Business Income Underreporting And Public Finance

>> John Taylor: We're really happy to have Ellen McGrattan speak to us today about this very important topic, which I lost when I started to read it. It's really very important to get it going. But the title is Business Income Underreporting and Public Finance. It's joint with Bhandari and Yao, and Ellen is spending a few days here, so she's happy to talk more about it.

I looked up, you've given several presentations here. Many, many, in fact, you got your PhD here. I won't say what it was.

>> Alvin Rabushka: Don't say when.

>> John Taylor: Wasn't that long ago

>> John Taylor: And next week Bob Hall is gonna present, like I had to tell, the active role of the natural rate of unemployment, it's a good title.

But anyway, Ellen, thank you for being here, and we'll ask many questions, but go ahead.

>> Ellen McGrattan: Okay, so thanks for coming, this is actually a paper about tax evasion. It's part of a bigger project that we're going to be doing with the IRS. So that's a big issue for them.

This is a starting point, so oftentimes, people, when they have access to 15 terabytes, they tend to jump right into the deep end. We're going to start by doing kind of the Minnesota way of getting a theoretical lens. We're going to tap into a lot of publicly available data that will be useful for us to actually theorize about the topic, which is gonna be business income underreporting.

And I noticed that Jerry Houghton is online, so a lot of the data will be Jerry and co-authors have gathered, and I'll be showing it. And he can yell if I'm not true to the punchlines. But the main mission here is to set out a theoretical framework that will guide us when we do go into the microdata at the IRS.

Okay, since I'm a macroeconomist, I wanna start at the top of the trees. So let's do the macroeconomics of US public finance, Sebastian, get out your pencil. You're gonna wanna throw these numbers around at your next cocktail party. So I tried to use a year where everything was kind of round.

So net government saving in the US in 2018, so these are 2018 dollars, was a negative $1 trillion. So that's federal, local, state, everything, this is Table 3.1, one out of the BEA, for those of you not into the Survey of Current Business. So that's a big number, we bring in about 5.6 in receipts, and we're spending about 6.7.

Another $1 trillion dollars, just to, again, keep things round, is in the national accounts, we have business income that is imputed by the national accounts of $1 trillion. What is that about? Well, they use their raw data for our national accounts that we're using to study business cycles and other little, tiny things.

Has a big imputation in there, and that is that they start with the income reported to the IRS. And then they, before even doing adjustments to put it into the NAPA definitions, they first add back estimates of misreported business income. Which adds up to $1 trillion, now this is income, it's not the taxes.

So you'd have to multiply that by $0.40 or something like 4.4. But I wanna do one thing before we start getting into the taxes, which is to say all that untaxed business income, a lot of it is through the pass throughs and smaller C corporations. Now if I just did the pass-through businesses, so sole proprietors, partners, S corporations, those guys would be at 700 billion.

Okay, so they're putting on their tax forms 1.3 trillion, and they're not putting 700 billion. Okay, so for any of the macroeconomists in the audience who don't know this, you should know this, you should know where this is coming from. Now if I added small C's to that number, we'd get back towards the 1 trillion.

Of all the corporate tax evasion, they're about half of it. Okay, in my favorite language, everything relative to GDP, these numbers, we're at a net government saving of more than negative 5%, cuz our receipts are 27, current expenditures at about 33. And that income that we're adding to the accounts is 3.4%.

Okay, Pat, you already have a question, I can tell. I bet you don't know table 7.14.

>> Patrick Kehoe: Not as well as Fabrizio does. No, don't be quested, so when you do the GDMP accounting the different ways, c plus I plus g plus nx on the final product side.

And you do it on the wl plus rk, the income side.

>> Ellen McGrattan: Yeah.

>> Patrick Kehoe: Do you see this level of number and the mismatch?

>> Ellen McGrattan: No, so that stuff called corporate profit proprietors' income, which is proprietors plus partners, that will add this in. So actually, the BEA does a nice service for

 

>> Patrick Kehoe: They added c plus l plus is a little bit bigger, so we're missing something. So I'm gonna put this in to make that thing add up, no.

>> Ellen McGrattan: They actually use the data that I'm gonna talk about. So there is independent, you can have the thing called statistical discrepancy truly is, we did this one way.

We did this another way, so there's going to be like, how different are they? It's gigantic, yeah.

>> Speaker 5: Is national income that excludes capital gains and includes imputed business income, or is this Haig-Simons, or what is it?

>> Ellen McGrattan: Yeah, okay, great question and we're gonna get really into it, but for the tax nerds, it's Schedule C, E, F for many of the pictures that we're gonna work with.

So those would be sole proprietors who file a 1040, think partners and S corps. Now, the data that I'm gonna pull from Jerry's work will also include other Schedule E income. But our ultimate goal is to actually look at the business filings for the proprietors, the S corps and the partners.

 

>> Speaker 5: No retained, no retainer or-

>> Ellen McGrattan: Well, this is net income on their tax form, we can get in, I'm happy to go to the board at any time.

>> Speaker 7: So another tax note, I just wanna get the definition right. So if I take my consultant income as an LLC and not as a C corporation-

 

>> Ellen McGrattan: LLC is a legal thing, not a tax thing.

>> Speaker 7: S corporation.

>> Ellen McGrattan: Okay, S corporation, good.

>> Speaker 7: S corporation instead of a C corporation, you define the difference as underreported evasion. How do you call it? Because I call it simply a choice. It's not obvious that my consulting income should be a C corporation to be taxed at the corporate level.

 

>> Ellen McGrattan: We're gonna get very deep into this. This is top of the trees and I love the excitement of this room.

>> Ellen McGrattan: but I'm gonna get into it and then ask again if I have not been stunningly clear.

>> Speaker 8: Definitionally, untaxed pass-through income is?

>> Ellen McGrattan: Say it again.

Pass-through means, pass-through from business filings to the individual. That's what pass-through means. Okay, so this has prompted, no kidding, has prompted the IRS for many years to ask for more money. Money's been going down, down, down. And they finally got like a doubling of their budget. So that has gotten people super excited.

 

>> Alvin Rabushka: On one side.

>> Ellen McGrattan: Yeah, super excited, okay. So with the Inflation Reduction Act, they got 80 billion and there's been attempts to claw it back over 10 years. The enforcement budget would, excuse me, would roughly double. And it's gonna look like kind of a coming in, because you have to train examiners.

So it takes some time. Right now they're answering the phone more often. That's how they're using the money. But interestingly, a lot of the attention outside of the University of Minnesota is like, let's do a return on investment. So if we spent another dollar on the GS 11, what would we get?

And so there's a lot of numbers floating around about how easy is it to get money. Pump blood out of some turn up and they're trying to pump blood. So numbers are coming out. Like if you spent one more buck on an examiner, you could get $5 to $9.

That's what the CBO and JCT were predicting. Then Boning, who's at treasury, and some co-authors at Harvard, they want to target, obviously they want to look for where the money is. They wanna target the high income. That's not necessarily where the money is and we'll see that later, okay?

 

>> Speaker 5: I just want to double clarify. So that 1 trillion, you're saying that none of that has to do with definitional differences in the way that income is defined? All of it is cheating.

>> Ellen McGrattan: That is an estimate of the BEA. Well, really, it's the IRS. And I'm gonna show you how to come up with the estimate.

It's an estimate of the cheating. I'm gonna tell you where that comes from.

>> Speaker 5: That just means, the BEA is-

>> Ellen McGrattan: The BEA is the Bureau of Economic Analysis.

>> Speaker 5: You're telling me that that entire 1 trillion is cheating, not differences in the.

