Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable. ======================================================================== # The Impact of Third-Party Reporting on Tax Compliance: Evidence from Gig Workers Authors: Andrew Garin, Emilie Jackson, Dmitri K. Koustas, Alicia Miller Discussant: Susan N. Houseman Video: https://www.youtube.com/watch?v=aXOe_NS3iVo&t=11504s ## Talk (03:11:44 – 03:33:29) [03:11:44] [applause] Uh Dimmitri Kustas is our next speaker. [03:11:55] Again, speaking on behalf of a team. Okay, thank you very much. [03:12:40] Uh thanks to the organizers for including us on this really excellent program. Uh this is joint work with Andy Garin who's here, Emily Jackson, now at Ruckers and Alicia Miller at the IRS and the disclaimer applies that these are [03:12:54] our views and the results I'm going to show you today using tax data have been reviewed for uh to ensure no confidential information. [03:13:02] So, like the previous paper, we're going to be studying self-employed workers in the gig economy using IRS information, 10 returns, 1099s. We now have a a body of work uh Andy and I and uh Emily and [03:13:16] some at the IRS looking at this population over time. Our first paper uh from 2019 with Mark Payne and Brett Collins, we were looking at 1099 data back to 2000 through 2016. And this is a [03:13:29] period of time where there's really no changes in uh the thresholds or the rules and we were trying to answer the question of whether gig work was replacing traditional employment. So looking at uh this long period of time and one of our key findings from that paper is that there is a source of [03:13:44] growth appearing in the 2010s and that main source of growth is coming from new online platforms. Uh these are now household names. I'll I'll talk more about what I mean by that in in a few slides. among this new online platform [03:13:57] work uh it looked pretty different from other type of independent contracting 80% was secondary to W2 work whereas it's about 60% for other types of uh independent contracting and these platform gig workers they're [03:14:12] predominantly young and engaging with self-employment largely for their first time and there's associated challenges with that dealing with a tax system that can sometimes be complicated now there are challenges Martha mentioned some challenges allges and one [03:14:26] challenge for measuring uh the this population using 1099s is that this coverage is going to change over time particularly going forward uh from 2016 and I'm going to be talking in [03:14:39] particular about uh issues with the 1099K where some of this gig platform work is being measured. Um so I'll talk more about the 1099K. It has a higher reporting threshold than the 1099 NEC. [03:14:52] And then more recently uh there have been other important changes under uh the ARP and then most recently under uh OB3. [03:15:02] So our key research questions in this paper are how does receiving an information return or not impact what individuals are going to be reporting on their tax filings? Particularly how this is going to show up as as a schedule C [03:15:15] or schedule SC self-employment tax filing. Why does this matter? Well, it matters if you're trying to measure self-employment using these self-employment tax filings. [03:15:24] We also are trying to answer the implications for measurement of gig work using tax data and particularly the 1099s. Uh a few years ago before everyone was concerned that AI was going to take our jobs, we were also we were concerned about gig work taking our [03:15:38] jobs. Um and this is tying into the um technology agenda. I mean there has been a very transformative change here with these platform gig companies. They have certainly improved matching efficiency. [03:15:51] Uh I'm old enough to remember DC before Uber and it was uh before rid share companies and it it was hard to get around. Um so it certainly has produced consumer surplus through technology and uh the the complication is this these [03:16:06] platforms are largely relying on this independent contractor model and it's raising concerns about worker welfare in in the gig economy and the very first step before you answer any of those questions is we need to have accurate measures of the number of platform [03:16:20] workers or understand where the source of growth is coming from in independent contracting. So what are we going to do in this paper? Uh while there is this issue of missing federal data, 1099K gap I'll talk about, we're able to fill in [03:16:34] that missing data in two states that have lower 1099K thresholds. We're going to use a border design to get causal identification of the effect of receiving a 1099 uh versus a counterfactual where you don't receive a [03:16:48] 1099 uh because you're because you're in falling into this 1099K gap and to look at the causal effects on tax filing. And then we'll come back and quantify how aggregate measurement is impacted by [03:17:01] changes in federal 1099 reporting. and I'll make some uh some um extrapolations to OB3. [03:17:09] So the outline of the talk, I'll tell you a little bit more about data in background. I'll tell you about our our design here using the state data