Any measure that becomes a target ceases to be a good measure.
Easy, simple sentence, but such an unfortunately powerful one. We’ve covered Goodhart’s Law before, of course. But more’s the pity, the world insists on giving us fresh examples. And the latest entity to oblige is the Bank of Baroda:
On Tuesday, the Reserve Bank of India took matters into its own hands and ordered Bank of Baroda to stop onboarding customers on its mobile app “BoB World”. The central bank wrote:“This action is based on certain material supervisory concerns observed in the manner of onboarding of their customers onto this mobile application. Any further onboarding of customers of the bank on the ‘bob World’ application will be subject to rectification of the deficiencies observed and strengthening of the related processes by the bank to the satisfaction of RBI.”
https://themorningcontext.com/yesterday/what-is-going-on-with-bank-of-barodas-app
What are these “certain material supervisory concerns”?
Well, Bank of Baroda’s mobile application is called BoB World. When the Bank launched this application, they expected all of their customers to use it, and expected all of their employees to encourage their customers to do so.
A little while later, it would seem, the higher-up’s at the bank were less than satisfied with the number of sign-ups for the application. Reprimands were issued (of course), and targets were set. Sign-ups must increase, went the diktat, and employees further down the pecking order were asked to ensure “x” sign-ups per month.
Goodhart’s Law woke up, and went to work.
Last year, The Reporters’ Collective and Al Jazeera published a report with findings that showed Bank of Baroda employees in Uttar Pradesh, Rajasthan, Gujarat and Jharkhand had linked bank accounts of customers without a registered mobile number with any mobile number they could get hold of—including their own, those of other bank staff, managers, security guards, relatives and friends—and used the OTPs generated at the back end to register them on the mobile app.
https://themorningcontext.com/yesterday/what-is-going-on-with-bank-of-barodas-app
The employees allegedly resorted to these measures out of desperation. Reprimanded for not being able to reach the target number of customers to register on the app, bank executives – who later turned whistleblowers – found the workaround from their colleagues in other branches.
Imagine a poor, uneducated worker in some small town in, say, Uttar Pradesh, who has signed up for a bank account with the Bank of Baroda. Such a person may have chosen to not register their mobile number with the bank. So a junior employee at the bank figures out that:
- They have to register x accounts on the app
- Here is an account that is not yet on the app
- Said account holder hasn’t registered their mobile phone number
- So… let’s add this account on the app, and link it to some other number instead
But which other number? Could be the phone number that belongs to the branch. Could be a nodal officer’s mobile number. Could be the nodal officer’s wife’s mobile number! (This really happened, by the way).
Turns out, some 1200 mobile numbers were linked to anywhere from 30 to 100 bank accounts… each. That is, one mobile number could be linked to anywhere between 30 to 100 bank accounts!
It gets better (or worse, if you prefer). Another application of Goodhart’s Law in this case is that quite a few banks now mandate that a mobile phone number be entered at the time of registration. But what if you don’t have one? Simple, the employee helping you open a new bank account simply adds their own number, or the official number of their branches.
Where will the OTP for such accounts go? That is, to which number? What all can you do if you can get OTP’s associated with another person’s account on your mobile phone?
Here’s how it plays out. Consider A, who realises that their phone is linked to your account. Even without knowing account details, if A can generate an OTP using their number, they can access all your banking details, including the funds you have. Likewise, they can also transfer money using OTPs. And the confusion only intensifies once payments are made to the phone number directly through UPI.
https://themorningcontext.com/yesterday/what-is-going-on-with-bank-of-barodas-app
And this is exactly what happened. According to the report, which cited internal documents of the bank’s head office, bank agents withdrew thousands of rupees from customers’ accounts using mobile banking. These agents, or “business correspondents”, operate as freelancers on behalf of the bank providing services in remote areas.
An underrated subject, principles of economics. More people should study it.
Including, Navin Kabra points out, the folks at Wells Fargo:
Adding online banking to a checking account was a “product” that counted toward your quota of X products per day, same as adding a mortgage; a checking account with $1 in it that closed in a week counted as much as a credit card that carried a $20,000 balance for years. Really just a rough cut at measurement.
https://www.bloomberg.com/opinion/articles/2022-05-19/is-the-sec-unconstitutional
Knowing that, you can figure out what happened next. Wells Fargo bankers, struggling to meet quotas, opened a lot of low-value accounts. People who signed up for checking accounts got signed up for online access whether or not they wanted it; they got pushed to open credit cards too. And bankers often went a step further and signed customers up for accounts without their permission. Occasionally this helped Wells Fargo and harmed the customers: They had to pay fees on accounts that they didn’t want or know about, and Wells Fargo collected the fees. Sometimes this harmed the customers without helping Wells Fargo: Their credit got worse due to new debt accounts, but the accounts were never used and Wells Fargo never made any money. Often nothing happened to anyone — the customers had $0-balance accounts that were quickly closed with no money changing hands, or free “products” like online banking access that they never knew about or used — but the bankers got to meet their quotas.
The point is just that Wells Fargo had a crude measure (“products sold”) that was meant to approximate an important thing (something like “value extracted out of each customer”), and its overstressed retail bankers cynically optimized the crude measure rather than the important thing. And then Wells Fargo got in huge trouble and was fined a lot of money and it was, like, the major US banking scandal of the mid-to-late-2010s
So did they learn their lesson, the folks at Wells Fargo? If ol’ Goodhart is in the building, he doesn’t check out anytime soon, unfortunately:
Mr. Bruno is one of seven current and former Wells Fargo employees who said that they were instructed by their direct bosses or human resources managers in the bank’s wealth management unit to interview “diverse” candidates — even though the decision had already been made to give the job to another candidate. Five others said they were aware of the practice, or helped to arrange it.
https://www.nytimes.com/2022/05/19/business/wells-fargo-fake-interviews.html
The interviews, they said, seemed to be more about helping Wells Fargo record its diversity efforts on paper — partly in anticipation of possible regulatory audits — rather than hiring more women or people of color.
By the way, you might think I’m trying to be funny when I say that Goodhart doesn’t check out anytime soon, but I’m not. Culture is underrated by economics and finance:
I say “come on!” but, you know, this stuff is hard. Wells Fargo is a big bank. To some extent you have to manage a big organization with simple, crude, legible metrics. In a big enough organization, somebody will game those metrics. Ideally you create a culture that minimizes that gamesmanship, one where employees understand and buy into the organization’s real goals rather than just trying to maximize the dumb metrics. Wells Fargo seems to be having some trouble creating that culture.
https://www.bloomberg.com/opinion/articles/2022-05-19/is-the-sec-unconstitutional
My thanks to Navin for reviewing this post, and making it much better.
EconForEverybody 