>> Ellen McGrattan: Correct, estimate of cheating.

And it goes, they get the IRS number in, they go tack on. Then they start doing differential things in terms of definition, don't count this twice, don't do this. Take the foreign stuff and do that. Put this on BEA. So even before they do any of those definitional changes, they tack the 1 trillion in.

That's the cheating. And let me get into it, especially since you told me you're leaving early. Okay, what is this paper about? Okay, like I said at the beginning, this is our first step before we look at anything. Try to come up with testable predictions, try to understand the world through a lens we understand.

So we're going to use publicly available IRS compliance data, a lot of which has been compiled by the man on the Zoom. And so, if you have really hard questions, I'll get him to answer it. We're gonna use the information we get from those to kind of inform ourselves about theory, like what factors do we think we need in there?

And then we're gonna do the ultimate, what I think is gonna be our ultimate thing, which is compare. We care about public finance, we care about welfare gains of doing it one way or another. Ultimately, we care about optimal design of both tax policy and administration. We are never gonna collect all the taxes owed.

So we have to know what world we're living in. And we have to understand the behavioral responses to something like the Inflation Reduction Act and the things from the JCT and Boning and others. Those are gonna be very marginal. What did they do when they got audited? We're gonna be the, let's put it on the computer, match it up to the national accounts, make sure everything fits with the NAPA, make sure everything fits with the IRS, then run the experiment, and then do the counterfactuals.

Today I will show you compare economies and steady state. Our ultimate goal is to do the real welfare analysis, which means putting in the true dynamics, which is you start today, where we are and then go. That's the ultimate goal of this project.

>> Speaker 8: This is the problem, it wants to minimize non-compliance?

That's the step you're adding to the standard PF problem.

>> Ellen McGrattan: No, the problem is gonna be ultimately maximized welfare. If we are funding new programs or programs at all, if we are funding a budget. How do you do that efficiently? We're gonna be wanting to ask those questions.

People are saying magic wands can be waved, where some money can, $12 for every $1, that sounds pretty good. Larry Summers is saying $18 for everyone. If you go for the 0.001%, you get $36 for everyone. Again, are you gonna get blood out of that turnip? We're gonna try to figure that out.

 

>> Speaker 8: But you make a big deal out of compliance. So that's the angle you're gonna explore.

>> Ellen McGrattan: Okay, this is, do you tax more or do you enforce more?

>> Speaker 8: Should you?

>> Ellen McGrattan: Yeah, ultimately, should you? Today will be, if you did those and you raise the same revenues, then I'm gonna show you what happens.

The world looks completely different. You might think, it's just a tax. A tax is a tax, no. So what's new here relative to anything else that we could find on our shelves? We wanna bring in two factors that are truly dynamic, which is we want to assume that when you get caught, you actually lose reputation, brand.

You're Martha Stewart and it's not worth as much after tax evasion. Because these businesses are getting sold and they're getting sold, those brands are getting sold. And so we wanna build in intangible assets that the businesses are building. And if I put Peter or Chad in my world, they're going to say a lot of TFP is due to that.

Not just the productivity of the business, but that which they cannot see. And that's going to have, whether we enforce more or tax more is going to affect that differently. The other thing we want to include is recovery of back taxes. A lot of theory these days is kind of extensions of the Allingham-Sandmo, which is a very static, think static portfolio, cheap, no cheat.

Return is a little bit higher for cheap, I'll do it. But we wanna do the whole dynamic thing. Of course we're gonna have that aspect in there, but we wanna make sure that we're doing the cheat and uh-oh later on. Yeah.

>> Speaker 5: Is there a distributional component to this?

Who cheats versus who pays?

>> Ellen McGrattan: Yes, yes.

>> Speaker 8: Central.

>> Ellen McGrattan: Central.

>> Speaker 5: All righty.

>> Ellen McGrattan: Okay, so why relevant? Okay, business dynamics, productivity. And in the world we work with in these kind of heterogeneous agent models, so there's gonna be heterogeneity here. There's gonna be amplification of precautionary motives, cuz you have to save up for when you get tapped for the audit.

So that'll bring another motive for precautionary savings. And what I'm going to say, my punchline at the end is, depending on which way you go, the world will look quite different distributionally. Yes.

>> Speaker 7: Here you basically have some negative externalities of enforcement, right? But there are some positive externalities.

So if I am a minority shareholder, for example, and tax enforcement at a corporate level is done better, and managers are less likely to steal the money from me. So there is an improvement in corporate governance of a higher tax enforcement.

>> Ellen McGrattan: So I'm gonna defer you. We're going to look at the model and you can repeat your question in the context of the model.

 

>> Speaker 7: Okay.

>> Ellen McGrattan: Okay, let me start, though, with the lessons we're learning from the IRS data that's already out there for all of us to see.

>> Ellen McGrattan: Okay, let me start. This gets at the, you know, what is this data? So there are two programs that have been run since the early 60s, the Tax Compliance Measurement Program and the National Research Program.

These are random audits, okay? So you go out there and you just stratify everything and you run audits. It's typically most frequently done on the 1040. And then there might be some special programs where they might look at S corps, for example, or look at partners. But the most frequent ones were the 1040.

So they would be looking on the business side at people who file a Schedule C. Those would be the proprietors. Or a Schedule E, that's where we see the partners in the S corps, or an F, if you're a farmer.

>> Speaker 8: Condition random, if you file above or below certain threshold or the universal?

 

>> Ellen McGrattan: No, so they're doing a nationally represented sample, the 1040s. There is another set of audits which go on, which are operational audits, and those are that examiners might go for the money, okay? But we're gonna be talking about the random audits. To the random audits, to get the tax gaps when people talk about $763 billion are owed, what they're typically gonna do is they'll start with the random audits, then they will tack on adjustments.

These adjustments work, I mean, there's a maximum likelihood model in the background. But let me give you the rough idea. Imagine you have a bunch of GSers, okay? There's a GS-15 who has had a lot of experience, and a GS-9 who has little. They're given stacks of things to audit, and the GS-15, with all that experience, gets a high number, and the GS-9, not so high.

Well, guess what? Take that number and give it to the GS-9 as an estimate of what he would have gotten for this kind of filing if he had enough experience. So it's really just kind of a scaling up. Now, going back to my NAPA data, for many years, they were basically, for every dollar they were using.

 

>> Speaker 5: Sorry, you're pretending like the 15 gets it on the button. So when you scaled it up.

>> Ellen McGrattan: Yeah, well-

>> Speaker 5: You take what the 15 did, and that's the truth. And you pretend everyone else had the same efficiency as the 15, and that. Okay, okay.

>> Ellen McGrattan: That would be one way to do the scaling.

There may also be we didn't get to see blah, and there was some. I mean, Jerry knows much better than I do all the other possible undetected adjustments that could make. But this is kind of the standard one. What was I gonna say before you said that? No, what was I going to say?

 

>> Speaker 5: You're doing a second bullet on the left. Use data from an examiners with large-

>> Ellen McGrattan: Yeah, yeah, yeah, yeah, yeah, but there was something else. Back to my NAPA. The NAPA was, for a long time, they were taking, for every dollar of income they were multiplying it by 3.28.

So the random audit gave them a buck and the DCE gave them 2.28 bucks.

>> Speaker 8: What is this magic number?

>> Ellen McGrattan: Okay, what is that magic number, then? By the way, when I talk about the data, when I show you the aggregates of the tax gap, it will have these two.

When I do the distributional stuff, we're gonna go straight to the raw data because they don't really have. There's a bunch of fights about to happen on that, okay, as you can imagine. All right, so how big is this tax gap? Billions of dollars. It's now, well, in the latest, these are projections for 2021.

