a16z Podcast

Pandemic Relief and Fraud: Willful Deceit or Design Defect?

Bharat Ramamurti, Naftali Harris, Alex Rampell, and Lauren Murrow

Posted September 4, 2020

This episode examines the potential for misuse and fraud among those applying for the Paycheck Protection Program (PPP)—and how fintech and software provide overlooked tools to stop it.

On March 27th, the government enacted a $2.2 trillion dollar stimulus package called the CARES Act, the largest aid measure in history. The act provides more than $500 billion for the Paycheck Protection Program, or PPP, a low-interest, forgivable loan program designed to help small businesses and self-employed individuals retain workers and stay afloat during the pandemic. Since March, the Small Business Administration has approved billions of dollars in PPP loans. But it is also estimated that U.S. losses from coronavirus-related fraud and identity theft have reached almost $100 million. According to the New York Times, the Small Business Administration’s fraud hotline has received 42,000 reports about coronavirus-related cheating and misuse; by comparison, last year it had less than 800.

To date, the Department of Justice has charged more than 40 cases of PPP-related schemes, from claiming non-existent employees or non-existent businesses to identity theft, kickback schemes, fake tax documents, and multi-state fraud rings. Most of those cases have alleged fraud of more than a $1 million. But what about the countless others that may be cheating taxpayers out of smaller—but not insignificant—sums? How does the government decide who should get money and who shouldn’t among millions of applications from businesses of all industries and sizes—and what role do banks play? How does the program then distribute that money quickly and accurately—or not, in many cases? And what tools are at our disposal to catch those who cheat the system?

Host Lauren Murrow is joined by Bharat Ramamurti, the original member of the COVID-19 Congressional Oversight Commission, which is tasked with evaluating the impact of coronavirus relief loans; Naftali Harris, the CEO of SentiLink, a software company that builds technology to detect synthetic fraud; and a16z fintech general partner Alex Rampell.


Lauren Murrow (a16z): The potential for fraud was a concern from the beginning. All the way back on March 16th, before the CARES Act was signed, Attorney General Barr directed U.S. Attorney’s offices to prioritize investigating and prosecuting PPP loan fraud cases. But how prevalent is fraud in the program thus far?

Bharat Ramamurti (member of the COVID-19 Congressional Oversight Commission): We’re still trying to analyze that that. There have been some prosecutions on PPP, but that’s obviously not the full universe of fraud. I think one issue that’s come up already is transparency. A good way of detecting fraud is to make sure that every single borrower through the PPP program is publicly disclosed. But the government is not doing that. They’re only disclosing recipients of loans above $150,000. And something like 80 percent of recipients had loans under that amount. So at the moment, we only know 20 percent of the companies who got money through this program, which is a serious issue.

Naftali Harris (CEO of SentiLink): Well, Bharat, from our perspective, we appreciate the fact that you were able to get at least some of the loans publicized because, my firm has actually gone through some of them. And even with the very limited data that’s available, we found some pretty clear instances of fraud, which we’ve escalated to the banks that actually issued the loans. There’s not too much you can do with more limited information that’s available.

Bharat: It’s really important to track what these companies are doing with the money. And I think a big part of what the Oversight Commission should do and will be doing is scrutinizing the companies that get money through these programs to evaluate how that money is flowing thereafter.

Lauren: Alex, after the stimulus package passed back in March, you wrote this op-ed that argued, I’ll quote, “The last mile of identifying, adjudicating, and dispersing assistance without a sea of fraud is a new challenge, one which the government is wholly unprepared for, and for which technology is the needed answer.” It’s now been nearly five months. Do you stand by that assessment?

Alex Rampell (a16z general partner): I do, unfortunately. I mean, there’s no time machine. If there was, I would do many things with my new piece of technology.

The only way of explaining this whole process really is as a very, very complex, worst way to do it, in my personal opinion, Rube Goldberg machine. And it took such a long time for many businesses to actually apply and figure all of this out, and their bank couldn’t do it. And, of course, somebody like some giant burger chain got the PPP loan and not the 10 businesses that really needed it. A lot of businesses have failed. And I think it’s terrible.

