On the surface, today's G.19 update, aka the monthly Consumer Credit Data, was a big disappointment due to a major miss in consumer credit, which in July dropped by $3.3 billion from $2.708 trillion to $2.705 trillion. The drop was, as always, on a slide in revolving credit, which dropped for a second consecutive month, this time by just under $5 billion, while non-revolving credit, aka student loans and GM subprime debt, rose by just $1.5 billion: the lowest monthly increase in this series since August 2011, when it declined by $9 billion. Expectations were for an increase of over $9 billion. There was a far bigger problem, however. The problem is the spike on the chart below which represents the November to December 2010 transition (source: Fed [6]). What happened there is that 3 months after the Fed revised the consumer credit data last, it decided to re-revise it again. Frankly, at this point nothing the Fed releases has any credibility, as the central planners are literally making up data every three months as it suits them.
[7]
This is how Bernanke explains [8]the epic spike in the most recently fudged data:
Because administrative data and comprehensive private data are not available to serve as a sample frame for finance companies, the Federal Reserve developed a procedure for identifying the industry universe of eligible firms within a list sample frame obtained primarily from a third-party private company and, to a lesser degree, from other sources. The CFC questionnaire (FR3033p) was mailed to each company on the list. A large-scale nonresponse follow-up study was designed and undertaken to assess the nature of nonresponse in the CFC as well as the effects of nonresponse on the universe estimate. With the information collected from the CFC and the nonresponse follow-up study, an estimate of the universe of U.S. finance companies with respect to company size and loan specialization was constructed. The subsequent SFC (FR3033s) was sent to all of the identified finance companies in the CFC, collecting detailed information on company balance sheets and receivable portfolios as of December 31, 2010. Analysis weights were constructed for companies in the SFC (after taking into account survey nonrespondents), and the finance company industry universe statistics of balance sheet and receivable portfolios (referred to as the benchmark estimates) were estimated and employed to benchmark the monthly and quarterly finance companies statistics.
Because of the benchmarking process, the G.19 data have been revised from December 2010 forward. Due to changes in the representativeness of the monthly sample since the earlier benchmark in 2005, there are considerable discrepancies in some categories between the pre-benchmarked and the benchmarked estimates. Such discrepancies are reflected in the statistical releases as series breaks in December 2010 (the benchmarking month). Therefore, the level differences between November and December 2010 due to benchmarking discrepancies do not affect the published growth rates of the corresponding series over this period. Also, analysis weights of the monthly longitudinal sample were re-estimated using the benchmarked industry universe statistics, and the monthly estimates after December 2010 were revised accordingly using the new weights.
In other words, the Fed just made any historical studies looking at consumer credit for data collection purposes impossible due to the embedded break in the continuous data which the Fed itself admits can no longer be reconciled.
And this is a "data set" on which the Federal Reserve of America is basing monetary policy that determines the cost of money in the entire world! Certainly, all this will have a Hollywood ending.
For those curious how this data set looks like post the "estimated" break, here it is . It is largely meaningless.
[9]
Broken down by sources of credit, we find that the US government was for the first time in ages, not the primary source of cash, adding just $1.1 billion in NSA debt.
[10]
What is most problematic is that while the SA data was adjusted as per the viagra spike above, the NSA data still have smooth continuity in the Nov-Dec 2010 data. In other words it is now completely made up by some Fed intern.
To everyone making budget decisions based on how the US consumer is treating credit, good luck. You are on your own.
This latest print was the largest 2-month contraction since March 2009...
[11]
and the biggest miss in a year...
[12]
Charts: Bloomberg