to estimate the effects of the 1099K gap and then we'll come back and look at these aggregate trends to try to fill in [03:17:22] what's missing. Um so just quickly since we got some of this already, if you are an employee of a firm, you receive a W2. [03:17:32] If you're an independent contractor, if a firm pays you for non-employee work, you can receive a today, you would receive a 1099 NEC. Prior to 2020, this was reported on 1099 MIS box 7, but that [03:17:45] there's no change there in reporting requirements. It's just uh showing up in a different place or since 2011, you could get a 1099K if you're a payment processor and some platforms fashion themselves as payment processors. uh [03:17:59] that earnings the the gross earnings could show up on a 1099K. [03:18:04] So these are uh reported by firms. That's why these are solid lines here. [03:18:09] And then what ends up on an income tax return? Well, not necessarily. These are what the dotted lines imply. So if you're a someone who doesn't have a a tax obligation as a W2 worker, you may not have a tax filing obligation and you [03:18:23] may not file a 1040. and that's okay because the payroll taxes are already automatically taken out on your W2. Um, that's not the case if you're an independent contractor. You're getting gross earnings. Now, there's some uh [03:18:36] maybe ambiguity on whether you have to file a schedule C if you don't have if you don't meet the filing requirements, which is $433 in profits for the payroll tax that you would show on a schedule C and then also file a schedule SC. [03:18:50] There's some ambiguity there. Uh and then there's also a large document documented uh literature on non-compliance and self-employment. And so there's breakdowns here. Uh and then there's this other group who are non-independent contractors, consumerf [03:19:03] facing, self-employed, and they would uh they would file you would not see an independent you would not see uh any type of information return. And you would only see them potentially in the tax universe if they're filing uh if if [03:19:14] if they're filing a a tax form. Um in other work we've shown that on average conditional having a 1099 uh about 55% in recent years are uh ending end up [03:19:26] filing a schedule C. I talked about this distinction of a 1099 NEC or 1099K. Well, it's kind of largely up to the it's kind of a choice [03:19:39] variable of the platform. one can uh only uh make guesses about why some platforms would choose a K or or an NEC. [03:19:47] Uh but you can see uh in particular there's uh variation across uh industries and types of tasks. One note this is prepared using entirely public sources. Nothing in the tax data uh did [03:20:00] came came here and there's no I'm not saying that any of these companies are or are not in our subsequent analysis. [03:20:07] Um let's focus on ride hailing and delivery which we've shown in previous work to be uh a large share 90 approximately 90% of the growth that's occurring among platform the platform gig economy. Uh the main ride share [03:20:21] companies they're using the 1099K and an NEC but for measuring for for reporting different things. So the payments from passengers are being reported on the K. If the drivers have any types of bonuses, they might get a 1099 NEC. [03:20:36] And then for the most part, with the exception of one large ride share platform that's also a delivery platform, uh the delivery companies are using the NEC. [03:20:46] So this is important for measuring this activity during COVID and I'll come back to that at the end. [03:20:55] These two forms have different ti different statutory minimum requirements. So the 1099 MISK box 7 which became the 1099 NC that's been around for a long time and that's had a [03:21:07] $600 threshold until recently. The 1099K was introduced in 2011 and [clears throat] it has a it was originally designed for processing of credit card transactions. As a result, that was the motivation for setting a [03:21:22] higher threshold of $20,000. Uh, under ARP, the American Rescue Plan is was supposed to come down to $600. It never did. Uh, eventually was we started a [03:21:35] phase implementation of reducing the K down to 5,000. Was supposed to go to 2500 in 2025 and 600 by 2026. Uh, that did not happen. OB3 again raise the [03:21:48] threshold for the 1099K back to $20,000 as well as increasing the threshold for the 1099 NEC to $2,000 and it will be adjusted for inflation moving forward. [03:22:03] So if you've been paying attention, there's something really interesting going on in this figure on the left. So this is the 1099K distribution for platform gig workers in 2016. [03:22:15] And what we see here is there's actually a lot of mass below $20,000. In fact, there's no structural break at $20,000 at all. So what happened, what we think happened is that platforms started to use the 1099K, [03:22:29] but were reporting at lower thresholds. They were kind of experimenting with using the 1099K, but didn't want to run a foul of any types of uh requirements on reporting if this was potentially should have been reported on a different tax form. And so we see that we see the [03:22:43] full distribution here in 2016. Starting in 2017 and by 