I didn't show you every single year because that's like, too many numbers. I don't wanna Chris Simms this. So I'm going to just show you a tiny, tiny numbers and maybe even three is too many and, like, kind of put in red some things that kind of stick.

So at your cocktail party, kind of and in billions, that's what the billions are. Yeah?

>> Speaker 5: In the audits, what underlies the cheating estimate?

>> Ellen McGrattan: This is the cheating estimate.

>> Speaker 5: But it's based on the audits or something else?

>> Ellen McGrattan: Based on the audits plus the adjustments. So this is the add the two together.

 

>> Ellen McGrattan: Yeah, this is when everybody goes on the TV and goes, we've gotta do something about this. This is the number they're doing, okay? I'm sorry. That's right, that's right. I was doing income earlier just because those are the things our tribe looks at. So I like to scare you when you're doing your research.

 

>> Speaker 5: The income part was 1 trillion and this is like 76% of that is taxes.

>> Ellen McGrattan: No, no, so I was doing businesses, this is the whole thing. Okay, I'm working my way down, but this is like a headline number you might have seen. So I'm giving you info for your cocktail party, yeah, 3%.

I shouldn't even have round it, how would you guys have known? Okay, what's the main source of the gap? The main source, so there'll be three things they look for, underreporting, what is underreporting? That's when you get cash receipts, but you just don't write it down. You expense that car for your spouse, but you call it your business expense.

It's those things, that's the underreporting. The underpayment will be, there's late payments or they just haven't clawed it out of you. I had to sit in the IRS listening to examiners, when do you think you can pay this, begging for the money. And then, there's nonfiling, which is, you literally just don't send in a tax form, which actually happens.

I have some colleagues who went 15 years without filing their taxes.

>> Ellen McGrattan: So it's a thing, those are the sources. The main thing is underreporting, obviously, I'm showing you the three numbers in a row just because you're seeing kind of a steady thing over the 20 years. And all of the underreporting, the main culprit is business, that's why we're looking at business.

Wages and salaries have tons of information, there's tax withholding, there's third-party information. It's very hard to cheat on your wages and salaries, well, unless you don't file. But the business has very little informational reporting required, so that's kind of the source of most of it. Now, the other will contain things like gambling, pensions, asset income.

So it's every other line item on the form.

>> Speaker 5: So this is like when I say to my painter, if I give pay you cash, will you charge me 10% less?

>> Ellen McGrattan: Yeah, yeah.

>> Speaker 5: If he says yes, they won't give you a receipt, right?

>> Ellen McGrattan: Yeah.

>> Speaker 5: Basically.

 

>> Ellen McGrattan: That's that.

>> Speaker 5: So the implication is it's a small business, it's not big business.

>> Ellen McGrattan: Yeah. Aah, okay. So hold on. That's it. Second anecdote. Our electrical system on our car went out when we're going up to five. It was in Downey, got to a wonderful garage, but they wanted cash.

And so we went to the ATM. So you can actually look about whether ATM withdrawals are greater around some of these small businesses.

>> Ellen McGrattan: I'll let you do that one.

>> Ellen McGrattan: Okay, so how widespread is the cheating? So now, I'm gonna be showing you evidence from the NRP random audits, no DCE anything, no fudge factor.

I'm gonna look at work that Auten, who's online, and Langetieg have done. They're looking at all owners, so the Schedule C, E, Fs, and then there are nice studies from the General Accounting Office that look at the sole proprietors. So I'll show you that, both set of study, I mean, it's all based on NRP so they have the same raw data.

But both reveal that cheating is widespread, but few owners account for a lot. So I want you to imagine you're on the highway and everybody's like, yeah, they're not at 60, okay, but they're maybe at 65. And then, there's somebody whipping down the street at 100, okay? That's kind of what we're dealing with, it's the okay, now, yeah,

 

>> Speaker 5: Do they have any sources of estimating unreported income that they don't pick up in the audit, is there any other sort of check? Presumably this is like, how much they catch in an audit, right? Is there any other way of estimating how much they don't catch in an audit, is there any other gap?

 

>> Ellen McGrattan: That's called research.

>> Speaker 5: So, there's no way.

>> Ellen McGrattan: Well, I mean, that's a good question for Jerry. There may be, again, other side things like, one thing we wanna do is look at the business filings. And only if the examiner decides in the random audit to go over to the business filing will it be done.

So it's a bit, I'll let you know after I go to that pond and start figuring it out.

>> Speaker 5: Consumption data to try to back that we're missing income, so it comes through the budget constraints.

>> Ellen McGrattan: Yes, probably nobody heard what you just said, there is other way.

Yeah, use other data to try to infer cheating, yeah.

>> Alvin Rabushka: Yeah, okay, so they have a very big program of tracking down 1099s, which is done independent of audit.

>> Ellen McGrattan: Yeah, 1099 is a third-party thing, it's not a big deal thing, we'll talk later. Okay, so I wanna show you the patterns, now, first I'm gonna show you a blank graph because I wanna do the, it's like a trick to keep Pat from asking me a billion questions before I've said what it is.

 

>> Patrick Kehoe: It's okay, this is done-

>> Ellen McGrattan: They're not that good, they're not that good.

>> Ellen McGrattan: Okay, so-

>> Alvin Rabushka: Try harder.

>> Ellen McGrattan: Yeah, please try harder so on this graph, we're gonna rank by reported business income. Notice there's some negatives, right, cuz lots of businesses have losses and then the others are ranked on the x axis.

Now, notice the y axis, though you can't cuz it's so hard to read these, but on the y axis is gonna be average unreported to reported ratios. Okay, I'll use you as an example, Patrick, and I'll use Erik, both Patrick and Erik get $10,000 of, they report $10,000 of Schedule C income from some consulting, okay?

So they're on the x axis, like in the 0-40, okay? Patrick really made 100,000, so his unreported is 90 and his reported is 10. So 9 multiplied by 100, he's gonna have a high number. Erik, super honest, he wrote down 10, he made 10, he's a 0. Now, when we're averaging over these people, you know, obviously there's a mix up, but let's look to see, are there lots of Pats.

So it's a big high number or are there lots of Eriks? How does it look and how does it look across the distribution? Now, we'll look across the distribution, I'm pressing my computer like that's gonna do anything. Okay, here's the picture, and the bars are the different NRP programs.

So these are the different years. So if your eye just sees one bar, good, it's just one bar. It's kind of rainbowy, but one bar, all right? And so, what we see is, in the 0 to 40, those are people who are going to be getting, it's almost a little bit, Jerry, in fact, maybe we want to do.

If you were doing this picture again, you almost wanted to put the unreported plus reported cuz there could be some 0s. Now, I know they cut out outliers, but you could get very high numbers because a lot of these guys are going to try to get their net incomes to 0.

And in fact, in our model, we cannot make this picture cuz a lot of our guys go down to 0. They're so good at tax cheating, so we'd have to make a different picture. But the point being that these small guys are doing a lot, I mean, they may make little, but they're misreporting or underreporting a lot.

 

>> Speaker 5: You said a few of the people do most of the cheating, so how's that possible?

>> Ellen McGrattan: Yeah, I'm coming to that. We're not even, wait, hold in, we're not done with this. So let me take those guys out and look at the rest of the distribution. So losses, one thing that Auten and Langetieg found was there's a lot of cheating in the people who write on negatives, they overstate losses.