Lauren: The Small Businesses Administration then started accepting loan applications just a week after the CARES Act passed. And that pace, as you mentioned, raised some concerns. For one, to help quickly disperse those funds—which, of course, speed was essential—the SBA allowed lenders to rely on borrower certifications to determine their eligibility. Was that a mistake?

Alex: That’s not how I would have done it. I think there was a little bit of a disconnect in terms of the people that set the policy, the speed at which the policy needed, in all fairness, to get rolled out, and then the level of technical understanding in terms of how the financial system works in 2020 with a lot of tools that have been developed, frankly, in the last 5 or 10 years.

But just to walk through what I meant by that, what does identify mean? Well, if you’re born in this country and you’re a U.S. citizen, you’ve got a Social Security Number. And you might think that every business has a Federal Employer Identification Number, which would be the equivalent, but what about sole proprietorships? Or what about people that just use an illegally fictitious name for their own quasi business? There’s not really a way of identifying them. There’s not really a way of knowing what their income was before coronavirus. And, after coronavirus, how much of an impact that actually had.

Like, you’ll say that you’re self-employed, but there’s not really knowledge that you have this business. So you’re one of the 30 million small businesses in America, but it’s just not really known as an entity to the government.

Naftali: The PPP really shined a light on how bad identity verification is in the United States. In addition to really not knowing who the people are or who the businesses are, if you look at the applications themselves and look at who’s on payroll, really no good way of verifying that. And I think if the United States was able to know who citizens are, and if there’s a way of identifying them, this whole system would have been a lot easier to run.

Lauren: How could that improve?

Naftali: If you look at other countries like India or like Estonia, the government actually takes a very active role in identifying its citizens. In India, they have the Aadhaar system. In Estonia, they have government-issued IDs which even have public-key cryptography on them. And so, in the same way that you can give everyone a bank account, you could also give every citizen in the United States or every business, public and private, key pairs and actually allow them to verify their identities that way. If you did that, you can even prove that an individual actually was on payroll at a given company.

Alex: I think the way to have done it is to say, “Let’s get your banking information.” And there’s a way of doing that, because there’s a company called Plaid where if you go sign up for a mint account, or if you go sign up for any kind of account online, it asks for your banking credentials from your current bank so that it can import all of the data. It’s an API, an application programming interface, to go read that information from the bank. It really is effectively as close as you can get to a certified copy of the financials.

It’s much, much harder to somehow falsify information that is actually coming directly from your bank, coming directly from your credit card processor, or coming directly from your payroll company. And the great thing about 2020 is that there are APIs largely through companies like Plaid that allow that to happen.

And I just feel like PPP it was done in arguably the worst way possible because you’d go to the SBA website and it would say, “Okay, to find a lender that can give you a PPP loan, enter your zip code.” You’d enter in your zip code, and no matter what zip code you entered, it would give you 56,000 results. As a small business, you’d go to bank A and they’d say, “Sorry, we can’t help you.” So then you went to bank C, bank D, bank E, bank F, and then you actually applied 10 times not out of fraud, but because you need the money, so you don’t have to lay off all of your employees and declare bankruptcy.

Lauren: Bharat, do you agree with that assessment?

Bharat: Yeah, I agree. And I think it goes back to the original policy choice that Congress made when this crisis hit. I like to point to Denmark, which took a different approach. Denmark said, “Regardless of how big your business is, if you are taking a hit from all these public health restrictions, we will step in and cover a significant percentage of your payroll costs. And you look at the result of that: In Denmark, the unemployment rate was about 4 percent before the crisis and it peaked at about 5 percent.

In the U.S., it jumped from 4 percent to about 20 percent. And it’s still in double digits as of today. And as Alex said, a lot of companies have payroll processors that have this information handy in order to feed the money into the right places. The government could have partnered with those companies to make sure that the money was distributed. You could have done something through the IRS, which is an idea that Congress is currently looking at.

Alex: I totally agree that other countries have done this better than we have, on the how do they handle their economic woes. And if we just said, “All right, here’s what we’re going to do. We’re just going to support people on payroll not for an arbitrary two and a half months, but we’re going to support them for an indefinite period of time, provided that sales for the business are below some percentage of baseline,” maybe that’s a better way of doing it.