2018 it's it's pretty much complete. There's uh essentially no platforms that are reporting under $20,000 by then. Okay. So this all this [03:22:56] mass has disappeared by 2018. Massachusetts and Vermont in particular were on the forefront of realizing that there is this gap in these different thresholds. And so they passed early [03:23:09] laws effective January 1st, 2017 that lowered the 1099K thresholds uh in those states for reporting in those states for people who who work in those states. uh five other states including neighbors [03:23:23] here Virginia and Maryland have uh also lowered their thresholds down over [clears throat] time. Um but that but that came later um effective in 2020 and 2021. [03:23:37] So what we do in this paper through a data sharing agreement uh at IRS uh we were able to uh acquire the Massachusetts and Vermont state data for uh the purposes of the research in this paper. And so we're able to combine that [03:23:51] with the federal data basically to see the activity that's missing. And uh so here is the full distribution once we also add in the what we see in the state returns. And there's a couple of interesting two interesting takeaways [03:24:03] here. So one remember you saw no uh mass below 20,000 in this figure. Well, I was excluding Massachusetts and Vermont. So if you look in Massachusetts and Vermont, there is some mass. So, one way that firms complied with the state laws [03:24:18] was to issue a federal 1099K to individuals in those states in Massachusetts and Vermont. Uh, but not all platforms. Some platforms only issued a state 1099K and not a federal one. And so, you need both of these [03:24:33] pieces to see that full distribution uh without the without the missing mass. Um but but the state laws did lead to some increase in uh federal 1099Ks in those states. [03:24:46] We're next going to explore this question is how does receiving a 1099 affect tax reporting? And for identification here we're going to use a border design or slight innovation on that border design. If you know anything about Massachusetts, I'm from [03:25:00] Massachusetts originally. We have really big counties uh and they go kind of from across the whole state. Uh so the one twist here we're doing on the borders and we also do this in our paper on pandemic unemployment assistance is to restrict to zip codes zip code zip code [03:25:14] centroidids within a certain distance bandwidth from the border spatial bandwidth here we're going to use 15 miles from the border so these are the counties that are in our analysis we're not including Boston here but we're [03:25:26] including uh border counties uh border essentially these border zip codes and so the [clears throat] idea here uh So, we're going to focus on people because we don't see we wouldn't see uh new entrance in these other states. [03:25:40] We're going to focus on people we knew were gig workers in 2016 when we have complete coverage uh all all of this reporting on the 1099K and then we're going to follow this population forward. [03:25:53] So, one limitation here is we're not our estimates are not going to include new entrance. The treated side is the Massachusetts gig workers who continue to receive a 1099K. And the control are these uh border states, Connecticut, New [03:26:06] York, New Hampshire, and Rhode Island. And so our specification is uh is pretty standard here. [03:26:13] [clears throat] Um we'll uh I'll just go through these outcomes. So here is the first stage. Uh we can show this in an event study framework. So there's no pre-trends. Once we have, you know, Massachusetts may differ, Boston may differ from New Hampshire, but once we [03:26:27] focus on these border counties, it looks like we've uh established a good counterfactual. So we don't see any pre-trends for people on these border counties uh in these border county areas. And then after uh 2017 in [03:26:39] Massachusetts, you're there's two things going on here. There's still people in the control who are getting 1099s if they're above the threshold or had an NEC. Um, and there's also attrition away outside out of gig work, but there's about a 20 percentage point higher [03:26:54] probability of getting a 1099 at all if you're on the Massachusetts side. It falls a little bit by 2018. And on average, this is about $2,000 more in uh in gross earnings being reported on a 1099. We can next look at reduced form [03:27:08] for filing a schedule C. It's noisy. seems to be increasing by 2018 and on average is about $400 more in profits being reported on a schedule C. This is the reduced form though. We have to scale it by that probability of getting [03:27:23] a the difference in getting a 1099. And so we do that with an IV specification. [03:27:27] So let me walk through our our main estimates here. Um, so what this estimate says is that uh getting an additional dollar reported on a 1099 leads you to report 28 cents more in [03:27:40] receipts on a schedule C. You also report higher expenses and in the end you're reporting 17 cents more in profits. [03:27:50] [clears throat] We see a positive point estimate here on filing a schedule C, although it's noisy. Uh, but we see a a really big and statistically significant increase in