how it works. If the politics only allow definitional changes that make the economy look better, then there could be some changes that are known to be needed, but left undone because they would make the economy look worse. This would make the reported statistics biased toward a healthier looking economy, and this is why the agencies that measure the economy should be as free as possible from political influence, so they can adjust the measurements in either direction as needed.):

Numbers Racket: Why the economy is worse than we know, by Kevin Phillips, Harper's Magazine v.316, n.1896 1may2008: If Washington's harping on weapons of mass destruction was essential to buoy public support for the invasion of Iraq, the use of deceptive statistics has played its own vital role in convincing many Americans that the U.S. economy is stronger, fairer, more productive, more dominant, and richer with opportunity than it actually is.

The corruption has tainted the very measures that most shape public perception of the economy—the monthly Consumer Price Index (CPI), which serves as the chief bellwether of inflation; the quarterly Gross Domestic Product (GDP), which tracks the U.S. economy's overall growth; and the monthly unemployment figure, which for the general public is perhaps the most vivid indicator of economic health or infirmity. Not only do governments, businesses, and individuals use these yardsticks in their decision-making but minor revisions in the data can mean major changes in household circumstances—inflation measurements help determine interest rates, federal interest payments on the national debt, and cost-of-living increases for wages, pensions, and Social Security benefits. And, of course, our statistics have political consequences too. An administration is helped when it can mouth banalities about price levels being "anchored" as food and energy costs begin to soar.

The truth, though it would not exactly set Americans free, would at least open a window to wider economic and political understanding. Readers should ask themselves how much angrier the electorate might be if the media, over the past five years, had been citing 8 percent unemployment (instead of 5 percent), 5 percent inflation (instead of 2 percent), and average annual growth in the 1 percent range (instead of the 3–4 percent range). We might ponder as well who profits from a low-growth U.S. economy hidden under statistical camouflage. Might it be Washington politicos and affluent elites, anxious to mislead voters, coddle the financial markets, and tamp down expensive cost-of-living increases for wages and pensions?

Let me stipulate: the deception arose gradually, at no stage stemming from any concerted or cynical scheme. There was no grand conspiracy, just accumulating opportunisms. As we will see, the political blame for the slow, piecemeal distortion is bipartisan—both Democratic and Republican administrations had a hand in the abetting of political dishonesty, reckless debt, and a casino-like financial sector. To see how, we must revisit forty years of economic and statistical dissembling. ...

The U.S. Economy Ex-Distortion

The real numbers, to most economically minded Americans, would be a face full of cold water. Based on the criteria in place a quarter century ago, today's U.S. unemployment rate is somewhere between 9 percent and 12 percent; the inflation rate is as high as 7 or even 10 percent; economic growth since the recession of 2001 has been mediocre, despite a huge surge in the wealth and incomes of the superrich, and we are falling back into recession. ...

The question is which gives the more distorted picture, the old definitions from 25 or 30 years ago, or the newer definitions. I'm sticking with the newer definitions, though sometimes it doesn't matter much, e.g. take the unemployment rate figure. The figure above is simply the broadest definition that the BLS measures (U6 in Table A12), and this measure adds several percentage points to the standard measure (it's available every month - there's no secret here). But the difference is fairly constant over time so that while the number is higher, the relative picture doesn't change much in terms of when unemployment was lower or higher, i.e. when things were better or worse for labor market participants.

Or take core inflation, one of the things criticized in the article. As explained here many times, there are theoretical reasons for preferring the core measure, and some of the discussions criticizing the Fed's use of core inflation don't seem to realize that the "alpha-trimming" is symmetric, i.e. both large increases and large decreases in prices are excluded from the preferred inflation measures used for policy (that's not true of CPI les food and energy, but that's not what the Fed uses for policy). Nor do they realize that the Fed is not trying to measure the cost of living. (Just to say this once more, one use of a price index is to measure the cost of living, but another is to determine the course of monetary policy, and it turns out the best price index for the Fed to use for policy excludes highly volatile prices. The Fed is not trying to measure the cost of living, it is trying to find the best signal for the course of monetary policy - see here). The CPI less food and energy measures of prices reported in the news  aren't the best measures of "core inflation," and I'd prefer that the headline number be the symmetrically trimmed PCE figure, but the two measures aren't that far apart, generally, and making such a change would likely create as much confusion and suspicion (see above) as it would improvement in communication about economic conditions. As for how to measure prices, the old way which gives the higher inflation figure in the last paragraph (does the author really want the Fed to increase the federal funds rate that much?), or the new way, again, I prefer the new way.

    Posted by on Sunday, May 11, 2008 at 02:43 AM in Economics, Methodology | Permalink  TrackBack (0)  Comments (51)



    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Numbers Racket?:


    Feed You can follow this conversation by subscribing to the comment feed for this post.