Bharat: I think that that original decision by Congress to treat companies of different sizes differently, and to send them money for different things, created a lot of the complexity that we’re seeing now.

Alex: I largely agree with that. If you look at a company like Amazon that’s grown into the pandemic, arguably they don’t need any help supporting employment. But this is the this is where you get information from the financial statements of a company, whether it’s public or private. Because I totally agree with you, there’s a lot of: “This company shouldn’t be getting a Paycheck Protection loan because they’re rich.” But it’s like, yeah, but if their sales are down by 90 percent and they’re going to fire 90 percent of their workers, how would you feel if you’re a worker where you pick the wrong company to work at, and therefore that’s why you’re losing your job?

So, I think you actually could have threaded the needle on this, which is to not base it on employment, but to base it on: Were your sales impacted? Basically, there were companies that saw massive decreases in sales. And I think the most unfortunate thing about this entire situation is the government did end up picking winners and losers, where mom-and-pop retail just got eviscerated. And Target just announced their best quarter ever. It’s like this giant transfer of wealth from a lot of small businesses to large businesses. If more money is going into your bank account after COVID than before COVID, that’s a pretty good hint that you probably shouldn’t be getting any kind of assistance from the government.

Bharat: Alternatively, what you could have done at the beginning when maybe you don’t have a sense of which companies were going to benefit and which ones weren’t, is that you just get the money out the door quickly to everybody and then you can recoup it on the back end, on 2020 taxes, which are collected in 2021.

But I think the larger point stands, which is that if you care about helping workers, then you shouldn’t really care about whether that worker happens to work at a really big company or a really small company. And I think that the congressional response resulted in very different outcomes for different types of workers and different types of businesses based on factors that I think aren’t really relevant.

Lauren: In addition to the application process, one aspect that seemed to pose a particular challenge was the disbursement. To give one example, the Treasury Inspector General for Tax Administration found that as of April 30th, nearly $1.4 billion had mistakenly gone to descendants. How might the disbursement piece have been done differently?

Alex: The government has never had an organ where it’s like, “We need to send money to people.” I think in a future world, if you were building a nation state from scratch, you’d probably just have your Social Security Number also be a bank account. Where if I, as the government, wanted to give everybody money, I could just push a button, and then boom, it gets deposited into the government bank account. Because you’re already paying the government anyway.

You know, my idea would be: every citizen should have some direct channel of actually both sending money to the government and receiving money from the government, which would ideally be some kind of quasi-government bank account.

Bharat: That is a proposal that is working its way through Congress. A number of Democrats support an idea called Fed Account where every person and every business would automatically have an account at the Federal Reserve, like banks have now. It’s a way of sending money out to individuals and businesses very quickly. But I think we’ve seen now that there are significant costs to not having that ability both for businesses and individuals.

Alex: The problem with the disbursement in many cases was that you’d apply with 20 banks, not because you wanted to apply with 20 banks, not because you wanted to cheat the government and get 20 loans, you just had no idea if you were going to get your money in time.

Because many big banks just don’t do SBA loans. Before this happened, the largest bank by count of SBA loans per quarter was Wells Fargo, which did 692 loans in the quarter before COVID hit. And now we have 30 million small businesses in this country, many of which are going out of business. The SBA just isn’t set up for this. And it was so complicated, partially because the government didn’t want to be in the business of picking winners and losers in terms of, “This bank is competent, this bank is not competent.” They just said, “Here’s every bank, we’re going to let the private market figure this out.”

My proposal would have been—every business now accepts credit cards. So the credit card processing company now has a pretty good idea of what your sales are. Most companies use an electronic payroll system, so they have a login for paychecks, or ADP, or Gusto. And most businesses also use QuickBooks or some kind of online accounting system. So to identify them, to adjudicate them, and to disperse with much lower fraud, there are tools to do this. But the tool is not either a community bank or Citibank. That was the worst organ for disbursement.