filing a schedule SC, 12.3 percentage points higher if you're on the [03:28:03] Massachusetts side of the border and getting the 1099. [03:28:08] Um, so I think I went through these results. I'll move on in the interest of time. The next step we do is try to understand what this means for the aggregate 1099 K gap. Um so we estimate so the black line here is if you look at [03:28:22] uh our measure of of gig labor platforms either measured via 1099 NEC or a 1099K you see this flattening. This is precisely because of the 1099K gap. We can then estimate if the rest of the country growth in the rest of the [03:28:36] country followed the trends in Massachusetts and Vermont. Here is the number of workers that are missing because of the 1099K gap. It's about 770,000 workers. If we multiply that from the estimates from our border design, it implies uh about half a [03:28:49] billion dollars, 560 million is under reportported in profits nationally due to the 1099K gap. Many of these people are not don't have high tax liab federal tax liability. So the implications for taxes is uh that there's about 43 [03:29:04] million in revenues lost. Uh the number though for SECA collections is about twice that. uh that that is now not being reported because of the 1099K gap about 85 million. Now this is through 2018 might be wondering what has [03:29:17] happened since then and so this is from our our paper forthcoming in the CRW volume that that Sue Hman is editing. Um so here you can see that it was flat because the 1099Ks are going down but at the same time the 1099 NEC's are going [03:29:31] up and this comes back to what I mentioned earlier that the delivery companies are largely reporting on the 1099 NEC's and so there is an explosion in gig work platform gig work in co driven by delivery [03:29:45] we all know what the dynamics were uh during co and why that's the case and so we as of our latest numbers have five uh have o over five and a half million in uh gig workers. Much of that is coming [03:29:57] from measurement uh on the NEC. Although in 2024, we also see more activity on the 1099K from the reduction in the 1099K threshold. Now, let me talk in my last few minutes about the [03:30:09] implication for the changes under OB3. There were two important changes. One is again an increase in the 1099K threshold from it had come down to 5,000 was going back up to 20,000. If we look in 2024, how many workers that was and how much [03:30:24] how many payments were being reported, there were 844, 1099 KS, that's not exactly workers if somebody is getting multiple of them. Um so, so it's a bit of an upper bound u but 8.8 billion in payments that is now [03:30:37] going to uh no longer be uh reported because of the increase in the threshold. For the NEC, there's two populations. There's one, the platform gig population that I've been talking about and focusing on in this paper. Uh [03:30:49] and so there were uh 2.8 million NECs from platform gig companies in this range and 3.2 billion in payments. But also more broadly, if we look beyond just platform gig work, there's 11.3 million total NECs in this $600 to [03:31:02] $2,000 range and 12.9 billion of payments. Um, so if you take our estimates of pass through and apply them uh to the these numbers, it's going to imply 3.5 billion less in profits reported to the IRS and then a [03:31:17] corresponding impact on SECA and federal tax revenues uh about 350 billion less in in uh 350 million less in in uh in federal taxes in in income taxes and uh [03:31:30] twice that much for Sika. All right. So to conclude uh in this paper we talk about the 1099K gap which really starts to bind beginning in 2017 when there's widespread reporting of gig [03:31:43] work on the 1099K becomes subject because platforms change the way they report to this higher threshold. we combine federal and and state tax data and leverage a border design uh to estimate that the K gap as of 2018 was [03:31:57] about 770,000 individuals uh and that this led to uh 560 million in unreported profits. Now other states have since lowered their thresholds. Uh if there are people in the room or who have friends in this room from these other states, we'd we'd love to talk. [03:32:12] Illinois, Maryland, Virginia, California. California is interesting because it only did it for for platform gig work, not for the uh 1090 for independent contractors more broadly and and New Jersey. I mentioned that OB3 [03:32:27] further changes this landscape and it's going to affect 12.2 million independent contractors or that those are the numbers uh as of using 2024. [03:32:36] Um there's other implications here for measurement. I'll just maybe in my last 30 seconds advertise a new project that Katherine Abraham, Andy and I, and John Rothbomb are doing over at Census. Our goal is to continue this measurement [03:32:48] work. Our goal is to link think about how this is being reported on uh on household surveys as well as to uh better understand how gig income fits into uh the total house fits into total household income and measurements for [03:33:03] poverty. So one implication for measurement is if we look pre207 uh or if we look if we want the most complete uh year since then it would be 2024 when the threshold is lower if we want to compare across across two years. [03:33:17] Uh so that certainly we're going to be facing these challenges uh moving forward. So looking forward to uh Sue's comments as well as the general discussion. Thank you. ## Discussion — Susan N. Houseman (03:33:29 – 03:46:34) *Shared discussion for Session 5: The Nonemployee Workforce; the discussant covers every paper in the session.