Abstract: This study evaluates the performance of feed-forward neural networks to model and forecast recovery rates of defaulted bank loans. In order to guarantee that the predictions are mapped into the unit interval, the neural networks are implemented with a logistic activation function in the output neuron. The statistical relevance of explanatory variables is assessed using the bootstrap technique. The results indicate that the variables which the neural network models use to derive their output coincide to a great extent with those that are significant in parametric regression models. Out-of-sample estimates of prediction errors suggest that neural networks may have better predictive ability than parametric regression models, provided the number of observations is sufficiently large.

This seems like a straightforward application of neural network to predict LGD. As with any usage of neural network models, It'll be more difficult to get business buy-in. If nothing else, it could help suggest some risk factors overlooked by veteran modelers, who might have tendency to pick the old work horses of factors that worked well in the past.

Financial Stability Board Chairman Mario Draghi renewed calls for capital rules for the largest lenders that go beyond the Basel III proposals earlier this month from global regulators.

Draghi said in Paris today that systemically important financial institutions should have the ability to absorb more losses than smaller banks. Draghi, who is also governor of the Bank of Italy, made similar comments in newspapers in the U.K. and his home country earlier this month.

...Adair Turner, the chairman of the U.K. financial regulator, last week also said he would have set higher capital requirements for systemically important banks.

The regulators and central bankers at the FSB are studying what types of extra capital beyond the Basel III requirements could be sought for the largest lenders, including capital surcharges, contingent capital such as convertible bonds, or “bailed-in assets.”

“The FSB and the Basel committee have been very clear about this issue -- there will be some sort of additional capital requirement coming for the biggest banks soon,” said Barbara Matthews, managing director of BCM International Regulatory Analytics LLC in Washington. “Big banks should be worried, and are already worried, because the combination of all these new rules will have a great impact on their business models and profitability.”

Given the issuer-pay model, there is an ever-present concern that rating agencies have an incentive to understate the risks and provide ratings that are too high. In other words, a conflict of interests exists.

Broadly we are faced with two choices to make the system more functional. The first option is to use partial measures that deal only with the symptoms; this option essentially shields the incumbent ratings agencies from competition.

The second choice is to take bolder action--we abolish the "Nationally Recognized Statistical Rating Organization" status that credit ratings firms currently have (as provided by the U.S. Securities and Exchange Commission) and open up the way for competition that encourages the growth of new and diverse types of credit assessment firms that cater to investor needs with objective and dynamic measures of risk.

Rule #1: Use The Small Edge

Says Rafe: "Let's say there are 10,000 decisions that you make in the course of winning a poker tournament, each with different odds of a positive outcome. The key is taking very small edges (55/45 here, 60/40 there) over and over and over again. If you keep waiting for the perfect hand, you'll lose all your money in antes. Most people’s perceptions of how much of an edge they need to make a ‘correct’ play is totally off."

The daily decisions we make follow a similar pattern. We can either wait for the odds to be obviously in our favor before making a move or we can take hundreds of smaller chances where we have a small edge. So ask yourself: Are the odds of this particular job, conference, or opportunity going well better than flipping a coin? Yes? Then do it again and again and again.

Rule #2: Decisions are Different From Outcomes

We've been raised to think that good decisions always lead to good outcomes. Problem is, they don’t. If you follow the small edge rule, any decision you make that has a better than 50 percent chance is a good decision. You both win and lose, but over a long period of time, you eventually make it big.

In 2009 the theory was that, as heads rolled in the post-Lehman landscape throughout the investment banking industry, risk was a relative recruitment sanctuary as firms looked to bolster this function to appease rising public and political anger.

In reality, hiring in risk was also almost non-existent at the height of the crisis, suggest recruiters, and it's not excluded from the slow-down now.

...

This is not to say that hiring in risk is dead entirely. Recruiters point to numerous roles within Bank of America Merrill Lynch, Deutsche Bank, HSBC and RBS, as well as within asset managers.

The type of roles, however, has changed: "This year we've seen a strong increase in the number of operational risk positions, where as the focus in 2009 was very much around credit risk," says Priya Mariannie, senior consultant, risk at PSD Group. "Banks are organising themselves to absorb a highly regulated environment.

I have a few concerns about the measures that were either announced or proposed under Basel III because of the differences between developed and emerging markets," she said at an international banking conference at the Chicago Federal Reserve.

This divergence on ratings agencies is a striking disagreement between Dodd-Frank and Basel III. Perhaps it doesn’t bode well for the level international playing field, that favourite rhetorical tool of bank lobbyists seeking to water down regulation; but at least someone’s started to have a crack at the problem.

...

The role of RWAs is another big divergence between Dodd-Frank and Basel III. In Basel II and III there are rules that (roughly speaking) map from the riskiness of an asset (deduced from credit ratings and creditor type) to an amount of capital to be held to cover possible underperformance of the asset; this is the concept of risk-weighted assets, RWAs.

Certainly Basel III toughens bank capital standards, probably in idiosyncratic ways depending on the nationality of the bank (some of those old capital loopholes crept back in). But the RWA system also has a big role in defining the capital requirement. So boosting the capital requirement without fixing the weaknesses of the RWA approach, or finding a way to chuck it out altogether, leaves a big unsolved problem flapping in the breeze. In particular, the mythmaking around  the chimerical riskless asset, “AAA”, is still intact, with the promise of a zero capital weight for a suitably rated  asset. That is a good game, of course, with dumb reg-arbitrage-aware banks accumulating vast holdings in the ‘riskless’ stuff: see CDOs 2004-2007 (step forward ML, UBS, AIG, Landesbanken etc etc) or sovereign bonds right now (step forward SocGen, BNP, and, one suspects, Landesbanken again). Sport, too, for the smart black-hearted fiends who spot the illusion; last time that was Magnetar, GS, Lippmann at DB , Michael Lewis’s Big Short crew. I’m not sure who is going to profit from sovereign bond defaults or restructurings but I imagine the positions are already “on”.