Bharat: I agree with a lot of that. And had similar critiques of the PPP program as they were still trying to think through how to design it. Number one, because it operated through banks, and the way that the bank regulations work, there’s a significant incentive for banks to cater to their existing customers first. Which meant that companies that had an existing small business relationship with a bank were put at the front of the line. But for a lot of small businesses who may not have that kind of existing relationship with the bank, they did not get the loan that they needed. And a lot of minority-owned small businesses are in that category. And that’s why you’ve seen this highly disparate racial impact when it comes to small business closures.

Alex: I would differ on this correlation of which people got the loans quickly versus not quickly. There was this just kind of like randomness around which bank you worked with. A lot of it was just this crapshoot: if you happen to bank with a bank that knew what they were doing. And a lot of members of Congress were like, “Oh, see, it’s the local and community banks that knew what they were doing.” But actually, think about the old-fashioned way of making a car, where somebody would assemble it by hand, and the new-fashioned way of doing it, which is an assembly line. These big banks are assembly lines and the small banks are effectively doing things by hand.

And it turned out, in the early weeks of this, the small banks were a lot more effective than the big banks. And that was really what caused this disparity. It’s like, if you happen to be in the assembly line bank, you were screwed. And then you had to scramble to go find a local, do-it-yourself handyman bank.

Bharat: Yes, on that last point, there’s a lot of good PPP data by state, and you’ll notice that the state that was the most successful at getting PPP loans to businesses was North Dakota. And North Dakota is the only state in the country that has its own state-run bank. And so, it’s another data point in support of building out our public infrastructure in terms of delivering money to people and to businesses.

Naftali: Some organizations were very excited about this. “This is free money from the government, let’s pass out as many of these PPP loans as we possibly can and collect some money that we have no risk in.” And others were, I think, a lot more conservative and said, “You know what, this is a new program. I’m not even sure we want to do this at all.” And only based on significant pressure from their customers did they even do this. And so, you ended up with a very uneven patchwork approach and very uneven experiences for different small businesses.

Alex: There’s always this rule in anything where you’re allowing transactions, which is the best way of stopping all fraud is to stop all transactions. And when you need to move quickly to stop businesses from failing, which was the debacle that we were in and, unfortunately, still are in, you’re going to let more erroneous or fraudulent transactions through. It’s inevitable.

Lauren: Right. To that point, there are many different means of fraud. So, some who have been charged have claimed non-existent employees or misrepresented the number of employees. Others claimed non-existent businesses. Some submitted forged bank statements or used to fake tax documents. There was a software engineer who created fake tech companies to obtain over a million dollars in funds. We’ve had several instances of kickback schemes. And there was a pretty well-publicized PPP fraud ring in which five people from three different states were paying each other for services that were never rendered. Naftali, in a program of this scale, what methods of fraud are you most concerned about here? And what are the tools at our disposal to catch these bad actors?

Naftali: I have been thinking about this as three different categories of fraud or, put more mildly, misuse of funds. The worst, for sure, is creating totally fictitious businesses and fake employees and just using it to steal money from all of us citizens. That’s one category.

I think another category is someone that does have a business, they actually are employing people, but they’re misusing the funds that are coming to them. They’re using it to pay themselves a better salary or take money from the loans that they shouldn’t be doing. The last category is about who should really be getting this: Is this business really one that was at risk of going out of business?

I think the ones that I’m most concerned about are really the ones in that first category. Because that’s where you’re really, really ripping off the government and taxpayers.

Lauren: Your company, SentiLink, aims to detect synthetic fraud, which, as you mentioned, is creating a fake person or a fake identity from the ground up. Typically, my understanding is that kind of fraud is coming from more organized crime rings. Is that a concern here?

Naftali: Absolutely. And we’ve seen quite a lot of that in the PPP because it’s very helpful, in this case, to be able to have a bunch of fake people that are on payroll, or to have a fake person that you can list as the owner of a business so that nothing will tie back to the fraudster themselves. It’s actually relatively simple to create a synthetic identity. And so, for the PPP, we’ve also seen a lot of lone wolves attacking it, people like that software engineer, for instance.

