One of my favorite things about reading a good book is rummaging through its citations to look for more good books. It’s like paleo-wikipedia. Today’s find, from Jerry Davis’s last chapter in Managed by the Markets, is this interesting article from Harper’s on manipulations of macroeconomic statistics by Kevin Phillips. Much of the piece recounts interesting and politically salient examples of how technical changes in the calculation of key macroeconomic indicators – GDP, CPI (which tries to measure inflation), and the unemployment rate – led to “polyanna creep”. That is, successive administrations pushed the statistics to always make the economy look better, such that compared to 25 years earlier, the statistics of the 2000s reported inflation and unemployment as being far lower than they otherwise would have been.
I have a lengthier take on the whole problem – which hopefully may someday be part of an article or dissertation chapter – involving the idea of technopolitics, and the problem with forgetting that statistics are always everywhere political tools and never technical measures of an objective underlying reality, but for this quick post I just want to highlight an interesting connection between measurements of inflation the mortgage crisis. So, story part I goes back to the Reagan administration and changes made in the CPI that helped lower interest rates:
In 1983, under the Reagan Administration, inflation was further finagled when the Bureau of Labor Statistics decided that housing, too, was overstating the Consumer Price Index; the BLS substituted an entirely different “Owner Equivalent Rent” measurement, based on what a homeowner might get for renting his or her house. This methodology, controversial at the time but still in place today, simply sidestepped what was happening in the real world of homeowner costs.
But, as we know from (good citation needed), rental prices and home prices diverged fairly dramatically in the past few years. That’s one of the signs we missed in the whole housing price bubble. Because the house price wasn’t in the CPI any more, the measure was lower than it would have been and thus interest rates were lower, which in turn fueled the bubble, as Phillips notes (citing a law professor):
As Robert Hardaway, a professor at the University of Denver, pointed out last September [2007], the subprime lending crisis “can be directly traced back to the [1983] BLS decision to exclude the price of housing from the CPI. . . . With the illusion of low inflation inducing lenders to offer 6 percent loans, not only has speculation run rampant on the expectations of ever-rising home prices, but home buyers by the millions have been tricked into buying homes even though they only qualified for the teaser rates.”
A (Budding) Sociologist's Commonplace Book 
Mike3550
/ May 24, 2009This reminds me of scenes in The Wire when they talk about “juking the stats” for both crime and education and how it can make mayoral administrations look good.
On a more serious note, I think that you have a valid point in saying, “the problem with forgetting that statistics are always everywhere political tools and never technical measures of an objective underlying reality.” I would caution, however, that it is very easy for this kind of argument to be turned into a statement of anti-intellectualism: if everything is political, then my interpretation is just as good as yours (e.g. “evolution is just a theory, so is creationism; therefore they’re equal“). It makes me queasy, especially as someone who uses statistics (I am a quantitative sociologist) to examine social processes. I think that it is important to understand their use, how they gain legitimacy, and whether statistics can become an implement for quant folks to beat over the heads of others to claim “truth.” But, I also think that statistics can be useful for trying to compare large quantities of data to each other in ways that other types of interpretation are not as well suited.
I don’t think that you are doing this, and I think that you are asking a different question which is (correct me if I am wrong), what political calculus is used by people/organizations/administrations that leads them to change statistics and metrics by which markets/crime/education is measured and how does that intersect with the technical knowledge of metrics to legitimize them? That, to me, is a fascinating and important question and I look forward to reading more about it here.
Erwin Rode
/ May 25, 2009There is a basic flaw in the argument that house prices should necessarily be reflected in the CPI. Here are a few points to ponder:
An owner-occupied house consits of two components: an investment and accommodation (roof over the head). The investment component is just like investing in shares, and surely we are not going to argue that share prices should be included in the CPI? But we will all agree the accommodation component should be reflected by the CPI (the cost of a roof over the head). This is why the American CPI now tracks residential rentals rather than prices to reflect the cost of accommodation. And this is why South Africa has followed suit as from January 2009.
Dan Hirschman
/ May 25, 2009@Erwin – I am not arguing that house prices should be included in the CPI, and apologies if it came across that way. Most of the changes in calculation method discussed by Phillips are defensible for some purposes. But that’s the key – for some purposes. When we stop thinking of the CPI as a particular, arbitrary (in the sense that choices were made that could have been reasonably made differently) measure and start thinking of it as the (best) measure of inflation, and when a particular measure gets enmeshed in all sorts of other processes but without the nuances of its creation, whacky and sometimes unfortunate things happen. Another way of putting it: if we take a statistical measure and its meaning for granted (i.e. CPI measures inflation), and then a change is made in how that measure is calculated (justifiably or not), the consequences can be profound. Does that make sense?
@Mike – I am definitely not arguing that all theories are equal, nor all statistics. But I think I am using a non-standard notion of politics and power. Politics, for me, is not the opposite of truth. Rather, drawing on Foucault, I see truth and power as intimately connected. So I am actually not so interested in the obvious examples of governments cooking the books except to show that people care about these things a lot – it’s like how science studies folks aren’t quite so (or solely) interested in Lysenkoism, but are very interested in the politics of everyday, ‘good’ science.
So, I definitely think large numbers can be useful – obviously, they are to governments, other organizations, scholars and etc. across the world. The question for me is, what does using numbers the way we do do?
Mortgage Man
/ July 14, 2009this post gave me reason to think, thanks…