So, for example, they're putting down -100 instead of -50, okay, and that means they're gonna get a 50% ratio here on average. Now, if you look at the numbers across the distribution, remember, these are averages. So I'm gonna break down, I'm gonna look even farther under the hood.

But the averages, you can see a lot of underreporting in the middle of the distribution, okay? So it's not just, wow, the top guys are doing all this misreporting, it's all throughout, okay? Now, let me get at, so that's the wide-spreadness, let me get at the few do a lot, okay?

And so what I'm gonna do is I'm gonna take all the people 40 and above, so I don't have to deal with negatives or zeros. And let's take those guys and keep the x-axis the same, but look to see what that distribution looks like. Where are the Pats and the Eriks?

Okay, so there's the 40 up to the top, and now the height of the bars are the underreporting ratios at the 95th, you can see the 95th and 90th clearly. And the 50th, I just drew a line there because I couldn't, when I tried to be super precise off of Auten and Langetieg's thing, I couldn't get it.

So I just made it blue. And what it's telling you is, half the population are Eriks. And there are some negatives, some make a mistake and pay too much. That's rare, but I'm sure it's accidental. But what this is showing us is, it's very concentrated in the Pats.

 

>> Patrick Kehoe: Pick a number and go through.

>> Ellen McGrattan: All right, let's pick the 40 to 60, okay? The orange, the orange is the 90th percentile, those guys have an underreporting ratio at the 90th percentile of 200.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: Okay, and the ones below them have much smaller. They're like the guys, the green bars are driving a little bit fast, but not so fast, the orange and red bars, they're driving fast.

 

>> Patrick Kehoe: I see.

>> Ellen McGrattan: That's what this is telling you.

>> Patrick Kehoe: You keep saying, one second, you keep saying underreporting, if I overreport expenses, that's underreported.

>> Ellen McGrattan: Yes.

>> Patrick Kehoe: So as opposed to just flat out lying, if I said, that car, like you said, that car is only for business.

 

>> Ellen McGrattan: Yeah, yeah, yeah, yeah.

>> Patrick Kehoe: 20K, I did 2%

>> Ellen McGrattan: Cars, trucks, meals.

>> Patrick Kehoe: So a lot of it, I presume is-

>> Ellen McGrattan: Yeah.

>> Patrick Kehoe: Could be enough for your business.

>> Ellen McGrattan: Yeah, yeah, I call it consumption on the job.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: Consumption on the job.

>> Patrick Kehoe: That's much trickier to figure out, than to say, I just flat out got money I lied about.

 

>> Ellen McGrattan: Agreed, but we're gonna be able to dig into, a lot of it hides in certain places and we can use machine learning tools to get it, okay.

>> Speaker 7: How do they come with these numbers? The examiner, what they do is comparables, what do they do? How do they-

 

>> Speaker 5: Do they come to your house and ask you to show them receipts?

>> Speaker 7: They come to your house?

>> Ellen McGrattan: Yeah, these are audits.

>> Speaker 7: Okay.

>> Ellen McGrattan: Okay.

>> Alvin Rabushka: These audits are unfair. Maybe I just didn't keep a good record, I'm honest, I'm just-

>> Ellen McGrattan: These are those guys.

 

>> Alvin Rabushka: I'm not very good at keeping records, and those guys come in there and they eat me alive.

>> Ellen McGrattan: Yes. There are those, and we're gonna talk about attitudes. Let me come to the attitudes, cuz you're a category.

>> Alvin Rabushka: Up to three years or something.

>> Ellen McGrattan: No, I'm gonna get to that.

Okay, I'm running out of time, there's so much excitement, such a buzz in the room that I'm gonna-

>> Speaker 7: If I'm a Chinese restaurant owner and I get paid with Alipay in China, which happens all the time, is that reported anywhere? Could they get it?

>> Ellen McGrattan: Did you write it down?

 

>> Speaker 7: No.

>> Ellen McGrattan: Okay, then how do they get it?

>> Speaker 7: So this is an underestimate of the-

>> Ellen McGrattan: Well, yeah, so, we'll talk later about the foreign part, that's where some of the DCE adjustments they're trying to get the foreign

>> Alvin Rabushka: Cash is the same thing. Look at expenses.

 

>> Patrick Kehoe: You bought food, it had to go somewhere. I mean, there's other ways to figure it out, right?

>> Speaker 7: If I have a restaurant

>> Ellen McGrattan: I'm gonna move on. I'm gonna skip the GAO evidence, because I'll run out of time. So, one thing, I also wanna say, one cool thing, the IRS does surveys to elicit.

So we've got the guys driving just a little bit, the Eriks, a little bit fast. And everybody's like, why didn't Erik go faster? He could have gone faster. What is he doing? And they're completely confused by Eriks. So they're asking, why are the people who actually could have, they didn't have any reporting requirements, what's going on there, okay?

So they ask, they figure out their DIF scores, these are scores the IRS uses. They're the likelihood you're going to need an adjustment if you were audited. So they, knowing the DIF scores, they sent out survey questions to people who were likely non-compliant and people who are likely compliant.

And they found that there was an agreement that taxes are complicated, that's you. There was agreement that they all know the consequences of underreporting, it profess moral obligations to both of them. Where do they differ? Their trust in the IRS, whether it's fair, whether or not they're doing the right thing.

So there's a difference, mostly in the trust. Do they trust preparers? Do they trust the government? And it's not just the small guys are the low compliant and the high guys aren't. No, they find that the ones with bigger employment are actually the low compliant.

>> Speaker 8: On the other hand, if I look at the distribution of these ratios across the percentile of the reported business income, it's the 40 to 60s that have the highest ratios.

Whereas isn't there, if I have a Ferrari, don't I have a greater incentive?

>> Ellen McGrattan: Yeah, let me talk about the distribution where the money is.

>> Speaker 8: Because it's surprising.

>> Ellen McGrattan: Yeah, I think I'm gonna get at what you're asking. So one thing that Auten and Langetieg give us is they give us the NRP shares.

Like where is the underreporting occurring? In this case, they gave total income. Jerry, I would love to see this for the business guys. Just a want, there. But he has it for the total incomes. Let's have a look at where the money, the shares of the non-compliance are.

So this is reported total income, it's not the business guys. This is the share, this is just how much unreported income was in those total income reported. Now I could do adjusted, I could do just positive income, you get the same pattern. It's not at the top. The way you'll get big numbers at the top is the DCE adjustments.

So that was your earlier question about, that's where the fights will be when they do the DCE adjustments and add it. Ellen, I know you might have said it, but why?

>> Speaker 5: There's going to be a skewness if I under report, my reported income is always going to be low.

So if I'm a Pat and I'm really the richest guy, and then I underreport everything, I'm going to be on the left hand side of that picture.

>> Ellen McGrattan: Yes, so this same picture, they have this same picture for that corrected one where on the axis is corrected.

>> Speaker 5: Yeah, and it looks very similar, yes, okay.

 

>> Ellen McGrattan: I just, I'm already like going to run out of time. So here's the lessons, big tax gap, underreporting is the main source of it, business owners are the main source of that. It's widespread but concentrated. It occurs across the income distribution. And we need not just economic deterrence, there has to be something else in the model because we won't match these data.

Okay, theory. Here are the key factors we're going to build into our model, there's going to be occupational choice because we want business owners. So there's going to be paid and self employment. There are going to be taxpayer types, there's going to be an Erik, and there's going to be a Pat.

Erik is always compliant, regardless. He might be running a business and he'll be compliant. Pat, if given an opportunity, he'll take it. I think you're a good choice for those.

>> Patrick Kehoe: I'm a maximizer.

>> Ellen McGrattan: Yeah, you're a good choice for you. The non compliance source will be business income.