I’d also say for the payroll stuff in particular, since there’s very little verification, if any, of the people that are supposed to be on the payroll itself, that’s a case where it’s even easier. You know, there’s even an indictment we saw where the identities that were used as employees on payroll were actually literally generated online in one of those fictitious name generators. So for a program like this, you really need to have a strong ability to audit it at the back end and prosecute in cases where the government and all of us citizens have been ripped off.

Lauren: Right. The DOJ has filed, I think, between 30 and 40 fraud cases [ed: now more than 40]. Most of them are for more than a million dollars.

Naftali: My personal favorite is the guy with the Lamborghini.

Lauren: Describe.

Naftali: Oh, It’s a really great indictment. Essentially, some guy has this lame business he claims has like 107 employees, which is patently not true. He gets the money and immediately uses it for, I think, a Lamborghini, and a diamond watch, and a whole bunch of crazy stuff. But my favorite part of this whole indictment is that it ends with: “These are not approved purposes for PPP loans.” Very understated.

Lauren: [laughs] Yeah. Though while those individual headlines are sensational, so far we’ve only heard about, a very minuscule percentage of those funds that have been fraudulent, versus the funds that are going to help small businesses. We don’t know yet.

Alex: Well, that’s part of the problem. Fraud is not black or white. I mean, in most cases people think there are good guys, there are bad guys, and the good guys chase the bad guys. But as an example, what if you didn’t buy a Lamborghini? But what if you paid yourself a slightly higher salary than you used to, as the business owner? There’s also something that you certified on the PPP loan application, which was: “These funds are absolutely necessary for the continued operation of my business.”

That one’s a little bit more subjective. But I think there’s a whole spectrum of, “I didn’t even need this thing, but I applied for it anyway because why not? Free money from the government never hurt of anybody.” There’s all of these different gradations of, perhaps we shouldn’t throw people in prison kind of fraud like we should the Lamborghini guy, but this is not being used as was intended. It certainly didn’t meet the spirit of the law, and it certainly didn’t meet the letter of the law. How much of that is going on? Or: I wanted to try to hire back all of these employees, but I can’t.

Lauren: Right. And more than 85 percent of the loans issued were for under $150,000. So, arguably, the speed and the scale of the PPP program means we’re likely dealing with much more fraud than usual, but it’s not necessarily the usual fraudsters. It could be—according to the cases charged thus far—a woman with an Etsy store, or a guy who runs an auto body shop, or someone who just takes the money and spends it at a casino. So, are we missing the smaller schemes because of the need to actually get this money out quickly? Does it matter?

Naftali: You know, I’ll tell you, we’ve been helping some of the organizations process some of the smaller loans. Not for every PPP lender out there, but for a subset of them. And we’ve certainly seen a lot of fraud in that smaller dollar range. To your point, it’s around $100,000 or so, not so huge that it would be triggering huge alarm bells, but also not so small that it’s not really worth their time. And the number of fraud attempts that we’re seeing has actually been pretty significant.

Alex: We’re not going to know for a while. Because you have to kind of play the game theory here. Right now, these are loans and there’s an application to apply for forgiveness of the loan. Now, if you’re a criminal and you already bought a private jet with all your PPP proceeds and put your Lamborghinis on the jet, and fled to some country that doesn’t extradite to the U.S., would you apply for forgiveness for all the loans? No, of course not.

So ironically, the companies that plan on surviving, that apply for forgiveness, I would argue are going to be that second tier of fraud. The actual, abject criminals that created fake businesses that don’t exist, that already got the money and ran off with it, they’re not going to apply for forgiveness or anything. And the SBA is going to be on the hook to reimburse the banks and the banks try to collect on that money. So I think you’re going to know once the applications for forgiveness start coming through, and the announcements by the bank to go claim money from the SBA from businesses that defaulted, and, by the way, that have been completely non-responsive to any kind of collections effort.

