In Europe and the US, credit ratings have become a political football. Consequently, the risks of policy mistakes and also unintended consequences are great. Enormous investor losses and even bigger tax payer funded or guaranteed bail outs mean that western politicians need to find someone else to blame, and fast. The dogs of war have been sent out. The easy scents for them to pick up are those left by smoldering financial instruments that anyone unfamiliar with finance would find hard to understand: derivatives and structured products. The motives behind those responsible for these instruments?bankers and credit rating agencies?have been questioned. Their business models smell bad. In general, profits for a few have given way to losses for many, fuelling a moral outrage that in turn leads to a demand for more regulation. The financial system needs better regulation. But regulation is not like pouring milk in your tea, with a little more or less to taste. We can have good or bad regulation and bad regulation is a lot worse than milky tea.

Indeed, the most recent regulation on rating agencies in the US before the credit crunch, to provide more disclosure on the ratings process, may have contributed to the destabilising behaviour of arrangers building structured products ?to rating?. The essential point is that to guide us towards better regulation, we need to have a good understanding of the role of credit ratings.

The way to think about credit ratings is analogous to Nobel Prize winner George Akerlof?s tale of ?the market for lemons?. In American parlance a lemon is not just a citrus fruit, but also a bad second hand car. Mr Akerlof explained that in the absence of independent ratings of quality, ?there is no such thing as a good second hand car??well, within reason.

The argument is as follows. The only person who knows whether a second hand car is good is the current user. He or she knows all the creaks and cracks, how temperamental the gearbox is, whether gremlins live in the electrics or not and more. The potential buyer does not. So the potential buyer hedges his bets and is prepared to buy the car for the average price between a good and bad second hand car of that make, vintage and outward wear and tear. The owner of a good second hand car knows this average price is too low a price and, if he can afford to, he withdraws his car from the market. The owner of a bad second hand car knows this is a good price and sells his car. Informational asymmetries cause adverse selection.

The Akerlof story suggests five important things about the second hand car market (or any market where there is uncertainty about the long-term quality of a products). First, the distribution of second hand cars for sale will be skewed towards a disproportionately large amount of ?lemons? for sale and a disproportionately small number of ?good? second hand cars. Second, the market will be small, as sellers of good second hand cars stay away if they have the choice, and buyers too, as countless stories of disappointment and suspicions keep all but the brave or expert away. Third, because the market finds it hard to differentiate between good and bad second hand cars, there will be a low correlation between price and the underlying fundamentals. One of the consequences of this is that owners invest less in the innards of a second hand car (its fundamentals if you like) and more in its outward appearance ?paint and body work. Fourth, this lack of differentiation suggests that the range of prices will be narrow. Fifth, the more other cars have independent ratings or checks, the more the absence of a rating causes suspicion and leads to a price discount in market prices.

Clearly these first three characteristics are not good for sellers, buyers and even manufacturers of second hand cars. The response has been the ?approved used car scheme? where the manufacturer?s dealer has checked a range of things that is considered important (the manufacturer has an interest in supporting prices obtained in the used car market).

The economics of the market for lemons with its information asymmetries and adverse selection can be applied equally well to the market for credit. The asymmetry of information is there. The issuer has a better idea of the underlying credit quality than a buyer. This asymmetry is even more built in to a bond than a car because credit quality is determined by the issuer?s future decisions.

The Akerlof theory predicts we would observe the following characteristics of a market with credit ratings compared to one without:

* The distribution of issuance will be more ?normal? with more issuance of loans and bonds by good quality credits and less by bad quality credits.

* The market would be larger as good quality credits are enticed to the market and buyers have greater confidence.

* There will be a stronger correlation between price and the underlying fundamentals.

* Increased differentiation of credit quality will allow for a greater range of credit quality to come to market.

* The more bond and loan issuers are rated, the greater will be the yield premium for issuers who are unrated.

Interestingly, the Akerlof story does not say anything about observed average credit spread. Yet almost all of the studies on ratings have focused on an analysis of spreads. In the Akerlof story, ratings can lead to higher as well as lower observed spreads. A good quality credit should receive a lower spread than before its rating, though this would be hard to observe because before its rating, it would not have issued. A poor credit may find that its yield spread has risen as a result of a rating, and as a result it may not issue going forward.

We can attempt to falsify each of the five predictions above with an empirical study. Jean-Louis Warnholz and I have carried out-the largest, and arguably most in-depth study of credit ratings looking at over 1,000 sovereign loans and bonds over the past 30 years using data across all of the major rating agencies, namely Standard & Poors, Moody?s and Fitch IBCA.

We found the following:

(a)The distribution across ratings of credits that have a rating, is very different from the distribution of credits across estimated ratings where they are not rated. To paraphrase Akerlof, without ratings there is no such thing as a ?good? credit (well, not much). Ratings support relatively more good quality credit issuers and a wider range of issuers.

(b)The fundamentals are better at explaining yields where there is a rating than where there is not. This is consistent with the Akerlof story that where it is hard to differentiate between good and bad bonds, buyers offer a price that is an average between the two.

(c)The more other loans and bonds are rated the more the absence of a rating drives up the yield.

(d)As ratings have become more common, the market for credit has grown significantly.

These results will be intuitive for anyone familiar with ratings, but they have wide reaching economic implications. Ratings raise the amount of borrowing and lower its cost. Ratings provide greater opportunities for investors to match and hedge liabilities, reducing risks for a given return. Ratings significantly improve the incentives of governments to improve their credit fundamentals. We need to improve the credibility of the business model of rating agencies?perhaps through standardised definitions of what a rating means or through a government pays model?but we must not in the process destroy the independence of ratings. We need them to deepen India?s credit markets.

?The author is chairman of London-based Intelligence Capital, governor of the London School of Economics and emeritus professor of Gresham College