Underreporting will assume paid employees have total withholding. The dynamics, like I had talked about at the beginning will be, there will be loss of rep in addition to the usual dynamics of a heterogeneous agent model where you have financial assets and other things. The main additional things will be that you will lose what I call sweat capital.

You know that you've built this business and you've built your brands, your customer base. And will include a recovery of back taxes if you're caught. And you don't know, it happens in the future, so you have to have a stock ready to pay off something.

>> Speaker 5: Can I just ask a clarifying question on that in the IRS, can I pay my back taxes out of future income?

I mean, it's just going to depend how strong the precautionary.

>> Ellen McGrattan: You have to have it in, you're going to save, it's going to be in the bank. You're saving for it because you don't want to go. In the data.

>> Speaker 5: If the IRS comes to me and says, you got no money, you owe nine.

 

>> Ellen McGrattan: Jail, man. You're off to jail.

>> Speaker 5: I can't make a payment plan with them.

>> Ellen McGrattan: No, actually, I've heard the guy on the phone make the payment plan, so there are.

>> Speaker 5: Precautionary motives in the model if that's going to do a lot of.

>> Ellen McGrattan: So the question is, how examiner specific is that?

Like, did you get somebody who's like, gosh, I feel bad.

>> Speaker 8: Ellen, just real quick, an interesting extension would be to have three taxpayer types. For one, whether they comply or not depends on whether business people that they know, because then you can have some interesting.

>> Ellen McGrattan: No, I agree with you, and that is one of the things that low compliance people talk to friends and the friends also don't comply.

 

>> Patrick Kehoe: You want to put a network on top of this.

>> Ellen McGrattan: Yeah, not now. Not now. Not now. John.

>> Speaker 5: In other kinds of tax evasion avoidance, there's a capital you have to set up if you want to become a corporate, you have to set up corporation. It takes time, I'd imagine, for business tax evasion, you have to persuade your customers to start paying in cash or set things up.

So just thinking about your evasion, usually the long run elasticities are much higher than the short run because it takes time to set things up in illegal ways or semi legal ways that are pretty complicated.

>> Ellen McGrattan: Yeah, we don't.

>> Speaker 5: Delaware Corporation.

>> Ellen McGrattan: I understand, we don't have that.

The only slowness here is that you're building up your customer base, not that you're building up the cash base, which is what you have in mind.

>> Speaker 5: Yeah.

>> Speaker 7: What do you have in mind for the loss of reputation? You use the.

>> Ellen McGrattan: I'm gonna give it to you.

I'm gonna be like mathematical about it, which is your language, I think. So I'm going to get to it. Give me a minute. So, okay, you gonna choose business or work and you're going to be solving this dynamic program. So just to, I hate to do so many letters, but the state of your world is your financial asset holdings, your sweat capital.

That's the, let's call it the capital of, in your brands and your reputation. There's the back taxes, what you've accumulated in unpaid taxes, and then there's productivity in the two types of activities, self employment and paid. And then there's a whole bunch of choices which I'm going to get to as I describe the underlying pieces of this dynamic program.

I just put it there since I had the x here, I figured I better tell you. But it's going to basically be the states, the consumption choices, leisure, the factors of production, and the reporting part, the consumption on the job part.

>> Speaker 5: Does bankruptcy help in these cases or they're not available?

 

>> Ellen McGrattan: You still owe your taxes.

>> Speaker 5: Yeah.

>> Ellen McGrattan: Okay. So the continuation value. Remember you're making decisions and then what's going to. This is dynamic in a sense. You're going to then possibly get audited next period, okay? And so that pi of d, I actually did things to help me since I'm so far away.

That's the probability of the audit, and the fines that you pay will depend on it. So obviously, you got to pay what you owed back, and then there will be penalties on top of that and interest on top of that and maybe lawyer fees on top of that.

 

>> Patrick Kehoe: And so that's in fa as all of those things.

>> Ellen McGrattan: That's all in fa, yeah. Then the reputation loss, we're gonna make this for everything you see today, zero. If you are non compliant, not if you get an audit, but if you're non compliant. So if you have some d turned on, so there should probably be a comma d there, d prime.

Then we're gonna assume you lose the whole thing. And we're playing with every possible thing here. So I'm happy to discuss, but it'll be just speculative if I tell you.

>> Speaker 8: Very quickly, what's the subscript R?

>> Ellen McGrattan: Say it again.

>> Speaker 8: Subscript R.

>> Ellen McGrattan: Reputation.

>> Speaker 8: F of R, because K Prime, I thought K Prime is the future sweat.

 

>> Ellen McGrattan: Yeah, that's future sweat.

>> Speaker 8: All motion is inside K prime, what is F of R K prime?

>> Ellen McGrattan: Yeah, I'm getting there.

>> Speaker 8: Okay.

>> Ellen McGrattan: Yeah, this is a well orchestrated, planned out, everything has like.

>> Alvin Rabushka: Thanks Faith for that.

>> Ellen McGrattan: Okay, what technologies do they have available?

Okay, they make goods and services and they make new sweat, that's just building the brand. So the z, that was the productivity, they used their brands and their customers to sell stuff. And then right now we have just rented physical capital and owner time. There's one thing that needs to go in, which is going to make our aggregates fit better, which is out external labor.

So that's the next step to go in. Right now, anything I show you at the end, they just have their own time. So they're like the one guy, but we're gonna have external labor in there. The sweat investment takes their own time, so they have to divide their time between production and brand building, and then they have to pay wine and dine their clients or whatever.

All right, there's a bunch of constraints, let me just take them one at a time, because I don't wanna torture you. So there's the budget constraint, here's the next period. These are financial assets. This is the true and reported income, so the true income is whatever the sales they had, less their payments to outside capital, less expenses, and what they reported was what they got.

Plus I'll call it consumption on the job. I could have written it in cash, I could have written it as consumption on the job, they come into the mathematics the same. So we just did consume, we wrote it that way. But that CR term is like the under reporting.

So if we go back to the pictures earlier, that's the underreporting, the flow. And then we have taxes on business consumption, and in the model, on every possible thing we see taxed in the US economy, so it's all the taxes are there. We assume that goods produced by C Corps and pass throughs, there's some CES over those.

And then we assume you get transfers, there's some growth, that's the gamma, and then there's some return on assets R. Sweat capital, that's the sweat investment I showed earlier, so it's just accumulating. Obviously, that accumulation is gonna get screwed up, if you're tax audited and you're found cheating.

The back tax-

>> Speaker 5: Brand name, that's like your brand name.

>> Ellen McGrattan: Yeah, yeah, the current misreporting that's showing up in the back taxes, we allow for some forgiveness. Now, the way it works in the real world, typical returns are checked at most three years back. But if you're underreporting, they can go six, if you're a fraud, they can go infinity.

So they're somewhere between six and infinity.

>> Speaker 7: Do cheaters reduce the profitability of honest people? Does that change the volume of honest people in the economy?

>> Ellen McGrattan: You mean if there were more of these guys, would that affect you over and above prices? Everything of me is showing up through general equilibrium price.

No other competitive like, wait a minute, my guy down the road is cheating and he's gonna get me out of business. It's only through price.

>> Speaker 5: The question was, does a person change type as a result of experience?

>> Ellen McGrattan: Is that what you were asking? Cuz then I totally, no.

 

>> Speaker 5: Would have been a good question.

>> Ellen McGrattan: Thanks, okay, then there's a borrowing constraint. Normally this would be zero, but again, here's this other precautionary motive that you have to save up. If you're a cheater, you're saving up for your audit. Closing the model, I'm not going to go through these things, this is all standard.

