[matasar]

It seems to me that rating agencies should be paid by the institutional investors that buy the bonds, not the banks trying to sell them. That might help the incentives line up better.

Am I crazy? Nobody suggests this, and I think I'm missing something crucial here.

[jey]

Rating agencies seem like an absurd idea altogether. Have they never heard of the perils of having a single point of failure? Much less a single point of failure that's a government-created oligopoly? Yikes.

It seems that businesses are just running on an outdated model developed when information sharing was a lot harder, so only condensed forms like quarterly reports and press releases were feasible. But now that information sharing is a lot easier, shareholders should be demanding more openness from businesses -- yes, today a lot of that information is considered to be a trade secret, but at the same time shareholders need to stop allowing businesses to get away with what's equivalent to deceit through "creative" accounting (e.g. apportioning profits/debts amongst subsidiaries). That's just a textbook example of exploiting information asymmetry. Shareholders should expect more information about the business' books and transactions. There must be some natural equilibrium between being open about transactions and protecting strategic advantages, but right now we have the functional equivalent to price fixing, er, information fixing -- companies don't release relevant information because "no other company does" and shareholders don't expect real operating information either because that's not part of the status quo.

Wouldn't an investment bank that was so confident in the principles behind its analysis that it was willing to list all its transactions in a continuously-updated XML file, despite the risk of copy-cats, seem like a pretty damn good investment? They wouldn't have to divulge their proprietary methods of analysis, just make their holdings public. This would greatly improve the efficiency of markets as investors could use their own proprietary methods to estimate the risk of an entity's portfolio management strategy when deciding whether to invest in it.

And there's definitely historical precedent for this -- the fabled value investor Benjamin Graham didn't go to great lengths to hide his trades, instead he'd use them as examples in his classes, and yeah, people copied him. And he still made boatloads of money.

[gaius]

Well, it's more subtle than that. The big ratings agencies are Fitch, S&P and Moody's. They are competitors for the business of issuers. So each one is incentivized to rate higher than the others, without blatantly being seen to take the piss.

The problem with complete openness is that it encourages short-termism. You see this even with quarterly results, companies that have recently gone public (and thus have minimal reputation) manage from quarter to quarter to quarter and are hugely volatile. Imagine working for a manager who only cares about the share price tomorrow.

The risk of your XML file is not copycats, it's front-running.

[jey]

The short-term price motivation is a great point. I'll think about that some more.

But how does this scheme encourage front-running?

[btilly]

The problem is that the institutional investors buying the bonds are much smaller, and there is no way to keep the information from leaking. Therefore most of those investors won't be willing to pay for the bonds, and the ones that can pay will be paying less money.

That doesn't mean that this model is not viable. But it isn't viable at the kind of margins that the rating agencies would like to remain accustomed to.

[matasar]

Maybe investors could pool resources to start their own rating agency that would be answerable to them, and publish ratings for everyone to use? There has to be a better system.

[btilly]

The problem with that is that the ratings are a public good. Public goods have some counter-intuitive properties. Investors would have every incentive to individually contribute as little to the pool as possible as long as action happened. Economic theory says that one of four things is likely to happen. Those are:

1. The knowledge is worthwhile to a small group (often just one) of investors, who fund it. The memorable phrase for this is "the exploitation of the large by the small." (A practical example of this is OPEC and high oil prices. The complicated negotiations OPEC engages in illustrate the desire of members of the group to contribute as little as possible to provisioning the public good.)

2. An organization exists that investors belong to for some other reason which funds the ratings. (A practical example of this how people belong to AAA for membership benefits, but then AAA lobbies for road improvements.)

3. A coercive organization intervenes and forces the matter. (Virtually all government regulations fall into this category.)

4. The public good is not provisioned. (What happened to investors who wanted trustworthy ratings.)

Read The Logic of Collective Action for the classic introduction to this topic.

[jimbokun]

This seems somewhat similar to open source software, where there is also a free rider risk. This has not stopped large corporations from contributing to open source software, because there are often strategic advantages stemming from the software's existence and wide distribution.

I'm not sure how this applies to an investor funded rating agency. I don't know that there is any strategic advantage to large firms having accurate bond ratings publicly available, but it would be interesting to hear business models where there would be.

[btilly]

Open source is an interesting example.

Normally there are transaction barriers that inhibit cooperation in providing public goods. However software has far fewer of those barriers. And so you get a situation where a network of loosely connected people each pursuing individual goals can collectively provision a public good.

I remember running across a footnote in The Logic of Collective Action saying that this was a theoretical possibility, but no example was known of it. (The book was written back in the 1960s.) In any case it is an extremely unusual example.

[cma]

http://www.cepr.net/index.php/beat-the-press/krugman-nails-t...

Dean Baker has been suggesting that either the exchange or the SEC take over selection of the rating agency for over a year now.