* [03:33:29] [applause] The discussant will be Susan Houseman. [03:33:44] >> Great. Thank you. Well, while we are um setting up, I I just want to uh say I'm really excited to be discussing some papers on uh self-employment or non-employee employment. It's a really [03:33:57] undeveloped area in statistics and I think uh you know needs a lot of work and I'm very excited to as I mentioned uh you know see increasing amount of work in this area. [03:34:33] Great. Thank you. Well, let me start off with the first paper which is on the non uh uh non-employee workforce that Martha presented with many co-authors. So, I think this is an incredibly important new re uh resource for research and [03:34:48] policy. Um their major shortcoming of the LEHD and related products is is that the analysis is limited to workers uh in employee W2 employment arrangements and the new data is going to use uh [03:35:02] information from tax filings to help in for self-employed businesses and the workers um uh whom we might call self-employed or non-employees. [03:35:12] uh for many self-employed will be able to link to demog uh graphic information and to the businesses using their services. So we'll get that that link that you have nice link that you have uh with the LHD along with the characteristics of those businesses. [03:35:25] It'll enable insights into workers holding both employee and non-employee jobs arrangements within a year and enable us to see some really interesting uh analyses of transitions between and across employee and non-employee [03:35:40] statuses. Um, and I want to underscore the different types. This will be available for the different types of non-employee business owner arrangements across years as Martha described and I won't go into that. So the one thing I [03:35:54] do want to flag is is that this new um, you know data product will help capture non-employees who are not business owners. So um there is a a ma this is a major gap in standard measures of [03:36:07] self-employment which focus on business ownership. Um the authors called it a blind spot in administrative and surveybased measures of self-employment. [03:36:18] Um in this paper these workers are identified as receiving 1099 NEC's but do not file a schedule C. Um they are termed latent businesses. Um I you know [03:36:31] Martha mentioned that she got push back on calling everybody business owners and I might suggest that they drop the latent business owner um uh language. [03:36:40] These are people who are probably most likely working uh as independent contractors for businesses and don't think of themselves as business owners and probably will never set it up. [03:36:51] Analysis of characteristics shows that these workers are younger, disproportionately black. Um, I'm going to kind of start my remarks here by mentioning some shortcomings of the [03:37:04] product. Um, I don't mean this to be a criticism, but rather I think it's important to recognize what it does and what it's not doing. Obviously, it's going to miss some people who don't receive any uh 1099. [03:37:17] That may be uh because these individuals receive cash payments and are part of the underground economy. Of course, you're not going to uh catch those, but they may also be paid through a third-party payer like PayPal, Stripe, Venmo, or large platform companies like [03:37:31] Uber, Lift, Upwork that issue 1099Ks. And as Dimmitri explained, for the most part in most states, um the threshold for the 1099K is much higher, $20,000. [03:37:45] And if they're earning less, they're not going to be captured here. uh the current analysis also misses anyone who receives a 1099 payment even even if it's above the threshold and a key question is how big is this phenomena in [03:37:59] cases also I want to point out where third party payers issue the 1099s we can't link the worker to the business sector that's kind of a really interesting part of the analysis that's [03:38:12] coming out here to the extent that businesses are increasingly paying people through uh you know, systems like PayPal, we're just going to miss that link. We won't know. We'll capture, we may capture them in the tax, we will capture them in the tax state, but we [03:38:27] may not know who they're working for. I want to quickly note that there's a comparable uh problem on the employee side. When employers hire workers through a contract company, they may be [03:38:39] a employee of the contract company um and they're paid a W2, but we don't know where their labor services are actually provided. We don't know where they're actually working. The prototypical [03:38:51] example of this is staffing companies um like temporary health agencies. So uh another sharp claim is the data frequency is annual which may be inadequate for some purposes uh insufficient for much