What the workers are doing, what the C Corps are doing, and the public financing looks the same. Except now I'm gonna add in the fines on the evasion. But this part is boring and it's at the end of the slide deck if you want it. Okay, let's do qualitative predictions then I'm gonna do quantitative predictions before I lose all the teachers.

Okay, qualitative predictions, when we increase enforcement in this world, we are going to be lowering precautionary motives. Because those, all that saving you were doing was to, when the probability of audit was really low, you needed to have a bunch of stuff in the bank. So, and then it's affecting the borrowing constraints, it's making them less binding.

The lower sweat capital stocks occurs because now you're exposed to brand assets get lost when you're exposed. So what's happening when we increase enforcement? We're going to see, we take a whack at business brands. There's gonna be, business ages are gonna be much lower, there's gonna be much more churn in our economy.

And what's gonna happen is TFP gains will be through higher selection. So the Z will get better selection, but the kappa will be lower. So when I put Chad and Peter in, they might not notice, but I, the modeler would notice that there's a different, the source of those TFP is totally different if I tax versus enforce.

Yeah, John.

>> Speaker 5: The enforcement is pi d, or is it the fines that you get caught?

>> Ellen McGrattan: Enforcement is the pi d and-

>> Speaker 5: That's a different policy, that could also be, aside from taxes, there's also the punishment from being caught.

>> Ellen McGrattan: The Becker kind of thing.

>> Speaker 5: You're studying probabilities, but also the punishment might be an interesting.

 

>> Ellen McGrattan: Well, that is all into the fines, is that what you mean? Like where's punishment?

>> Speaker 5: That you could also play with.

>> Ellen McGrattan: Yes, yes, yes, don't worry, everything that you've seen so far, we're thinking about all those things, yes.

>> Speaker 7: How did you model the probability that you get caught?

Is it that, you keep drawing, eventually you will get caught. So then cheating is just transferring, I'd rather pay in the future than paying today. Yes, I think they all have in mind that they might get away with.

>> Ellen McGrattan: Yes, so we are gonna have a selection issue.

There's gonna be people who sit in there, last a long time, kind of, they're relatively unproductive. They don't have much kappa to lose. And for them, that's the economic deterrence part, what's the rate of return now versus later.

>> Speaker 7: Model? Is there a positive probability that they never get caught or not?

 

>> Ellen McGrattan: It sets, I mean, over their lifetime. Yeah, they could not get lost.

>> Speaker 7: Because it's a finite life.

>> Ellen McGrattan: I mean, if you go to life cycle, yeah, they're gonna be probabilities where you didn't get caught, cuz if it's pi equal 0.01. It happens once in 100 years.

 

>> Patrick Kehoe: It's a function of D prime. So if you don't have much, if you didn't cheat much.

>> Ellen McGrattan: Yeah, if you didn't cheat, you could get tapped and you don't get. Nothing happens, if you're Erik, he doesn't care if he's audited. He loses nothing. There's no cost, Radek?

 

>> Speaker 5: Kind of a folk wisdom is that there are various red flags that trigger audit. Like there's an excessive loss one year compared to previous years. So I wonder if this pi is also a function of maybe even pressure.

>> Ellen McGrattan: Yeah, so that would be those kinds of things, that there are these DIF scores.

Yeah. Now, right now for everything I'm gonna show you. So that's running on that. Here's another Chris Simms, that's running on the computer with that going up with the d prime. Right now, we're gonna have a constant pi, and I'll show you that. So right now we just did the constant pi cuz that's gonna be our baseline.

And then we're playing with it, it ramps up cuz we need to find some data to discipline that. And then we assume the fines are four times that. That's probably too high. It was 1 for the tax, 0.75 for the penalty. Then there's the interest, then there's the lawyers.

It's probably more like three. But anyway, I don't think that's gonna make much of a difference. And none of these numbers should be like, my God, I can't believe she used 4 and not 3. I mean, we're nowhere near precision.

>> Speaker 5: Can I just ask Sebastian's question?

>> Ellen McGrattan: Yeah.

 

>> Speaker 5: From earlier, so how do we know that the reputation cost goes to 0 if non-compliant as opposed to 0.95 or something.

>> Ellen McGrattan: We don't.

>> Speaker 5: Reputation cost.

>> Ellen McGrattan: No, that's where we're going to try to use micro data to discipline that. Well, I have no idea.

>> Speaker 5: Okay.

 

>> Ellen McGrattan: This is an extreme. Yeah. No, so what we're gonna be doing in the answer is we're gonna be like trying to understand all of the elasticities. What would we look for in the microdata? Before we even touch it. Cuz otherwise we're just gonna drown in that data, yeah.

 

>> Patrick Kehoe: The enforcement act, being caught acts like a tax on a thing you can accumulate.

>> Ellen McGrattan: Yes.

>> Patrick Kehoe: Rather than-

>> Speaker 7: Some independent of whether you accumulated that thing or not.

>> Patrick Kehoe: That is a margin that matters in the model. Whether you model it one way or another.

You chose to model it one way.

>> Ellen McGrattan: Yeah, I'm-

>> Speaker 7: Are you gonna pay a loss of $1 million as you get caught, regardless of whether you did the thing that build up sweat capital. So that's what you're building into the model. I don't mean maybe it's the right thing to do.

 

>> Ellen McGrattan: Yeah, but we should talk about that later cuz I'm only gonna be speculating. But let's talk about possible ways, yeah.

>> John Taylor: Wrap up a little bit.

>> Ellen McGrattan: Yeah, okay, John.

>> Patrick Kehoe: Why don't you hold question for a few minutes?

>> Speaker 8: Anna, to explain the model. One-

>> Speaker 8: You can postpone, but this one-

 

>> Ellen McGrattan: What I wanna do is just get to kind of the main results.

>> Speaker 8: That is before the end.

>> Ellen McGrattan: Yeah, cuz I'm gonna lose the teachers. So I'm gonna do three sets of results. Comparative statics, distributional impacts for one, comparative statics as I vary that pi. Distributional impacts for one pi.

And then I'm gonna look at my counterfactual, which is I'm gonna raise a certain amount of revenues one way versus another, taxation versus enforcement. And I'd really like to get to that last one because I think that's cool. Okay, the comparative statics, I'll give you the punchlines, as we go from say, up very, very low.

Kind of what we see, we see something between like 1 and 2 right now for a lot of these businesses, as we start to get higher and higher, you get a complete compositional shift, no surprise. But it's really like dramatic. By 7% we can compositionally shift this. We get a large drop in precautionary saving.

Okay, again, when we're at the point, at 7%, almost everybody's compliant, not that they want to be, but that they're audited to be. And so at that point, this stock of savings is not needed anymore. So we see a big change across these economies as we change the probability.

 

>> Patrick Kehoe: The compliant guys is general equilibrium effects. They're not changing what they're doing.

>> Ellen McGrattan: Yeah, compliance. Compliant guys, prices are changing on them.

>> Patrick Kehoe: It's all prices.

>> Ellen McGrattan: It's all prices.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: And they might change because of what you guys are doing. They might switch, it might be better to be paid versus up, so there's going to be.

But it's all general equilibrium for them.

>> Patrick Kehoe: Through prices.

>> Ellen McGrattan: Yeah.

>> Patrick Kehoe: They're all paying all their taxes all the time.

>> Ellen McGrattan: All the time.

>> Patrick Kehoe: The prices shift cuz the other non-compliers are shifting.

>> Ellen McGrattan: Yes.

