The Psychology of Paying Taxes | Alex Rees-Jones

The Psychology Behind Paying Taxes

Dan Loney: Did you know that there’s an element of psychology about how we think about paying our taxes, especially when you look at things like losses and gains? Wharton business economics and public policy professor Alex Rees-Jones has taken a deeper dive into this component of taxes, and he joins me to talk about it. Alex, what first got you thinking about psychology in taxes?

Alex Rees-Jones: When I was in graduate school in the late 2000 aughts, I was very interested in behavioral economics, which is an area of economics that’s all about trying to build more models from psychology into the way we do in regular economics. At that stage, that was a very well-developed field, but it wasn’t yet as successful as it could be because a lot of it was focused on, say, small lab experiments, and there weren’t an enormous number of demonstrations of this stuff being really useful to think about in big economic behaviors.

The thought I had was, “I think people could be influenced quite a lot by psychology when thinking about taxes. If that’s true, that would count as a big economic behavior, and understanding how to model it better through psychology would be really useful just for doing regular economics.” I turned out to be right, and there have been many other examples like that in the years since, but that’s what initially was my point of entry into the field.

Loney: I know you wrote about the losses component, about how we think about working the losses that we have in our tax preparation every year.

Rees-Jones: Let me summarize the idea on how losses and gains map into it. The thing I was trying to get at in this study was the idea of loss aversion would play out in tax settings. Now, the term “loss aversion” sounds pretty general. You can imagine it applying to a bunch of things, but when behavioral economists use it, they mean a very particular thing. They’re referring to this reproducible finding that people seem to value slightly increasing a gain less than slightly decreasing a loss.

To illustrate, let’s say I’m asking you how much you value a dollar that I’m holding. You think about it, and you’d hope you would say, “Oh, a dollar is a dollar. I’ll think about my value for that.” To put it into different frames, I could say, “What if I already owe you $10, so this dollar is getting mixed in and turning 10 into 11?” Or I could say, “What if you already owe me $10? This is turning (from) you paying me $10, into you paying me $9.” In either case, I’m just giving you an extra dollar. A dollar is a dollar. But the thing that you see, very reproducibly, is that people care a fair bit more about making the loss a little bit smaller, compared to making the gains bigger. That’s not so surprising that many people are like that. People don’t like losing. People don’t like losses.

What would that matter for taxes? The reason I got to thinking about this is there’s a very strong and natural gain/loss framing inherent in most people’s tax experience. Come every April 15th, when you have to fill out and submit your tax forms, a lot of this is going to be centered around doing a kind of gain/loss calculation, where the main thing you’re doing is going through documenting all the types of income you’ve had over the year and figuring out your tax liability. That’s what the point of this really is. But the last stages of it are going through and saying, “How much tax did I already pay,” say, through withholding from your employer or estimated taxes or something like that? Then what’s the difference, the balance due? And based on that, I decide how much money is changing hands between me and the IRS.

If I go through this, and I’m one of the 75% of people who over-withheld over the course of the year, whose employer kept more taxes than they needed to, I then determine, “Oh, actually the IRS owes me money.” I face a gain, or I could be one of the quarter of people who are under-withheld, in which case I determine, “Oh, I need to send a check to the IRS.” Now I owe the IRS money. The thing I want to think about is how much you care about doing something that changes your total tax liability, let’s say by a dollar. You could again say, “A dollar is a dollar,” but you can imagine that being in this gain or loss frame could matter. A person who is looking at a refund of $500 says, “I don’t really want to spend an extra half-hour looking for tax credits,” whereas a person who has to send a check to the IRS, they might be more motivated to find extra things they could deduct or cheat a little more or any of those types of things.

Loney: Is there a component of this that impacts the collection process by the IRS?

Rees-Jones: Absolutely, yes. To just briefly mention the finding, which will play into the collection bit — the thing that I find by looking at distributions of balances due that are reported is pretty clear evidence that something is changing, as gains turn to losses, and the way I rationalize it is by saying, “It looks like $34 more gets taken off your tax bill if you face a loss, versus a gain.”