policym. We [03:39:06] actually can't tell the paper uses the term moonlighting. I'd say we actually can't tell if people are moonlighting because we don't know the sequences of jobs. Uh it may be uh you know a main job for some period of time. Studying [03:39:20] dynamics is limited. it misses income data for those who receive 1099s but who don't schedule the 1090C um and uh demographic characteristics are available and administrative data are far less than what could be [03:39:34] collected in surveys so many possible uses in the interest of time I'm going to skip through this um you can think of them yourselves but um I want to mention that the findings about non-employees [03:39:48] who are not business owners underscores the need to improve worker class questions in government household surveys. So, uh for now, as others, uh Brett and others have have mentioned today, household survey is an important [03:40:01] complement to administrative data on non-employee workforce. They potentially capture those who don't file taxes, uh provide more demographic information, and are often collected with shorter [03:40:13] reference periods. Um the paper mentions uh a a paper by um uh Katherine John and others um that compares the uh tax data with non-employee um uh miss uh [03:40:27] non-employee sorry tax data with the CPS showing that they're missing a lot of uh non-employee employment. Um, but I think we have a pretty good idea of uh the problems with workerclass questions in [03:40:41] surveys like the CPS and uh some work that uh Katherine Abraham, myself and other colleagues at UpJon have been doing have been kind of testing possible questions with the Gallup organization. [03:40:53] One of them is ongoing and we think we're we're you know capturing a lot of these people with better questions. Um incidentally these people like what um Martha and co-authors found uh were less [03:41:06] advantaged you know non-white younger uh less educated and the like. So um uh finally on this paper I just want to mention that the uh NLSY cohort the new cohort I understand will be improving [03:41:20] those questions so they can be uh integrated into uh uh federal statistics surveys. So moving on to Dimmitri Andes and and their co-author's paper. Um it's a broad uh the it ties into the broader [03:41:35] literature on tax evasion uh by examining the effects of receiving a 1099 on filing of schedule C in federal tax returns. Um and exploits exogenous uh cross-state variation in the receipt [03:41:48] of 1099 and whether workers file A and B report how much they report in expenses and profits. [03:41:56] So um as Demetri is uh has explained uh the filing of 109C uh uh sometimes uh is issued by the in sorry I'm a bit ahead of myself but but [03:42:11] it provides really convincing causal evidence that this is having an effect on information returns. [03:42:19] So the empirical setup is uh such that um uh Dimmitri explained it and again is as in the interest of time I'm not going to go through the results but they find a small but insignificant effect on uh [03:42:34] filing of uh schedule C and significant effects on the reporting of receipts and profits. I want to close by saying by talking a bit about the interpretation of these findings. I think the causal [03:42:46] effects are clear. uh interpretations a bit uh a less a little less. So on the surface the finding the first finding in particular is a bit puzzling. Why is it affecting people who file? Um I think [03:42:59] you know they they kind of outline some some possible uh uh uh possible mechanisms. One is that the controls those are residents outside of Massachusetts don't behave as expected. [03:43:11] Um the analysis sample only includes incumbent gig workers. they're used to filing schedule C and they continue to do so even when they stop receiving them. The other another one is a treatments those insights uh [03:43:23] Massachusetts don't respond maybe because uh many of them just reserve a receive a state return and they know it's not going to be shared with the federal government so they don't file a [03:43:36] schedule C um but the and the analysis sample may be too small to detect an effect or finally no one responds [laughter] right we always have some people file taxes and comply with the law they know what it is and others [03:43:50] don't. Um uh and so uh you know that's not going to change. Um it would be uh useful to see the full distribution of this percent filing schedule C among recipients by their kind of level of of [03:44:05] of uh uh revenues. It's probably concentrated amongst the the low earnings folks. Um the other one interpreting uh the fact that it does have an effect on uh 1099 [03:44:20] uh K uh is conditional on on filing a 10 uh a schedule C. It does have an effect on the receipts and expenses reported. [03:44:29] Possible mechanisms, the filing without uh a 1099K. [03:44:34] Those filers um who don't have a 1099 K uh uh K uh information return or maybe they're cheating a little on their taxes. and uh filers with uh without a [03:44:48] 1099K and an alternative explanation is they just don't maintain good records and that having an information return is very important for getting accurate tax