Decisions Decisions...

I often check The American Economic Review in hopes of finding new research related to behavior change, choice, and utility.

I found ‘Speed, Accuracy, and the Optimal Timing of Choices’ and had to read it. The paper: ‘model[s]…the solution of optimal sequential sampling, where the agent is uncertain of the utility of each action and pays a constant cost per unit time for gathering information’.

You may be thinking, why is someone reading that, but as someone who has a business focused on choices, it sounded great: ‘Oh cool! How do people make the right choice when they don’t know which option is best, and when should they make the choice given the fact that gathering information to validate/justify said choice costs time?’ I could certainly be better at making optimal choices and knowing when to think fast or slow (if you haven’t read that book, know that I’m sending major side eye).

It's a highly technical paper with sets, probability functions that could be decomposed into multiple conditional probability functions, and standard Brownian motion. But the paper isn't just for Economists, it has actionable insights for anyone.

Can we gleam actionable insights from a highly technical economics research paper?

Let us see.

Here are my interpretations supported by anecdotal evidence.

1. The agent’s problem is to decide which option to take and at which time. Waiting longer will lead to more informed and thus better decisions, but also entails higher costs. What matters for the decision is the difference between the two utilities

My Take: Decisions require research and thought, they require time. But when making a choice between two options (a key assumption), what really matters is what will provide me with the highest utility (aka satisfaction).

For example: I love coffee and fresh baked goods, but I don’t love spending $9 every time I go to a cafe. Do I go for the coffee or the baked good?

Research Required: this is ‘required research’ because it impacts which option will give me the most utility.

• Is it after 2pm? If yes: baked good If no: coffee

• Have I already had a fancy sweet this week? If yes: coffee If no: baked good

What’s the point of this example? An ‘accurate’ decision depends upon the result of said decision being made. It’s a basic choice when it comes to coffee versus a baked good (to be honest, they’re pretty similar in terms of my utility derivation: both make me incredibly happy), but it’s important to recognize which option will provide the most utility. So I do the research steps (both of which have minimal time costs), and the choice is made!

Takeaway: When choosing, aim to maximize utility.

2. The optimal strategy depends on … the prior variance. For example, if [a] and [b] are two houses with a given utility difference [x], we expect the agent to spend…more time here than on a problem where [a] and [b] are two lunch items with the same utility difference [x]…We expect the prior belief of the agent to be domain specific…and the variance of the prior to be higher for houses than for lunch items

My Take: The time we need to take to make a choice varies based upon our existing knowledge within the respective area.

Using the house versus lunch example from the paper: people are good at choosing what to have for lunch, and it’s a relatively easy decision — we eat every day. But when it comes to house buying, especially if it’s a first purchase, most people don’t know where to start.

They might start by: figuring out their budget, mortgage rates, asking friends or family members who own property for advice, location vs cost, cost of future renovations, the foundation… all of which take time to understand. (Writing an accurate description of what’s important to know before buying a house would require me to invest some time increasing my own domain knowledge).

Take away: If you have to make a choice, and have no idea how to go about choosing — take the time to do some research.

3. There are two reasons for the boundary to decline over time: as data accumulate, additional data is less likely to lead to much change in beliefs, and if a decision hasn’t been made after a long time the two decisions are probably about as good

My Take: The more that is known about a decision, the less additional research is needed. If there’s not a clear choice to be made, the utility to be derived from either option is likely the same.

For Example: I want to get a plant for my apartment, and I am a newb.

Research Required:

• What indoor plants are hard to kill?

• What indoor plants are okay with indirect light?

• What size plant will fit on my table?

• Where is a good place to buy plants?

At this point, my search has been narrowed down to either a ZZ (Zamioculcas zamiifolia) or a snake plant (Sansevieria trifasciata) from a local nursery.

Between the ZZ and the snake plant, there’s not a clear reason to pick one over the other given their comparable qualities — so it doesn’t matter which one I choose, the utility will be equal.

Take away: Research until an option is clearly the superior choice (from a utility perspective) or the options seem equal, in which case just go with one.

Look at that, I gleamed actionable insights from a highly technical economic research paper after all. Here’s to making better choices!