Too Good to Question

The Federal Reserver Bank of New York posted about their study that confirms many people's biases about moral hazard and large financial institutions. The question is "Do “Too-Big-to-Fail” Banks Take On More Risk?" and the answer is yes.  The basic idea is that higher government support leads to riskier loan portfolios, which indicates to many people that Too Big To Fail (TBTF) banks were abusing their positions by loading up on risk.

I'm sure TBTF banks have taken on more risk - I believe moral hazard exists in the financial system. But I am not sure this study should give anyone more confidence on this issue.

After controlling for many variables, the study found that on average eight months after an increase in the perceived government support as measured by Fitch's "Support Rating Floor" the bank would have more impaired loans around eight months later (and vice versa). 

This is using data from March 2007 through August 2013, so the time period covered both the financial crisis and the european sovereign debt crisis. Given that, which explanation is more likely?

1. The average bank goes out and makes riskier loans after getting government support.

2. A negative economic shock created a scenario where government indicated support rating floors are needed. Banks who more obviously needed help got it first. Because problems in banks balance sheets show up slowly, it took a while for the banks that got support to admit that more of their loans were impaired. Support goes away when it isn't needed and slowly the loans are found to be performing better.

3. Only after a bank is assured that it is getting more government support (this happens only after the support has been promised for some time) do the banks feel comfortable marking down part of their loan portfolio. 

4. Banks that take over ailing financial institutions become TBTF and get boosts in support levels. After taking over troubled institutions, they find that many of those loans end up impaired.

The analysis controlled for quarter year fixed effects among other things, so the simplistic "Oh they were just pricing in the timeline of the crisis" argument doesn't quite work.  But even so points 2, 3 and 4 seem far more likely than the first scenario. In their paper the NY Fed researchers claim that because tier 1 capital ratios didn't decrease then their interpretation of moral hazard is more likely to be correct, but this doesn't account for the capital raising that occurred during the crisis.

Thinking of it from another perspective, it's likely that the age of the impaired assets are greater than eight months - the banks didn't rush out to make or buy bad loans just because they got some more perceived support. The relationship between changes in the support floor an subsequent changes in the bank's portfolio are both related to the bank being in trouble and this isn't adequately captured by the other variables. It is far from certain that the story played out as neatly as people would like it to play out.

There is moral hazard and many banks have abused their positions a TBTF, but studies that confirm everyone's biases should be examined even more closely than usual.