Lauren: Yes, I do want to talk about that question of loan forgiveness. In order to get Paycheck Protection loans forgiven, meaning you don’t have to repay the money, small business owners have to show proof that at least 60 percent of the money was used for payroll. They have to rehire and retain workers at the same salary levels. So, there’s currently a proposal in Congress for automatic forgiveness of loans that are $150,000 and under. Basically, it requires the borrower to say, “Yep, I used it like I was supposed to.” Some critics say that there’s not enough information available yet to say whether the program is actually working to save jobs, and that by easing forgiveness requirements you could potentially heighten the risk of fraud. Bharat, what’s your perspective?

Bharat: Yes, I share the concerns that you mentioned about that. As I noted before, they’re not disclosing the names of companies that received under $150,000. And what is the level under which you get automatic forgiveness? It’s $150,000. So that seems to be a troubling overlap. I think, at a minimum, you should be publicly disclosing the names of all these companies before you do blanket forgiveness for that set of companies.

Second of all, look, I am I’m in favor of figuring out ways to minimize the paperwork responsibilities for companies to get loan forgiveness. But there’s an important public policy issue here which is that the money was supposed to go to support payroll and keeping people employed. And if companies ended up not using the money for that purpose, then I don’t think that they should be getting their loans forgiven. And so I think it’s important to verify that that actually happened. And the only way to do that, I think, is to require them to jump through at least some hoops before the loans are forgiven.

Lauren: I’d like to talk about how we might potentially improve this, as additional stimulus measures are passed and extended. One issue in detecting and preventing fraud was also the issue of faulty data. SBA data showed that many companies claimed to have retained more workers than they actually had, in some cases not by the fault of the business owner. The analysis also showed that for more than 875,000 borrowers, zero jobs were supported, or no information was listed. Bloomberg reported data that at least a quarter of a million dollars may have listed the wrong location for the borrowers. How can we improve the data to better assess the success or the shortcomings of this program?

Alex: I think there are two things here. My favorite article just took apart the data dump that was provided by Treasury on every single person above $150,000. And Pennsylvania, the state, was spelled like 155 different ways. Because the E-Tran system for processing these loans was just not very good, business owners did not have access to it directly. So they filled out a PDF, sent it to the bank, and then it got transcribed. And then the information was wrong, but as long as you got an ICHRA number on the other side then the bank would disperse the money, knowing that if the business failed, they’d be made whole.

So, I’d say two things. If you’re a restaurant, and you took a PPP loan, and your sales went down by 70 percent, and they’re still down by 70 percent, it does not make sense to employ any people. That’s part of the problem; that would not be a case of fraud. The thing that’s missing from that data, my point, is what the sales have been. We just don’t have the data on how the businesses have performed, except in aggregate.

Bharat: Just a slightly different point, one of the things that I’ve found in government is that simplicity is probably the biggest virtue. When we’re trying to figure out how to design a program, I would say that the simpler the program, the easier it is for businesses to comply with it, the easier it is to get good data on it, the easier it is to monitor compliance, the easier it is to enforce whatever the rules are of that program. And my other rule would be: when in doubt, make sure you’re being over-inclusive rather than under-inclusive. Okay, so maybe some businesses get money that you don’t think should get money at the margins. To me, that is a small price to pay for, number one, getting money out quickly. And number two, making sure that companies that do need the money are getting it.

And so, I think that on those two metrics, the congressional response so far has been fairly shaky, in that, if the goal is to make sure that businesses get money in order to keep people employed, there are about 10 different ways of doing that more directly than the one that Congress chose. And the result is that we’re going to have to collect a lot more data and there’s just a lot more opportunities for people to slip through the cracks and for fraud to occur because we created all these intermediaries and we created all these different hoops you have to jump through.

And so, as a result, you create all sorts of weird incentives and outcomes for companies where there may not be a way of getting people to come back. Like, what if all of their employees have childcare problems and they can’t work right now because they can’t leave the home? And so, when you have a country of 300 million-plus people, and however many businesses, there’s always going to be edge cases that opponents of the program are going to seize on to say this was a bad idea. But at the end of the day, the goal is to get the most help to the most people as quickly as possible. And I think doing things as simply as possible is the key. And I think learning from this experience, to develop basic infrastructure so that you can deliver relief more effectively in the future, is also really important.