>> Patrick Kehoe: They then react by saying what they're doing and changing it.

 

>> Ellen McGrattan: Yes.

>> Patrick Kehoe: Okay.

>> Ellen McGrattan: Yeah, yes. Okay, now there's gonna be a large increase in productivity due to selection. We're getting rid of those unproductive guys who sit on a pile of cash and just say, this is pretty good, I'll pay it this way, rather than the other way, all fine.

It's just, to them, an economic return which is higher, okay? Once the enforcement goes up, they're selected out. Now, the problem is with the enforcement, we're also smacking the business assets. It's much less, an audit is going up, now, this is extreme, I agree with you, Erik, but why not look at the extreme, right?

It's a way you're going to lose those assets. And so it's really costly. And one thing we were interested in is also to think about the very large businesses. And would we see some difference if you got up to scale and wanted to hold back? So we're still trying to figure distributionally how that will affect things.

Yeah.

>> Speaker 7: But there's no compliance. So they reduce a lot the sweat equity, there's no more of them. So what happened to overall sweat equity?

>> Ellen McGrattan: No, but everybody in business needs sweat. You need customers to operate. So everybody needs customers, yeah.

>> Speaker 5: You spend a lot of time on the data to sort of, kind of establish some facts that you wanna build in the model.

But can you show us anything that says either companies sit on a lot of financial assets to pay these fines, or that the sweat capital actually gets damaged when you get caught cheating.

>> Ellen McGrattan: Okay, we have no idea about that. What we're setting ourselves up for is looking for testable predictions like on business age, assets, all the things that we sweat capital we can see when they sell.

So this is what we're looking for. We want to get our minds trained and then say, what should we see? How should it look different?

>> Speaker 5: You get mediocre business owners who are cheating a lot? Now switching to being workers because they know the

>> Ellen McGrattan: Workers, yeah, yeah.

If they're marginal, yeah. Unless they have really low epsilon, then they stick. Okay, and then business ages are dropping, we get much younger, more churn. What was keeping those businesses was without any of the compliance stuff is you're building up your brands, getting bigger and bigger and now that's getting killed off.

Here's the age distribution for non-compliant versus compliant, they look very different. So there's a testable prediction, how does that look as we change? Okay, distributional impacts. We're looking in the model to see if we see this concentration of Pats. And we're looking to see, do we see cheating across the income distribution.

We're looking for those, that's also a prediction from the data we have. We do see, okay, so I rank owners by their under reporting. This is a model with pi equal 2%. I'm just going to have you look at the right hand column because there's way too many things on there.

We can see that a ton of the people in our model are very good at getting their income all the way down to zero, too many relative to the data. So we see that as the, in fact, they'll hold on because it is a good, the non compliant guys, they're driven by economics, they're driven by these returns.

So they're gonna get all the way to the corner, which will be counterfactual if we see it in the data. There's going to be more people in the middle. The people in the middle are too poor. They're not wealthy enough to be non compliant. They have too few assets to be non compliant.

Those are our speeding a little bit guys. In terms of the cheating across the distribution, we see it across the distribution. That can confuse people if they think all cheaters are up at the top. Because these two sets of data, these two predictions will look very different. Because if you've got middle guys cheating and you've got top guys cheating, it's not going to look like the same data as the same distribution as when we did the other ranking.

 

>> Patrick Kehoe: I don't get it. Okay. I see a 57-

>> Ellen McGrattan: So we ranked owners two ways.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: Rank them by their misreporting.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: Or under reporting. Now rank them by their income.

>> Patrick Kehoe: Yeah.

>> Ellen McGrattan: Okay, the world out there would say, that looks about the same, why is it different?

We're saying, no, no, no, that looks very different because the under reporters are throughout the income distribution. I'll leave it there, just because I'm out of time. I wanna just do my counterfactuals and then John will be happy because I can stop. So we're gonna raise revenues two ways.

Through higher enforcement, now this is a steady state, steady state. Our ultimate goal is to actually do the dynamic transition. But here's two steady states, one at the audit probability is 2% versus 5%. Another is the tax rate on business income is 40% and 47%. Those yield the same amount of revenues relative to the baseline that had 2 and 40.

Okay, here is the percent change relative to that baseline of 2 and 40. If you look just right away, you're going to see on the left side, you see like the signs are flipped. Almost in every category it looks different, except one. So I'll drive your eyes right down to the bottom.

Both exercises, we get an increase that's identical in terms of business income, the true income, but for very different reasons. These are very different economies. One economy has very young businesses with little capital and there's lots of churn. And the other economy has a bunch of non compliant guys sitting around, paying their taxes when, or audits when they have to.

So we see very different data, much of which will be observable to us. So it's these kinds of predictions where we see the same kind of aggregate outcome, but the underlying distributions are going to look very different. The underlying things on their balance sheet are going to look very different.

So this is the kind of thing that we're after is, how can we distinguish? And then the ultimate question is, well, what's better?

>> Ellen McGrattan: Our goal is not to just get a certain amount of revenues.

>> Speaker 8: Given the revenue equivalence that is baked into the exercise of the different incentives, you have to have this contravailing-

 

>> Ellen McGrattan: Yeah, but it's not baked in that we got exactly.

>> Speaker 8: It's the same.

>> Ellen McGrattan: It wasn't, yeah, that was accidental. But what's happening is, again, the productivity's coming from different sources. They're totally different, one from selection and the other from the reputation. So there's something to be discussed as to what the world would look like.

And then we want to think about if you're transiting, because one is investment and the other is just getting people misallocation, I guess, those are very different and we want to understand what we should expect.

>> Patrick Kehoe: Compliant 29 and -18?

>> Ellen McGrattan: So these-

>> Patrick Kehoe: Just give us a sentence about the economic story behind it.

Is it the changing occupations?

>> Ellen McGrattan: Yes, so exactly. You're getting rid of with the more audits. That's selection. You're getting rid of those unproductive guys.

>> Patrick Kehoe: And they go back to workers.

>> Ellen McGrattan: Yes.

>> Patrick Kehoe: I see.

>> Ellen McGrattan: They work.

>> Patrick Kehoe: I see.

>> Ellen McGrattan: Yes. John, you had a question.

 

>> Speaker 5: So you just sort of increased the compliance or the audit rates are uniformly everywhere? Use your models to figure out like posterior probabilities of being a complying guy versus non compliant guy based on the observables that they generate?

>> Ellen McGrattan: Yeah, we can do it two, there's two ways to do it.

 

>> Speaker 5: Following the enforcement.

>> Ellen McGrattan: So we've done things two ways. One, we know your true stamp and the other, did you comply or not? Right, because the people who's true stamp are non-compliant, but they're not wealthy enough to actually cheat, you would want to know them, as well as, they look compliant on paper, but they're not.

Is that what you're getting at?

>> Speaker 5: Well, I'm saying that you have all these things that people do differently if they're compliers versus non-compliers, that then there's different age distributions, income distributions, whatever. And so you can use that information that they throw off to potentially sort of back out the probability of being a complier versus non complier.

 

>> Ellen McGrattan: Exactly.

>> Speaker 5: And target your enforcement, specifically, to the guys who look like non-compliers.

>> Ellen McGrattan: Exactly.

>> Speaker 5: And that might be a different kind of policy that has a better.

>> Ellen McGrattan: And this model could be used to try those kind of tax administration. See, the IRS refuses to do experimentation on people, but we can.

That's the beautiful thing about computer codes. We just can try to experiment on the people, and it would be to come up with, yes, clever ways to target it. So let me just, what am I doing? I keep hitting my computer. Let me just bottom line it. The higher enforcement versus taxation, most evident in the composition of the owners and the businesses themselves, not evident in the aggregate business income.

