Mitsubishi, measurement and Goodhart’s law: James Saft

Send a link to a friend  Share

[April 26, 2016]  By James Saft

(Reuters) - The Mitsubishi Motors mileage scandal illustrates a key weakness at the heart of most investment processes: management by measurement and target setting often leads to chicanery.

Shares in the Japanese automaker have tumbled about 45 percent since it admitted last week it falsified fuel economy data on several models as employees tried to meet internal targets that the company itself says may have been impossible.

The National Highway Traffic Safety Administration has asked Mitsubishi for information on cars it sold in the U.S., opening the possibility of wider repercussions.

While you cannot understand without measuring, you are kidding yourself if you think measuring and target setting is the same as control, a point investors have learned to their cost time after time.

“Shareholders might consider again the uses of Goodhart’s law, which predicts that measures become moribund as they become targets,” Deutsche Bank analyst Sahil Mahtani wrote in a note to clients.

“Just ask the internet advertisers that covet digital ad clicks but lost $7 billion last year paying excess fees for the two-fifths of hits that were faked.”

Not to mention the very similar Volkswagen emissions debacle or, for that matter, the way in which the gaming of credit ratings contributed to loss, dislocation and disorder during the last financial crisis. Volkswagen on Friday announced 16.2 billion euro charge to its 2015 earnings to help fund the costs of the Dieselgate affair, centering on the use of software to cheat emissions tests.

Named after the economist and Bank of England policymaker Charles Goodhart, the law has as its central insight that the information you get from data, be it miles-per-gallon figures or bond prices, becomes muddied by manipulation when that data is used as the basis for a target.

Sometimes, as with Mitsubishi, this becomes outright falsification but the effects and pitfalls for investors are more wide-ranging and subtle.

Agents, like fund managers working on behalf of investors, or employees doing the same for shareholders, have an uncanny ability to produce results in line with the targets they are set, a process which often has little to do with the maximization of profits or a firm’s future well being.

Attempts to manage these sorts of conflicts, between what is best for a principal and what lines the pocket of her delegated agent, have been less than fully successful.

CULTURE MORE POWERFUL THAN MEASUREMENT

The last 50 years or so have seen the replacement of a culture in which managers of investments or companies were charged with meeting broad mandates, such as “do what is best for the client”, with one involving contractual bells and whistles which, while aiming to align interests, often conflate measurement with management. Humans being human, the results are very often very far from those intended.

[to top of second column]

Paul Woolley, a veteran IMF official and fund manager, and Dimitri Vayanos, of the London School of Economics have explored the ways in which the measurement of the “success” of fund managers creates perverse incentives and bad outcomes.

Since most fund mangers are charged with beating a market-capitalization index without taking too many huge bets, or beating an index of other fund managers' performance, they end up buying stocks which rise in price not out of conviction but as a means of professional self-preservation. Stray from the herd and you may get fired, so buy what ever is going up lest you be left behind.

The end result, Woolley and Vayanos assert, is lousy allocation of capital, leading to the kind of bubbles and volatile investment returns which have been the hallmark of the last two decades.

They argue for a system built more around mandates for “value investment”, a style which requires patience and a leap of faith by investors that their managers will do well over a much longer period.

Mahtani of Deutsche Bank suggests that using a matrix of two or more targets might produce better results.

Discussions about regulation, and by extension, about all agency conflicts, often end with stressing the importance of culture in an organization.

You can set banks and bankers with all of the risk-mitigating targets, rules and regulations that you like, but ultimately in a corrupt organization or industry with corrupt values you will get corrupt results. New targets are just new games to play. So it is with executives or with fund managers.



In the field of medicine, the patient's well-being, broadly defined, is usually set as the highest goal. Perhaps investment and management need a Hippocratic Oath.

(At the time of publication James Saft did not own any direct investments in securities mentioned in this article. He may be an owner indirectly as an investor in a fund. You can email him at jamessaft@jamessaft.com and find more columns at http://blogs.reuters.com/james-saft)

(James Saft)

[© 2016 Thomson Reuters. All rights reserved.]

Copyright 2016 Reuters. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.

Back to top