Lauren: Well, to your point, we’re talking about potential improvements here. But the current administration says that 51 million jobs have been “supported” by the Paycheck Protection Program. The program has been credited with driving down the unemployment rate. Is it cynical to focus on fraud, which, as we discussed, may be limited, when in fact, we should be celebrating the program’s success thus far?

Alex: I mean, look, if a business has had sales go down by 50 percent and they got a loan that was basically handling two and a half months of payroll, and we’re in month five of this, it just seems like a very hard to believe claim. So I think actually the claim itself should be investigated and not just taken at face value.

Bharat: You know, I would say on PPP, I’m deeply skeptical of the number that came out of the administration. David Autor, a professor at MIT, did an independent analysis of PPP. He found that there was a pretty wide range when it came to the cost per job saved through this program. But the median amount, or the middle of his range, was about $200,000 per job. Which, frankly, isn’t that great. And I think this speaks to both flaws in the design of the program and inefficiency in delivering the money to the companies that that really needed it.

And, of course, that that also doesn’t account for the cost of potentially allowing tens of thousands of businesses that could have qualified for PPP to fail because we were unable to locate them and provide them with money in time.

Naftali: One thing that I’ve made a point to do is when I get takeout or go to a small business, I’ll ask, “Hey, did you apply for the PPP?” And more often than not, they’ll say, “No, I didn’t. I don’t want to have a loan like that,” or “I didn’t know how to complete it,” And that’s a huge shame. And I just hope that the next time that we do this, if we have to do it, we’ll be able to serve those deserving businesses.

I think if you look at the spectrum between ‘let’s lock down this money as well as possible and make sure it only goes to the most deserving businesses’ as opposed to [inviting] fraud, the program was skewed towards the ‘let’s give out money more quickly.’ And I think that was the right thing to do. But on the flip side, it’s taxpayer dollars that all Americans have worked hard for and given to the government.

And the fact is that I don’t think limiting fraud or finding the fraudsters is all that challenging or all that expensive to do. And so even if the amount of fraud is limited, I still think that from a matter of morality, we as a society should go after the people that have stolen money from deserving businesses.

Bharat: I totally agree with that. I think that it’s totally reasonable to focus on fraud. I would just argue that there are other important factors for the government to consider when developing these types of stimulus or relief programs. And that in some cases, there are there are trade-offs where you make choices that could potentially open the door to more fraud, but they have upsides either in terms of speed, or inclusiveness, or whatever the case may be, that that that make it worth it.

Honestly, I was frustrated after PPP came out and so much of the focus was on a handful of big companies that got money that were essentially shamed into returning it. You know, if they weren’t eligible under the rules of the program that’s one thing, and certainly they should be required to return the money in that case. But there was far more focus on that issue than on the fact that tens of thousands of companies that were eligible weren’t getting the money. And that a disproportionate number of them were owned by black or Hispanic owners because they lacked traditional banking relationships or there weren’t banks participating in the program in their area. That, to me, is a much bigger tragedy than a couple of burger chains getting money they shouldn’t have.

Lauren: So, you’re effectively saying some fraud is price we have to pay for getting people the aid that they need in a short period of time.

Bharat: Yeah. I would say that there is a trade-off between efficiency, inclusiveness, and potential fraud. Now, you can solve the fraud on the back end by identifying cases of it and bringing enforcement actions to cut down on that or to hold people accountable. And I’m all for that. but I find that what happens more often than not in government is that this idea that “the most important thing is making sure that people who don’t deserve the money don’t get it” leads to all sorts of terrible policy designs and outcomes that have much greater harm than, you know, some small group of people getting money that we think aren’t deserving.

Lauren: Naftali, you mentioned it’s not that challenging to catch this type of fraud, particularly in the first category, which was people who are claiming non-existent employees or misrepresenting their number of employees. How can we do this better?

Naftali: I think we should have, in addition to private banks or other lenders checking for fraud, uniform checks that are done by the SBA themselves. There are lots of products in the space to detect various different kinds of fraud, including a uniform check that the beneficial owners are actually real people and who they say they are, that the people listed on payroll actually exist, as well, so that we’re not just at the least common denominator of whichever lender wants to issue the most money as fast as possible.

