Bad data II

December 13, 2011

By Matthew Goldstein

Bad data continues to confound the U.S. government in its measurement of the economy, with the Federal Reserve Bank of New York noting it too has been a victim.

In the Fed’s most recent report on outstanding consumer debt, the nation’s central bank said it recently discovered it had been underestimating the total dollar amount of student loan debt for a number of years. In the report, the NY Fed said some of the under-counting may have stemmed from the methodology used by one of its vendors.

The Fed said it has fixed the problem, which was a significant one. The Fed says it may have been under-estimating outstanding student loan debt by some $290 billion.

But don’t worry, the Fed indicates there’s nothing wrong with its other data collection. And on that front, the Fed said overall consumer debt at the end of the third-quarter declined by $60 billion to $11.66 trillion.

Let’s hope the Fed hasn’t been under-counting outstanding home loans and credit card debt either.

Here is the Fed’s student loan debt correction:

Revisions to Student Loan and Total Debt Balances
From the inception of the FRBNY Consumer Credit Panel, we have frequently compared the aggregate
balances reported on our sample of consumer credit reports to other publicly available sources of data. For most
categories of consumer debt, our aggregate figures are close to other public estimates and/or we are able to
understand the differences we observe. However, we came to realize that our aggregate reported student loan
balance was at the low end of the substantial range of publicly available sources.

After several months of discussions with our vendor, we have now come to understand the source of this difference.
The revisions to the data are fairly substantial: as of our August report, 2011Q2 student loan balances were
reported at $550 billion. We now estimate that student loans outstanding in that quarter (2011Q2) amounted to
$845 billion, $290 billion or 53.7% higher than we reported earlier.


One comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

Hello Matthew!
Thank you for providing an inline link to the Fed’s [PDF] (so I could be lazy and one-stop shop from your blog).

So. Who is “The Vendor”?

That unidentified vendor caused an underestimate of scary magnitude for the two most recent quarters. But it seems like the Fed’s estimate of student loan balances outstanding versus other sources (who later proved to be correct) started to diverge back in 2008, based on footnote 1 in the PDF. Again, I wonder why they didn’t check on that sooner. Well, at least they disclosed and corrected now, which is good.

Re the data “vendor” at fault: First I looked at the results that INCLUDED student loan data, as denoted on the lower right of each of the rather nicely produced bar charts. On the lower left corner of those charts, it said “Data sources: Fed Reserve Bank/ Equifax”. But then I realized it said that on EVERY chart, with and without student loan data! So I’m uncertain if the vendor were actually Equifax.

An aside: One would think that the Fed could obtain student loan data (for public loans, probably not for private) from other government agencies, rather than paying a corporate entity. Though I’m unsure what data was provided by Equifax, so my aside was not meant as “a big government spending- what a waste” bicker and carp, not for now.

Posted by EllieK | Report as abusive