records. Um uh and I'm also going to say [03:45:01] that the simulations we should be very you know careful about uh I have some comments which I'll skip over in the interest of time. um but encouraged the authors to be explicit about assumptions that must hold for their accuracy when [03:45:16] they go beyond and and are and are are uh projecting the effects. My guess is is that because of this nature of the sample and other factors, it's possible that they're understating the effects of [03:45:29] these more recent changes. Um if you'll give me 30 seconds, I just want to say this is my final slide. I wanted to step back and in in the context of of both of these papers um [03:45:44] kind of ask the a bigger question research question is there a future role for private data to supplement government survey and administrative data. We have spent a lot of time at this conference already talking about [03:45:57] the use of thirdparty payer data to uh estimate uh W2 employment. [03:46:04] um these uh data potentially could also be used for 1099s and that would be a useful area to explore. We might also be able to use from these 1090s uh thirdparty [03:46:19] payment processors um might help shed some light on uh the size of the 10 1099K workforce which is a you know a big question mark. So thank you very much. ## Q&A (03:46:34 – 03:50:23) *Shared floor discussion for Session 5: The Nonemployee Workforce.* [03:46:34] [applause] Uh so we are running a little behind [03:46:43] schedule. Um maybe I could give Martha 30 seconds if she wants to say anything, Dimmitri Andy 30 seconds if you want to say anything and then we could continue [03:46:57] our conversations about this over lunch. Uh thank you for the great comments. Uh the 1099K is uh actively being linked this summer. [03:47:08] Uh there there's this fixed cost that sends us of releasing data. It's called disclosure review and uh it's as soon as you do one, you have to be very careful what you release next. So anyway, we opted not to release our 1099Ks in a [03:47:22] hurry. Uh but we definitely recognize that's an important component uh and want to to bring them in. And then I will say that I was very interested and this is something I did not know that the 1099 uh NEC's and KS are sometimes [03:47:35] issued by the same company. So thank you for bringing that to our attention. [03:47:47] >> Yeah. So I thanks for the comments and and for the presentation. I just want to say that we've been Demetri and I and and and Emily have been turnurning out um work in this area at the IRS for a long time uh in a way that um we hope [03:48:01] has been uh really kind of shed a lot of light on how these forms work, but it's not a kind of sustainable research infrastructure. It's not what the IRS JSRP was made for. So we really think the effort that's going on at census um you know not just what we're doing but [03:48:15] the but with the broader effort to kind of build these uh resources that can be um for uh uh research going forward is really important. Um I would say that you know most of the tabulations that that you guys present today we have put [03:48:28] out in various forms um and and in a way that we won't be able to do going forward. So it's very important that you can and and you know we've promoted these a lot. I think understanding it would be very helpful since we've done the same thing to do it for like 2018 we [03:48:43] have the demographic cuts um in 2016 we have um and and a couple of earlier years the most comprehensive cross tabs of all the various filings um and um you know you may have seen maybe one or two of our papers we actually have a a lot [03:48:58] of things up there um and and and seeing um do you match what we're getting I mean it should be the same universe what decisions matter and documenting that in a sustainable way will be really important for the research community to be able to actually use these returns um [03:49:12] as opposed to just kind of starting from scratch without any kind of regard. And it's not just us. There's actually other work that was going on when we were at IRS by um Katie Lim and Elanor Wilking and Max Rich who um you know we did less on who was hiring them because that was [03:49:27] their turf. and they obviously have a work a working paper that has almost like a hundred tables and and and graphs um that kind of take what you had started doing but kind of for 2016 go banan I mean it's it's it's um about the [03:49:41] you know the dynamics of the relationships and and do people go and come back a lot of information out there and we're happy to point you to these resources but the more you're able to kind of go back and and look at you know which definitions matter and which don't [03:49:54] I think will be um useful for these statistics to have impact that going forward. So, look forward to working with you on this. Yeah. [03:50:01] >> Great. Um, lots lots to talk about over lunch. Lunch today will be upstairs in on the penthouse level for those of you who were here for dinner last night. [03:50:12] It's uh that floor. You can go up on the elevator. We are in the jewel room. [03:50:18] Okay. >> Whatever room it was we were in last night, but there's signs on the doors.