Friday, August 5, 2011

Ratings agencies

I was going to blog about rating agencies tonight, and I guess that now I actually have a nice justification for it. S&P just downgraded the US credit rating by one notch from AAA to AA+. There are two big questions related to what I wanted to write about:
- Does it matter?
- Should it matter?

I was mostly going to blog on the second one as a follow-up to my post 2 weeks ago on polls and forecasts. As I said there, you have 2 possible objectives for those institutions: they can inform(tell you what the state of the world is now) or predict(what the state of the world is going to be). The latter is more complicated. Today's example is quite good to understand that: the very fact of downgrading the debt puts the US(and the world ouh lalalala) in uncharted territory. Thus, it is clear that the downgrade is not supposed to be a forecast, and it should be an assessment. But as Ezra Klein points out, S&P is actually trying to make a forecast on how the deficient US political system is going to deal with the rising debt.  That's one reason why S&P is probably in the wrong. It is not their job to forecast the US credit rating. Jared Bernstein said on MSNBC tonight that the agencies are there to tell us whether it's safe to invest in certain sovereign or corporate bonds. I disagree.  Their job is to inform of the current situation. And the current situation is clear: the US is creditworthy and the only problems are long-term And that's also why there will be no movement as a consequence of the downgrade. As Felix Salmon noted yesterday after the international market crash
if there ever were serious doubts about the USA’s creditworthiness, they’re gone now. The 10-year bond is yielding less than 2.5%: there are no worries in evidence there. 
More generally, it is really important to wonder why we are using rating agencies. At the time being, with private agencies, we cannot consider them as forecasters. The rating agencies have interests that are not aligned with perfect forecasting. The main issue in financial markets is that they are paid by the issuers of the securities that are rated. More generally, private forecasters usually have competitive and reputational incentives that lead to biased forecasts. Those two things actually have opposite effects. The former problem is discussed by George Henry

[A] technique for seeking attention is to produce a forecast that departs sharply from
the consensus
The reputational factor, instead, might lead agencies to herd, as suggested by Scharfstein and Stein
 herding may become more or less of a problem as a manager’s career progresses. One the one hand, there is apt to be less uncertainty about the manager’s ability, which should reduce the incentives for herd behavior. On the other hand, later in a successful career, wages are probably higher above the outside alternative. . . This latter effect can increase the propensity to herd. 

Usually, the evidence points to the latter. Owen Lamont showed that, for instance, that
ex post forecast accuracy grows significantly worse as forecasters become older and more established.
Errors are growing worse as the forecaster ages, which would point to a strong reputational incentive.

The bottom line is this: if you have private rating agencies, you don't want them as forecast, because their objective is not to have a perfect forecast. You want to use their ratings as information. Again, given the flawed incentives, you will have to adjust the information. Interestingly, this is what people in credit rating agencies will tell you, apparently. In an interview to Liberation, the left-leaning French newspaper, Carol Sirou, president of S&P for Europe, Africa and francophone Middle East,  argues that :
Nos notes sont dans le domaine public, chacun est libre de s'y référer ou pas. Mais le fait est qu'aujourd'hui, nos décisions sont souvent amplifiées et entraînent certains comportements, liés à l'anxiété ambiante. Trop d'acteurs utilisent nos notes comme seul critère de décision. Nous sommes favorables à ce qu'ils considèrent aussi d'autres paramètres. 
What does it mean in practice? The fact that rating agencies should not be used as forecast, but that their decisions should be evaluated as a partially informative signal by investors, means that ratings should IN NO WAY BE IN INVESTMENT RULES.  That is also why I don't really like when rating agencies are blamed for the pains in Europe for instance. We shouldn't care. The problem seems to be mostly political, and legislative.

As pointed in several papers(e.g. Steiner and Heinke(2001)), what you actually see is that downgrades have an impact when the downgrade is from investment grade to junk, and only then. Basically, the downgrades are useless, and their only effect is through the regulations that impose a sell-off by pension funds required to invest in certain categories of financial assets. All is told in the figure here:
You are interested in the lighter curve, which shows the "cumulative abnormal return"(CAR) around a downgrade. The downgrade occurs at time 0 in the graph. You can estimate what should be a "normal" return from the past. Given this prediction, you look at the actual value, and you get a day by day abnormal return(the black thingy). The CAR is just the sum of abnormal returns. What you see in the graph is that at the time of the event, the downgrade has been priced in, and the sudden fall just after the event is just an overshooting, since the situation goes back up after 15 days. Actually, as Steiner and Heinke put it(italics mine, obviously):
In the long run(which is what institutional investors are about!) the investor is better off not to sell the bond on the announcement day because there is a positive cumulative excess return from day 1 to day 90 of 0.526%.
You also see evidence that the size of the downgrade(how many notches) does not matter, evidence for the fact that ratings actually provide few information, and that their only impact is through the constraining rules imposed on institutional investors.

My bottom line, in theory, is this: ratings should not be in the rules. There should not be rules restricting investment to triple A bonds. This does not make any sense.

On the US problem now, it's actually clear, I think, that nothing is going to happen, for a couple of reasons. First, as mentioned above, investors not subjected to binding rules on what they could invest in will not react. As you saw yesterday, people were still flocking to Treasuries during the market crash, while the downgrade has been in the making ever since Barack Obama announced the agreement on the debt ceiling. Reuters most recently reported on a JP Morgan study showing that
The number of investors that are restricted to by Triple A mandates is minimal. Expectations are that around USD40bn could be pulled from US Treasuries as a result of force selling. 
[JP Morgan] see[s] little direct impact of a downgrade in the short term, but the medium term impact is Treasury yields in the range of 60bp higher than otherwise, which is a significant cost to the Treasury.
Third Way has a somewhat larger increase in yields, at 75bp. Fitch does not care
Second, the problem is, again , the downgrade from investment-grade to junk. AAA to AA+ does not matter. As Felix Salmon reported
“the long-term credit performance of ‘AAA’ rated versus ‘AA’ rated issuers,” says Fitch, “is statistically negligible”.
Third, there's nowhere else to go. The US issues 59% of triple A sovereign debt. The other AAA countries(yep, France) do not look that attractive right now.
Fourth, Treasuries have a special role in certain institutional investors(can't find the link, sorry). Basically, those investors can invest in Treasuries not because they're triple A, but because they're Treasuries. Because a downgrade does not change the name of the bond, there will be no impact for them.

Etc, etc... Bottom line: no biggie. I am taking a risk, I don't like making predictions. Well... I should probably start at some point.

No comments:

Post a Comment