I’m automating my finances on Twitter

Using A.I. to fix my credit card’s choice architecture

I recently set up @CacheAI, artificial intelligence to automate my finances, & to make sure of one thing: that the right money is in the right place at the right time. I’ve given Cache full access to my bank account, to automate transfers between my checking, savings & credit accounts at Bank of America. Cache will tweet at me when it’s going to make a transfer, stating how much & between which accounts. I can say no to a transfer, but it’s more fun to say yes.

Why am I doing this? For two reasons. First, automation is useful, especially for tedious tasks. For example, I’ve asked Cache to keep my credit card paid off:

Every few days, Cache pays off any new balance on my credit card, or as much as possible, while still doing its best to make sure my checking account has the right funds to cover upcoming bills, such as rent. As a result, I no longer need to remember when my credit card payment is due. I can treat it more like a debit card, but still benefit from its credit score building & cash back features. And if Cache ever lets me carry a balance, it’s evidence that I’m overspending. So far so good.

I could do this privately. Why Twitter?

Because I’m also doing this as an appeal. I care deeply about problems in consumer finance, perhaps to a fault. With the help of my colleague, Gael Reinaudi, I’ve been exploring solutions in the space for some time. This post, as well as our research page, is a call to see what you think. We could use your input. Building solutions is hard, especially in financial tech.

But finding problems is often easy. And as I see it, one problem is how credit card payments are designed.

To take out a student loan or a mortgage, you have to apply for it. You have to do something, with intention, to get it. But few of us apply for our first credit card because we need to take on debt. Rather, we apply for credit cards because we‘ve been enticed by some cash back reward, or because we’ve been told that credit card ownership is an essential step toward building good credit.

When we do actually take on credit card debt, it’s a passive act: we need only neglect to pay off the full balance when it’s due. In other words, rather than having to do something, actively, to take on debt, with credit cards we must repeatedly do something to not take on debt (we must make monthly payments)¹. In my opinion, this is a failure of choice architecture.

Choice architecture, a concept introduced by behavioral economists Richard Thaler & Cass Sunstein, suggests that how we present choices to consumers can influence the decisions they make². What makes it a remarkable topic of study is the extent to which small adjustments in presentation have produced significant changes in behavior. Take the Schiphol Airport urinal experiment, for example. The following is from Thaler’s & Sunstein’s article, Easy Does It³, in The New Republic, in which they describe the airport’s efforts to encourage restroom cleanliness:

Enter an ingenious economist who worked for Schiphol International Airport in Amsterdam. His idea was to etch an image of a black house fly onto the bowls of the airport’s urinals, just to the left of the drain. The result: Spillage declined 80 percent. It turns out that, if you give men a target, they can’t help but aim at it.

So many parts of our lives are choice architected. We find ourselves on mailing lists because we missed the pre-checked opt-in box when we signed up. Grocers often put their fresh produce at their store’s entrance; even if we end up buying the packaged goods at the back, we’ve been primed to feel better about our shopping experience⁴. When I shop at IKEA, I often have no idea where I am, but I always seem to know where to go next.

Defaults play an important role in choice architecture. Studies have shown that, all else equal, people are more likely to choose whatever has been provided as the default option⁵. “When in doubt, humans tend to do nothing,” Thaler & Sunstein write:

The default option determines what happens if the decision-maker takes no action… the usual answer is that, if you do nothing, nothing changes; whatever is happening continues to happen. But not always.

Suddenly finding yourself in debt is a change indeed. So this begs the question: have our credit cards been choice architected to facilitate unintended debt? Does the default option encourage, well, defaults on our payments? I think so. And more than $900 billion in outstanding revolving credit debt in the US⁶ suggests the same⁷.

Managing our finances is a hard problem. This is why I set up @CacheAI: to flip my credit card’s choice architecture right side up. My finances should be automated to do the right thing by default, & require my attention only when something important needs to be addressed.

So this is my appeal: Gael & I would like to build a smart credit card. One that never facilitates unintended debt, that helps us be wise about debt when we actually need it, & encourages healthy financial habits. Does this sound like something you’d use? If so, visit our research page & tell us. If not, tell us anyway. Let us know why.

This is the first in a series of posts I’ll be writing about how I’ve used technology to improve my own finances, & what I think the future of banking might look like if we place the consumer’s, not the bank’s, needs first.

Talking about money is taboo. And often, so is talking about the challenges we face as we search for our startup’s first product/market fit. But when we don’t talk about our challenges, we often don’t solve them. I set up @CacheAI to start a conversation around both. Maybe we can help each other out.


Footnotes

¹ One could argue that we take on credit card debt not because we neglect to pay the balance when it’s due, but because we are unable to pay when it’s due. In other words, that poor budgeting is the root cause of our credit card problems. I don’t disagree, but I believe that budgeting is a hard, unwieldy problem that runs deeper than the solutions provided by most budgeting apps. And I believe that our credit card payment UX is a part of that problem.

² Thaler, Richard H., Sunstein, Cass R.; Nudge: Improving Decisions on Health, Wealth, and Happiness; Yale University Press, 2008

³ Thaler, Richard H., Sunstein, Cass R.; Easy Does It, The New Republic, 2008

⁴ Nassauer, S.; A Food Fight in the Produce Aisle; WSJ, 2011

⁵ N. Craig Smith, Daniel G. Goldstein, and Eric J. Johnson; Choice Without Awareness: Ethical and Policy Implications of Defaults; Journal of Public Policy, 2013

⁶ The Board of Governors of the Federal Reserve System; Consumer Credit G.19 Release, June 2016

⁷ Is $900 billion in credit card debt too much? By comparison, there is about $1.3 trillion in student debt.