But we want to do the transitional dynamics and so that's our next step. Our next step is to get our hands dirty using this information that we've gleaned, using the studies that have already been done. We don't want to reinvent wheels at all. And then the theory, we need a ton of things, including the most important, I would say, is the transitional dynamics.

And we could then really run an inflation reduction act.

>> John Taylor: So Alvin has a question.

>> Alvin Rabushka: Okay, I want to bring just a little bit of history to this. In the early 1980s, there was already a concern about tax compliance, and the estimate was about 1981, 82. The tax gap was about $100 billion.

90% of it was the legal sector tax gap, and 10% was criminals, drugs, prostitution, gambling, robbery, whatever. So the American Bar Association established a commission on taxpayer compliance. And the easy way for you to read it is just to Google. American Bar Association, Commission on Taxpayer Compliance, 1987.

It's on JSTOR, you can get it on charge. Okay, so they appointed nine or ten people to this commission, and we had two former IRS commissioners. We had three academics, I'm the only one still alive. We had the leaders of some of the biggest, biggest tax law firms in the country and some corporations, and I'm the only one alive of the entire commission.

And so we had a staff that went to do all the academic research, gather every article they could find. And that was the time when the IRS was willing to cooperate more with academics, so this seemed to be an ideal time. So we met for a couple of years and then put out this report.

And I want you to understand, there's no computers, there's Smith Marchant calculators, there's no phones, there's no modeling, there's just some data and some people. Okay, so the commission reported there were two things that seemed to matter. One was tax rates. I don't hear any discussion about tax rates.

So they felt that lowering tax rates might improve compliance, because, after all, the difference between 50% and 20% is how much you get to keep, and then the risk reward ratio changes. So the commission then concluded that since the 1986 Tax Reform Act had passed, maybe we'll see in a few years if that works.

No, the first Bush and then Clinton got it all the way back up to 39%, so that never materialized, so the first proposition could never get tested. Then the second had to do with information collection and reporting. And so there was some question of how to improve compliance through collecting information.

Well, third party reporting was terrible. And cash under $600 is not taxable. So you see, Patrick quit his job as an academic, and he went into full time consulting, and he had ten projects a day from ten different customers, and he only took $500 a shot from each in cash.

No way to possibly add it up, compile it, test it, evaluate it. And he never wrote off anything remotely approaching his cash take. So he could write off 200, but then sometimes he wouldn't write off anything. Why bother? Okay, so that went on a lot, and that goes on a lot today.

A lot of people live on campus, some of you pay your gardeners in cash, and that doesn't get reported. But it's actually a lot more widespread. You could look at the Bureau of Printing and Engraving and forget the hundreds, look at the twenties. Because bags of twenties turn out to be convenient.

Hundreds become problematic, now with 9/11, FAFSA, everything else, so the casting becomes very important. But the problem is that setting up all of these things hasn't fixed the problem, as you rightly point out. And I'm not gonna spend 20 minutes, I could spend an hour talking about this stuff, but the point is that you're quite right about all of that.

I want to give you one example of something to illustrate this problem. The government puts a cap on the amount of money you as a professor can contribute to your retirement account, right? But if you also have a consulting income and you can take Keogh retirement, basic retirement contributions, okay, and the combination of the two go over the maximum, they didn't check.

From 1981 up to about, I don't know, a bit around 2000, I discovered this. And then it turns out TurboTax didn't have a provision for this. So I thought to myself, how many enemies can I make if I tell TurboTax, you guys are missing the entire academic community of consultants?

It may be just 2, 3%, but all of us were not audited, and we both took the maximum because it was automatic thing to do, but TurboTax never put the two lines in the 500 forms together. And I realized if I reported this, the IRS would probably throw me in jail because they don't want to admit they failed for 20 years.

Okay, now, this stuff is huge. I've talked to John about doing a seminar in the spring about tax shelters, I'm a world expert on tax shelters. I've been to a dozen countries that are tax havens where you don't pay income tax, sales tax, and how you hide money.

And a lot of the money that is untaxed is income that goes through offshore tax havens, and that's really hard to trace. I mean, to give you an example, in the Virgin Islands, US, British Virgin Islands, there are 375,000 offshore registered accounts, 40,000 from China. So I'm gonna stop at this point, and anyway, I just want to indicate that we've done all this stuff before.

And so what you said is that we haven't had that much great success, so this thing takes the IRS to task a little bit. And it also says, be nice and friendly, cuz when you're one of those people who get caught in TCM audit, they put your life through hell and you didn't do anything to deserve it.

And so this stuff goes around and then bad IRS publicity. And so Republicans make a living off this stuff.

>> John Taylor: Okay, what do you say?

>> Alvin Rabushka: Okay, so anyway, that's it.

>> Ellen McGrattan: Yeah.

>> Patrick Kehoe: Your model has a prediction, as it stands, if you jack up the tax rate, the maximizers, you call them the non complex, make the rational decision.

And I just mean the rational people in the world, based on that, will automatically cheat less. So that's all that factored in. It's already in there.

>> Ellen McGrattan: It's already in there.

>> Speaker 5: I think the last fact you put up, I keep coming back to facts, was the survey that trust predicts compliance, right?

That's what it was, right?

>> Ellen McGrattan: The biggest difference between the people who had low and high DIF scores for the amount of trust in the government, right?

>> Speaker 5: So the question is, does more compliance create more trust? Maybe that's a network and it's a modeling exercise, but I think everyone's cheating.

 

>> Ellen McGrattan: Yeah, these people also, they also are more local. In other words, say their businesses are more local. Their friend, they said, do you talk, how much networking locally are you doing your business? So these are sole proprietors, right?

>> Speaker 5: The fact of stronger compliance, that you don't have it all in the model right now.

 

>> Ellen McGrattan: And your customers and everybody's cheating. You're like, well, it's an unfair system, everybody's cheating, I'm gonna cheat.

>> Alvin Rabushka: Correct, but if you saw your neighbor kill somebody with a car, a child, you'd report that, but you won't report your neighbor for cheating. That's, I think, a big difference in the way people think about these things.

 

>> Speaker 5: The given tax budget of chasing people down, if twice as many people are cheating, you have half the chance of getting caught. Cuz it's just double the numbers, same money chasing you. That's just more rationality factors-

>> Alvin Rabushka: That's weird.

>> Speaker 8: The incentive to cheat being affected by the tax rate.

Alvin was saying that we didn't have enough time series variation, but we certainly have crossed state variation because different states have different income taxes. So that's something you might look at in the data.

>> Ellen McGrattan: Yeah, and we can see their geography, yeah. I saw Alessandra over there and thinking about kinda culture.

And it struck me because there are so many immigrants in the country, does their propensity to cheat on taxes depend on how much tax evasion there was in their home country?

>> Speaker 8: Can I talk about it very quickly? I mean, in terms of dynamic efficiency of an economy and the occupational choice impact of this tax policy.

So is there any scope for trading off? I mean if I care about scope, if I care about the entrepreneurial sector, the business owners in the economy, is there a trade off there, I may wanna get them? Yeah, we're a little bit of a break because I still want this.

If these guys are doing the economy-

>> Ellen McGrattan: You're gonna worry about growth and you're gonna worry about,-. I mean, I'll do the Peter and Chan, if you think that they're a big source of growth and investment, then you may worry about this.

>> Speaker 8: The planner there may have wanted to give them a break.

>> Ellen McGrattan: Yeah, that's what we need to sort out cuz it's gonna look different.