Lauren: We’ve talked about how the stimulus is unprecedented. Alex, do you have a perspective on how potentially tech could help detect fraud as additional measures are passed?

Alex: Well, we can only do it in the constraints that are provided by the bill that is or is not in existence. There’s a lot of stuff that you can change to better adjudicate what is fraud and what isn’t. I mean, I 100 percent agree with Bharat that focusing on this kind of “envy capitalism” or whatever—”Oh, this person got money, they didn’t deserve it. And we should have really made it a one-year process before anybody got money so that not a single cent was wasted.”—that itself is so pennywise and pound foolish that it could just topple over the economy. So, that doesn’t make sense.

But I think the main thing is get data from companies and use that as an automatic adjudication mechanism. And then ideally, just have a disbursement mechanism that comes from not banks but from the government directly rather than paying banks $15 billion to give away the government’s own money, which is the people’s own money.

Lauren: Bharat, from your perspective on the Oversight Commission, do you get the sense that in future stimulus measures they’re looking to standardize the process into something that would improve some of these issues?

Bharat: So, I think step one is strengthening the authorities that we have. So, for example, we don’t have direct oversight authority of PPP. We don’t have subpoena power to, force people to testify before us or to get documents from other agencies. So, strengthening those oversight authorities would, go a long way towards ensuring that there’s robust, independent, and bipartisan oversight.

I think that we have learned from the experience of PPP, or Congress has learned and is pushing towards different approaches. So using the tax code to provide direct support to companies without having to go through banks, that’s included in the HEROES Act, which is the bill that the Democrats in the House passed in May. And that’s been sitting in the Senate ever since.

And I think that there’s been a lot of discussion recently about what data is necessary for full transparency and accountability for these programs. So, the PPP discussion has led to follow on legislation about mandating that every single loan recipient is disclosed, even if they got under $150,000. Mandating, and I think this is critically important, certain demographic data about the nature of recipients, so that we can get harder data about black and Hispanic borrowers, for example, systematically being excluded from this program, either because of purposeful discrimination or because of things like access to traditional banking services. And, if so, is there something additional we should be doing to target those communities?

So, I think there’s been a lot of congressional interest in strengthening oversight authority, in improving the quality of the data that’s being collected, and in tweaking these programs so that the money is delivered more efficiently.

Lauren: Alex, if you were to write your op-ed again today, what would you adjust based on what you’ve learned over the past five months, since the rollout of the CARES Act?

Alex: Honestly, I wouldn’t adjust that much except the duration. So, I still think everything that I wrote, I still stand behind that’s the right way to do it going forward. However, what if this goes on for a year? What’s happened to businesses where, there’s still demand, there’s still supply, but the government has said you can’t go there, which has helped Amazon. That’s why you’re having this transfer of wealth from, in some cases, mom-and-pop businesses to big corporations that are able to meet their COVID requirements. Like, say I need to buy Plexiglas to have an ordering station. As a small Mexican restaurant, I can’t do that. But Chipotle can. There are all sorts of kind of winners and losers being created here, which is really fundamentally unfair.

If this is going to go on for a long time, it’s actually healthy in capitalism to have weak businesses fail. The airlines, should they have gotten bailed out? Maybe, maybe not. But if nobody wants to fly anymore, the answer’s no. If the government stops you from flying, the answer is probably yes. A little bit more nuance than that. So, I think the real question is, if this goes on for a very, very long time, and it’s not just a blip, does it make sense for the government to say, “Everybody’s a winner? And we’re going to protect businesses that weren’t doing a good job before?”

Because if you look at the restaurant space, a lot of restaurants go bankrupt on their own. So if this plays out for years, should we be on the hook as taxpayers for bailing out everybody who provided bad service to customers? I don’t think anybody would say the answer to that is yes, even if workers will be impacted. But where do you draw the line? And since this has been going on for a much, much longer time now than when I wrote that in March—it was like a stopgap measure. it’s very hard to have a one-year stop measure.

Lauren: Thank you all so much for joining us on a16z podcast.

Alex: Thank you.

Bharat: Thank you.

Naftali: Thank you.