Last December, the Board of Governors of the Federal Reserve published a brief article describing projections of bank assets under the scenarios used for stress-testing. The projections suggested that, rather than falling, bank assets tend to rise in recessions. As a consequence, the Fed suggested banks were being too pessimistic about potential lending volumes and asset size under the severely adverse scenario used in the Comprehensive Capital Analysis and Review and related stress-testing exercises. This conclusion is critical because stressed default rates applied to a larger asset base implies much greater losses and the need for far higher capital reserves by banks.

The Fed's projections have been derided by bankers and many others who are closely following the CCAR process. Unfortunately, the sources listed for the numbers quoted by the Fed are apparently unpublished and the procedure the Fed used to reach this controversial conclusion goes undescribed in the brief paper. Nonetheless, we can draw on public data sources to try to better understand the Fed's thinking.

Using different banking assets data  on large domestically-chartered commercial banks from the Fed board, we find very similar results for all three recessions analyzed by the Fed's researchers.  By our analysis, the nine-quarter asset growth rates are slightly lower for the 1990 and 2001 recessions and slightly higher (11.8% versus 9.9%) for the key 2008/09 downturn, which is assumed to begin in December 2007.  

In terms of loan growth, the same data source reveals a slightly more pessimistic view of the 1990 and 2008/09 periods (1.1% versus 4.3% growth) and a much slower rate of loan growth in the 2001 recession (8.8% against the 18.7% rate found by the Fed).  Though the numbers we obtain are slightly distinct from the Fed's, the picture painted is qualitatively the same. Based on these numbers, bank lending and assets clearly rise in a crisis.

As part of the description of its methodology, the Fed makes an assumption of "normal" bank lending supply during the recessionary period. Presumably, these supply adjustments account for much of the discrepancy seen in the data. One can see why the Fed wants to make this assumption, since its main task in a recession is to ensure that liquidity is maintained and credit keeps flowing through the economy's veins and arteries.

From an analytical perspective, however, isolating separate supply and demand channels is very difficult when one observes only volume and price outcomes. If loan growth falls, is this because banks stop offering loans or because businesses and households stop wanting them? Typically, there are few, if any, reliable and pure measures of lending "supply." As a result, one wonders exactly what the Fed is holding constant in its projections of asset and lending growth under stress.

The exact timing of the assumed start of the recession is also critical for this analysis —the Fed uses National Bureau of Economic Research dating, though this is in no way conclusive in terms of defining the beginning of a period of true banking "stress."

In the first half of 2008, long before the collapse of Lehman Brothers sent the economy into free fall, GDP was fairly stable and bank lending and assets were rising solidly. The economy during much of this period bore no resemblance to the immediate chaos implied by the severely adverse Fed scenario. After Lehman, banks were able to squeeze out one further quarter of loan growth before the inevitable decline occurred.

Starting our nine-quarter period a year later than the Fed, we find solid declines in bank lending (down 6.6%) and a smaller reduction in total bank assets (4.5%). These numbers are much more in line with the projections of the individual banks. Bank lending and assets decidedly fall as a result of recessions though, admittedly, the data tend to lag the trajectory of the broader economy by about a year.

One question that needs to be addressed in this analysis is why bank lending rose during 2008. Breaking down loan growth for that year, we can attribute about 46% to business lending (commercial real estate and commercial and industrial) while consumer loans such as autos and credit cards were responsible for 15% of growth.

In residential mortgages, the epicenter of the unfolding crisis, closed-end mortgages actually fell while home equity lines of credit — only 8% of all bank lending at the start of the period — contributed an outsize 18% of total bank loan growth. This disparate behavior of mortgages is revealing. Although banks can control originations and the size of unused credit lines, they have no means of controlling the utilization of existing credit lines. One suspects that much of the rise in lending in 2008 was driven by businesses and households maxing out credit lines in anticipation of a looming credit drought rather than egregious new lending by rapacious banks.

We generally see this type of pattern with revolving credit at the start of recessions. Clients are slow to react to changed circumstances while banks quickly lock down or reduce available credit lines. Utilization rates, therefore, rise rapidly. In the latter days of the recession, businesses and households tightened their belts, leading to deleveraging and an ultimate normalization of line utilization. It is this pattern that renders bank activity as a lagging indicator of the overall economy.

So is the Fed right to say that loan and asset growth should be positive during the nine-quarter CCAR severely adverse scenario? By our reckoning, the trajectory of the Fed's scenario most closely follows data from the Great Recession assuming a start date of June 2008. Using this definition we observe modest loan growth (2.3%) and robust asset growth (9.2%). The fact that all the growth comes in the first two quarters is in some ways irrelevant — over the full period, loans and assets outstanding do tend to rise modestly.

The derision of bankers is, therefore, somewhat unfair to Fed researchers, who are themselves being rather disingenuous in not telling the complete story. Stress causes bank assets to decline, though this happens with a considerable time lag.

Tony Hughes is managing director of credit analytics at Moody's Analytics, where he manages the company's credit analysis consulting projects for global lending institutions.