Originally published on Forbes.com Oct 17th, 2013
Alex Rees-Jones is an economist, a post-doctoral fellow at the National Bureau of Economic Research. The paper he just published – Loss Aversion Motivates Tax Sheltering: Evidence from U.S. Tax Returns makes an interesting point. According to his study, people who have balances due on their returns are more likely to take aggressive positions than people who are getting refunds. I’ll do my best to give you a sense of how Professor Rees-Jones reached that conclusion, but first I have to tell you, he could do some follow-up research to confirm his findings by interviewing tax preparers.
Take two people who visit their accountant on say April 10. Their life situations are absolutely identical even to the extent of the amount of their income tax for the previous year – $50,000 for each of them – except for two things. Mr. Happy has $60,000 paid in and will get a $10,000 refund that will bring his $90,000 bank account to $100,000. Mr. Sad has a $10,000 balance due which will bring his $110,000 bank account to $100,000.
In both cases, things are going pretty much as planned and expected. In principle, they should both feel the same, but experience tells me they will not. More likely than not Mr. Happy will be happy, although not nearly as happy as Mr. Sad is sad. What is really frustrating to me is that the accountant arguably did a better job for Mr. Sad, since he got to hold onto his money longer. Back in the good old days when money earned interest that was a big deal in tax planning.
I always suspected that there might be some science behind this phenomenon and similar irrationalities like refund anticipation loans and people who don’t pay off credit card balances even when they can. I even invented a name for the sub-branch of economics that explained stuff like that. I called it stupinomics. Apparently the correct term is behavioral economics.
Classical economics might be what you remember from college with all those supply and demand curves intersecting and Adam Smith’s explanation in Wealth of Nations about how the butchers and bakers make sure all of London is fed even though the butchers and bakers are selfish bastards. Classical economics is based on a model of humans called “homo economicus”:
In economics, homo economicus, or economic human, is the concept in many economic theories of humans as rational and narrowly self-interested actors who have the ability to make judgments toward their subjectively defined ends. Using these rational assessments, homo economicus attempts to maximize utility as a consumer and economic profit as a producer.
Behavioral economics takes insights from a variety of fields such as evolutionary psychology to study the ways in which actual people differ from “homo economicus”. One of the experimentally confirmed findings of behavioral economics is “loss aversion”. People have a tendency to strongly prefer avoiding losses to acquiring gains.
There is something called the “endowment effect”. It explains Mr. Happy and Mr. Sad. They will both end up with $100,000 in the bank after their taxes are settled but it bothers Mr. Sad more that his bank balance is going down than it pleases Mr. Happy that his balance is going up. Based on loss aversion, Professor Rees-Jones predicts:
… that loss aversion will motivate individuals in the loss domain to work harder to reduce taxes, and that an excess mass of tax filers will cease their sheltering efforts at zero balance due as a result of the sudden drop in perceived marginal payoff.
In other words Mr. Sad is much more likely to all of a sudden “remember” some deductible items that might be a bit on the sketchy side than Mr. Happy is.
Professor Rees-Jones worked with the IRS Statistics of Income 1979-1990 Panel of Individual returns. The dataset is the detailed numbers on the returns of randomly drawn taxpayers. “Repeated observations of the same individual over time allow direct observation of the manner in which taxpayers have modified their behavior, revealing new avoidance activities and changes in reported income”.
Now I have to confess that much of the math in the paper is beyond me so I will be oversimplifying here. Absent the “loss aversion” tendency you would expect that people’s balance due or refund status would tend to be smoothly distributed. Doctor Rees-Jones finds, however, that there is significant bunching near zero balance due and that bunching is more pronounced the higher the income. I suppose that you could attribute that to just being accurate in estimated payments. There is more.
The bunching at close to zero balance due is associated with what the Professor calls “income shocks”, the type of income events that are harder to predict
Bunching at zero was shown to be associated with having high income, with experiencing a large and positive income shock to difficult-to-forecast income sources, and with the pursuit of visible components of total tax sheltering consistent with predictions of the loss-averse model.
The most interesting policy take-away that Professor Rees-Jones has from this study is that the phenomenon of over withholding is extremely beneficial to the IRS. People expecting refunds are much less likely to do sketchy things to increase the refund than people facing a balance due might do to lower the balance due. A second policy recommendation, somewhat disturbing, is to focus audit effort on those showing close to a zero balance due. Nothing like punishing people for being right.
As I said, the math in the paper is beyond me, but thirty plus year of anecdotal experience pretty well supports the conclusion.
You can follow me on twitter @peterreillycpa.
Afternote
Another anecdotal observation here. I remember one time being utterly astounded by a return. It had a couple of W-2’s and some itemized deductions and the like. Clearly no one had intended the precise outcome, but it ended up that the withholdings were exactly the same as the tax. I showed it to a couple of others in the office who were equally amazed. The senior on the job called up the client, who spoiled it by coming up with some more deductions.
Professor Rees-Jones Offers Some Clarification
I heard from Professor Rees-Jones and I am pleased to say that he indicated that I had gotten the gist of his article. He did, however, offer a clarification:
There is one point I’d like to clarify. My results do not necessarily show that what I’m seeing is “cheating,” in the sense of illegal tax evasion. They could also be driven by legal tax avoidance. For example, you write: “In other words Mr. Sad is much more likely to all of a sudden “remember” some deductible items that might be a bit on the sketchy side than Mr. Happy is.” This is indeed a prediction of loss aversion. But loss aversion could also predict that Mr. Sad actually remembers (as opposed to “remembers”) some additional deductible items. It predicts that Mr. Sad will be more motivated to take the time to look for credits and deductions for which he might qualify, and more likely to take the time to document and report those behaviors.
The behavior I’m documenting is likely some combination of legal and illegal efforts to reduce taxes. Some of the behavior I see is sufficiently suspicious that I strongly suspect it’s evasion. However, in the interest of being scientific, it is important for me to be clear about the distinction between what the data clearly show and what I personally suspect.
In other words my use of the word “cheat” in the headline might appear to be a bit of an attempt at attention grabbing. What a thing to suspect a blogger of.