What that means from the perspective of tax collectors is, first of all, when I’m forecasting revenue, I need to forecast the amount of tax avoidance activities that will happen, and that will be a function of gain/loss status. If I’m thinking where to aim my enforcement activities, who I’m trying to audit and things like that, this suggests you might want to aim at people facing losses a little more. I kind of suspect they’re already doing that, but I don’t know for sure. That’s the largest bit for tax enforcement at the higher level, when you’re thinking about how you want to design the system. You could design the system to have more or less people facing a gain or a loss on tax day. That’s something that’s within our control, based on how we set up withholding rules and things like that.

Loney: But you would think that it also has an impact on the person who is filing the taxes, on how they think about things like deductions and ways that they look for more deductions, so they’re facing a smaller loss. I think you even touch on it in the paper that, in many cases, we focus more on how we look at the losses than maybe we focus on how we look at the gains.

Rees-Jones: I think that’s right, yes. My understanding of what’s going on is absolutely that people are changing the degree of attention and effort they put in to, say, pursuing deductions, or maybe pursuing evasion, although I don’t see that directly. I think that is moving around. In some cases, that’s desirable; in some cases, it’s not, right? We basically have two reasons for taxing things. One is because we need money, and we view it as a shame that we’re messing up your decisions by taxing you to get that money; in which case, we really don’t want you to respond to the tax. In that case, it’s really convenient if you’re viewing it from, say, the gain frame and you’re not going to be as responsive. In other cases, we really want you to react to taxes. When we put out a soda tax, we’re trying to dissuade people from drinking soda by making it more expensive. The whole point of doing it is to get behavioral response. In those kinds of cases, if you could make things loss-framed rather than gain-framed, that would crank up the behavioral response and be useful. So that’s a way to riff off of this kind of observation.

Does Taxpayer Behavior Impact the IRS?

Loney: Part of this research looks at how this type of mindset can impact business owners, specifically small business owners.

Rees-Jones: The relevant pieces for thinking about business owners here are, first, they have lots of opportunities for either legal avoidance or illegal evasion. Relative to a lot of individual filers who just have salary income, business owners can get away with a lot more. Why do I say that? The biggest determinant of tax evasion is just whether your income is matched or unmatched. What do I mean by that? When the University of Pennsylvania pays me my salary, they also tell the IRS what they paid me. It’s matched in the sense that two parties said exactly what it is, and if I submit a number that’s different from what Penn said, they look at my return immediately. I have no opportunity to successfully evade my salary.

If I’m running a sandwich stand, there’s no third party saying how many sandwiches I sold. If the IRS wants to dispute what I’m doing or what I’m claiming my income was, they have to show up and audit me. Even if an auditor comes, it’s kind of hard to say exactly what happened. It’s a lot of work to successfully audit people. As a result, businesses have more opportunity potentially to evade or do gray-area stuff or legally avoid.

If you look at the total tax gap, which is the difference between the total amount of money the IRS thinks they’re supposed to get and what they actually get, nationwide it’s about 15%. We’re missing 15% of the money that we think we should get, which is pretty good as far as countries go on tax enforcement. But for individual non-business income, it’s about 5%, which is really good because there’s so little you can get away with. For business income, it’s like 40%, so there’s just a lot more money missing because it’s harder to get it.

In those cases, there’s a lot more of this discretionary thing where you can go in small units at a time and say, “Do I want to report this income, or do I want to legally deduct this business expense or do all of these kinds of discretionary things?” It’s much less of a case where someone just sends you a thing that says, “Your wage was $100,000, so report that.” It’s just comparatively easy, and it’s more of a case where you’re constructing this laboriously over hours and hours and deciding at what point you say, “Oh, this is good enough,” and send things in to the IRS.

Loney: This overthinking that we have around taxes does benefit the IRS in terms of the collection, holding onto the money for the period of time that they do, and the benefit that they can find at least in a short period.

Rees-Jones: It’s mostly true. The fact that people are thinking about what opportunities to pursue and eventually saying, “This is more effort than it’s worth.” That is in some ways helpful because it results in people not claiming benefits they are due. If we can get that from the right people, for example, if you could use this to get a little bit of extra income from comparatively wealthy filers, that’s generally viewed as kind of desirable.

On the other hand, there are a lot of credits and deductions where the whole reason we set them up is we really want to transfer money to someone, like the Earned Income Tax Credit, which is our main anti-poverty program. About 17% of people who are eligible for it don’t claim it. We think that’s largely because it’s so complicated to deal with that we’ve successfully dissuaded people in that process. In cases where we can save some money from whatever the “right people” means, it’s kind of good. In cases where this keeps credits out of people’s hands, where we really want them to take the credit, that’s very bad.

Regarding the timing of it, there’s another element that’s favorable to the IRS that’s not specifically about prospect theory but that interacts with it. In our current regime, most people are over-withheld, so about 75% of people get paid money back by the IRS at the end of the tax year. For the people who are doing that, that’s kind of a bad deal, in the sense of what does it mean? It means you are giving the IRS money throughout the year, and then eventually they give it back to you, the exact same amount. Normally, when you give people money and they give it back to you later, they give you interest, but not so here. This is an interest-free loan that the government is getting from a lot of people, some of whom are low-income and high-income across the board. We have lots of people loaning money to the government. The general advice I have for people is if you can avoid doing that, you should probably avoid doing that, although very precisely targeting your withholdings is a pain, so you’ve got to decide how to balance these things.

This interacts with our discussion of loss aversion, in the sense that beyond generating some interest-free loan money for the government, this phenomenon of over-withholding is also making people less motivated than they would be otherwise to avoid and evade taxes. It has a double whammy of being favorable to the government to have a lot of people in the gain frame. The only reason the government would really want to intentionally put people in the loss frame is if they wanted people to respond to these tax incentives, for example, if they wanted to motivate people to take up tax credits.

Loney: What is your takeaway from doing this research?

Rees-Jones: There’s a narrow takeaway and a much broader takeaway. The narrow takeaway is straightforward in that it looks like this gain/loss framing actually matters to people. It influences what they do. It influences how much money the government is going to get, so we should pay attention to this stuff.

The thing that’s a little bit broader, thinking far beyond this paper, but this paper is an example of it, is about the value of taking psychology seriously when thinking about tax policy. There’s a field within public finance called behavioral public finance that has really taken off in the last decade, that’s focused on doing this. What’s the value proposition of all this in saying, “Look, the way economists and economics-related policymakers think about setting the right tax is generally going to be trading off two considerations. On the one hand, we want to have tax money, because the government need to build roads and have an army and do redistribution to low-income people and do a bunch of other things.” All those things, to varying degrees, we think are valuable. We can debate how valuable, and that the weight in this hand, right?

On the other hand, getting that tax money is costly for society. It’s costly for two reasons. One is if I take a take dollar away from someone, that hurts that person, but in some ways worse than that, when I impose taxes on society, it changes what everyone is going to do. If I put a $100 tax on apples, no one eats apples anymore. Everyone starts eating bananas. That made people worse off, because people who want apples can’t have apples anymore, because they’re priced in a silly way now.

If I make labor income really highly taxed, then people might work a little less and do something else with their time rather than work, if it’s no longer as profitable. And so on. What we’re fundamentally doing in optimal tax policy is we’re balancing our need for money to fund the things that we want to fund, and our desire not to mess up decisions too badly. Economists erect really elaborate models of this to try to make it look like a very fancy exercise, and it is. But at its core, it’s just that and getting all the details right.

When we’re thinking about this component where it’s trying not to change what people do too much, that’s fundamentally about forecasting what people will choose to do. We’ve historically done that by forecasting them to be really smart, and they do whatever the maximally smart person would do, which is an OK approximation to what people do. The real value proposition here is saying, “Look, we’re getting a better grip on what psychological forces are at play when people respond to taxes.” And by doing that, we’re getting something that allows us to model what’s going to happen when we impose taxes a bit more precisely, which fundamentally affects how you do this trade-off, which fundamentally affects how you decide what taxes to use. This paper is like a very small rock in a mountain that’s being built trying to do that kind of stuff.