Skip to content

Local Maxima vs. Global Maxima: The Case Against Nudging

August 23, 2016

Look at this. Can you guess where I’m going?

tumblr_inline_mqr8kd4lrq1qz4rgp

Image Credit: Andrew Parker of the Gong Show

I’ve been thinking about this image pretty often in the month and a half that I’ve been officially working in the behavioural economics field. Sometimes, it worries me.

Why? Because for the past couple years, my main interest has been in complex systems, and how they might be shifted into entirely different (and much better) ways of working. I went to San Francisco because I was interested in how Minerva Schools was drawing a radically different future for higher education. I went to India to learn from the incomparable Pradeep Ghosh about how social problems might be solved with a specific kind of innovative thinking. As I wrote then:

I am interested in how communities can be changed for the better. How does that happen? How do you address problems that part of “the system” itself? And by system I mean all the people and groups in a community and the ways that they interact. The system is partly created by the rules we lay out for ourselves as communities and societies (formal laws = rules, informal = values), partly by some things we can’t always control or change (geographic factors like how far away houses are, or biological factors like aging) and partly created by how we interact with these rules and with each other. It sounds messy, right? Let me take another pass: when we talk about a “community”, it’s tempting to want to picture only the people. That’s a community, right? If you want to fix a problem, you can just help individual people – give them a leg up or something like that.

I think that if you only look at people, and if you look at them all separately, you end up trying to plug holes in a leaky tub but never turning off the water.

Someone explained it to me like this: There are two people, and a river, and there are bodies floating in the river. One person immediately starts fishing bodies out, and another walks away. “Where are you going? How can you leave me when there are so many bodies to be pulled out?” says the first one. “I need to go figure out why they are there in the first place,” the other calls back as she keeps walking.

And we need both kinds of people (we need all kinds of people!), but I strongly identify with the second kind of person, the one who pulls back and tries to see the entire landscape, rather than helping people one by one.

The interesting thing for my personal integrity is that behavioural economics, at least the Version 0.1 that is being implemented in the public sector right now, is arguably more about saving the individual people in the river than it is figuring out why they are falling in in the first place.

For example, let’s look at tax collection. Studies around the world (but especially in the UK) have shown that when you send a letter to people with overdue payments telling them that the majority of their fellow citizens pay their taxes on time, and that they are in the “small minority that requires extra collection efforts”,  these delinquent payers actually get on the bandwagon and fork over more cash than they would have without that letter. That’s great, but it’s not going to affect the wealthy businesspeople who squirrel away as much of their earned income as they can so that it can’t be clawed back. How many ‘average’ people nudged into paying a little more will it take to equal the hidden millions of a few top families?

Nudging (a more colloquial term for behavioural economics) is be easier to do than reforming our tax laws, and that’s probably why the rewards are lower. Successful nudges result in bumps in the single to low double digits. Maybe 86% of people pay on time, and with a nudge, we can get that to 88%. Because nudges are so cheap, even these small bumps can result in huge returns-on-investment. They work quickly, and in a fast-paced political climate it is easy to love things that give results quickly.

I’ll throw up this picture again, because I think I’m finally ready to talk about it.

tumblr_inline_mqr8kd4lrq1qz4rgp

Image Credit: Andrew Parker of the Gong Show

When we tweak an existing letter, or train a social worker to ask slightly different questions to their clients, we are searching for a local maximum. We are trying to leave the core of the system or process intact, and instead trying to make peripheral changes that will make it run just a little bit better than it did before. If we are midway up a hill, we will keep walking until we get to the top. We won’t even consider the fact that there are other hills, because to get off of our current hill, we would have to descend – face failure, disrupt people’s lives, and generally cause chaos. Who wants that? There isn’t even a guarantee that there is something better out there!

But, if there is a better way out there, if there is a taller hill that we need to climb, if we refuse to move off of our hill, we will never find it. That’s why behavioural economics Version 0.1 is unsettling to me. We focus on making incremental improvements to existing systems, even though we know that more fundamental changes might be required.

I know I’ve been a bit uncharitable to behavioural economics in this post, though I am trying to make very clear that it is the Version 0.1 implementation that I am seeing in governments across the world to which I am reacting, not the field in general. I think there is a case to be made for behavioural economics, but that can wait for next week’s post.

Until then, keep climbing, my friends 🙂

Advertisements
No comments yet

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: