Unlock Online Signature Legitimacy for Non-Compete Agreement in European Union

How to eSign a document: online signature legitimacy for Non-Compete Agreement in European Union

>> Greg Pewett: Welcome, and thank you for standing by. At this time, all participants are in a listen-only mode. Today's webinar is being recorded, and the recording will be posted publicly. If you have any objections, you may disconnect at this time. Now, I'd like to turn the call over to your host, Bryce Hannibal. Bryce, you may begin. >> Bryce Hannibal: Hi, everyone. Thanks so much for coming out today to the FSRDC presentation series. We have kind of a double header today. We have Matt Marx, who will be giving his presentation on his RDC research, following which we'll have an introduction to RDC research by Emily Greenman, who is the RDC administrator at Penn State. Just a couple announcements – we'll be doing Q&A at the end of both presentations. So once Matt has finished his presentation, we'll do a Q&A, and then, we'll begin Emily's presentation, both of which – so we'll do Q&A at the end. However, feel free to put any questions you have in the chat window, and be sure to select all panelists when you do so. I'll be monitoring the chat, and then we'll get to each question after the presentation. And of course, if you have any interest in presenting your RDC research, please feel free to reach out to me, and I would love to coordinate with you to make sure that – yeah, to make sure we can get you on the list. Okay, so today, we are fortunate to hear from Professor Matt Marx. He is the Bruce F. Falling, Sr. Chair in Entrepreneurship, and professor of management and organization in the Dyson School of Applied Economics and Management within the S.C. Johnson College of Business at Cornell University. He will be giving his presentation, titled “Employee Non-compete Agreements, Gender, and Entrepreneurship.” So without any further ado, Matt, I turn it over to you. Take it away. >> Matt Marx: Thanks so much, Bryce. If anyone can't hear me, let me know, but I hope the audio was all set up well. And I just want to thank you very much for this opportunity, and the chance to talk about this research. Now, this is an article that came out last year on Labor Day, coincidentally, because we're going to talk about labor mobility, and freedom to do what you want as a worker. The article did come out open access, so I'll throw a link to this in the chat if you want to download the article and have a look. Of course, as with any article that uses confidential data, the standard disclaimer applies. I imagine this audience is very familiar with that disclaimer, so I won't read it to you. But before jumping into the article, I just wanted to acknowledge the many efforts of, I'm sure, many people on this call in creating these public and also special sworn access data files, which are truly remarkable resources. I just wanted to mention my own very modest contribution to the public data world, which is if you study innovation, and if you're interested in the relationship between the public world of articles and creation of science, and the commercial world of science, I've linked patents to articles through citations, and a website called relianceonscience.org. It's an open data set that you can download for use in your research, and I'll throw that in the chat also for later. Again, freely available for anyone to use in their research. Anyway, with that, let me turn to motivating this article, and I want to do so slightly different from how I do in the article, with a graph made from the company PitchBook. This is proprietary data that we licensed recently. It covers basically every venture-backed company on the planet since around 2000, and one thing that PitchBook does is, it collects data on founders, and also the gender of founders. And what you see in the PitchBook data is that barely – not even – or barely 20% of companies worldwide that have raised venture capital have even one female founder. And if you look to the right of that graph, that even dropped – that dropped during the pandemic, for reasons you and I can all imagine. So that's a massive gender gap, and understanding the gender gap in entrepreneurship is a first-order question for economic progress, for equity, and is the subject of this article. I'm far from the first person to ask this question. Many before me have tried to understand the reasons for this gender gap. Among the many articles is an excellent study by Brooks et. al in PNAS. They took the same PowerPoint slide deck and script, and had men and women read this – give the same presentation to randomly selected judges, who then rated the presenters, and ended up rating the men more highly than the women, even though it was precisely the same pitch. So there's many studies in this genre, but they tend to focus on individuals making conscious choices to discriminate towards men in favor of women. Which could be either for taste-based reasons, or for statistical discrimination reasons, but there's a new, more recent line of inquiry actually drawing from sociology called institutional discrimination. And that's what this article focuses on, and so, let me talk just a little bit about this. There's an article in the JEP by Small and the late Devah Pager where they talk about something that is more innocuous – not innocuous, but perhaps unintentional. It defined institutional discrimination as differential treatment that is either perpetrated by organizations or codified into law. It need not result from personal prejudice, i.e. taste-based discrimination, or from rational guesses on the basis of group characteristics, i.e. statistical discrimination. In other words, it could be unintentional. That is, laws or organizational practices that had nothing to do with gender may have gender or race-based effects. So one example of this is an article by Castellaneta that came out a couple of years ago, where they looked at the changes in institutional – sorry, in laws for incorporating new businesses, and they found when one country simplified the red tape for setting up a new business, this helped to close the gender gap in entrepreneurship, even though the rules said nothing about men versus women. And that's in the same spirit as what we're going to investigate today. Now, what's interesting about institutional discrimination is that although individual preferences, and practices, and biases can be very difficult to change, institutions, and especially laws, can change very quickly. And so, today we're going to study one particular institution, namely an employee non-compete agreement – now, in case that's not a familiar concept, let me show you a non-compete that I signed when I began my career in 1995. It said, more or less, “Mr. Marx hereby covenants not to work for any company in the speech recognition industry, anywhere in the world, for a period of two years after leaving Applied Language Technologies.” Now, I signed this non-compete – this is a paraphrase of it – in Massachusetts, and this is now – this would now be – it was valid when I signed it, but if I signed something like this today, it would be invalid and unenforceable, because the law has changed in Massachusetts just a few years ago, thanks to a lot of reform work that's been done. And the crucial thing is that the individual states choose whether or not to allow firms to require employees to sign these agreements. Now, the general idea behind a non-compete is to restrict an ex-employee, someone who leaves the firm, from joining or founding a rival business, and we'll focus on the founding in this paper, for one to two years, sometimes longer. And these are quite prevalent in the modern economy. I ran a survey of engineers about 10 years ago that found nearly half of knowledge workers have signed a noncompete. I then started a broader survey covering basically all fields, and found that one in five workers across the entire economy has signed a non-compete, so this is a very broad phenomenon. You may have heard of Jimmy John's, the sandwich maker, had its sandwich makers sign non-competes. And so – and that's the thing. The employer has to ask the employee to sign, but also, the non-compete has to be enforceable by the state. Now, in the U.S., there's no federal statute, at least not yet. If you look at this map, this shows you the heterogeneity that exists from state to state currently with regard to enforcement. Now, California famously will not enforce a non-compete, except in very rare cases, which is the sale of a business. Also, Oklahoma and North Dakota, but there are shades of green or gray across the rest of the country, and, as we will take advantage of in this paper, these policies have changed substantially over time. Now, to the – and in fact, last summer, the administration announced its intent, or suggested the FTC look into the possibility of a national mandate on non-competes, where some things may be underway. We will see, but for now, we can take advantage of this state-level variability to make progress. So, now, the question that may be on your mind is, well, why exactly would non-competes affect the gender gap? Because there's nothing in the contract I showed you, and probably nothing in any non-compete that I've ever seen, that mentions gender. Also, nothing in any law regarding non-competes that mentions gender, so why would it have an effect – why would it affect men differentially versus women? But here's the logic. So, first of all, we know that most startups fail, although this is less true when the founders take advantage of industry expertise. That is, they found a company doing something that they know, as opposed to something completely unrelated, and the late Steven Klepper made this the – this is the last epoch of his work, showing this in lasers, and tires, and many different industries. Founders who leverage their expertise are much more likely to succeed, yet that is exactly what non-competes are designed to block you from doing. Now, because non-competes add legal risk to the business risk already inherent in starting a company – that is, if you leave your company – you have an idea, and you leave your employer to start a company, you're not in California or a non-enforcing state, your employer can sue you to block you from continuing with your startup. So that's legal risk on top of business risk. Now, I've said nothing about gender. How does that work in? So one reason it might work in is that, if you do get sued, you would – you could defend yourself, but that's expensive, depending on who you hire as a lawyer, and as I will show you a bit later, women tend – using LEHD data, women tend to start companies earlier in their career, and having earned less money with lower average salaries. Therefore, their relative cost of defending themselves against a lawsuit is higher, which may mean it's more expensive on a relative basis, and they may be more reluctant to even bother to try to mount a defense. By the way, I'm seeing the chat now, and Wesley [assumed spelling] says, “Can you refuse to sign a non-compete?” You can refuse. However, the employer may say, as they said to me when I said, “Wait a second. What is this non-compete?” They said, “Well, you don't have to sign it, but then, you don't have to work here, either.” It can very much be a condition of your employment. One change in the law in Massachusetts is – that we made in 2018 is that the employer now has to tell you about the non-compete in your job offer, but that's very unusual, and does not apply in most states. So, great question, Wesley. Thank you. So what else is going on? Fourth, we know from prior literature that women are penalized more harshly for professional failure. Two of my favorite papers on this topic is one by Heather Sarsons, who looks at cardiologists. Now, cardiologists often don't choose their patients. Whoever comes to you, you operate on. You might get a patient who is going to respond well to surgery. You might get one patient where there's not much you can do. It's not really up to you, and so, controlling carefully for risk factors, Heather looked at what happens to the cardiologist after a certain patient dies. What she finds is that, when a male doctor loses a patient versus a female doctor loses a patient, that the female doctor receives fewer referrals after losing a patient, that there's a gendered effect of failure. Egan finds a similar effect in financial services – not with death, but with accusations of fraud. And, as I'll show you in the LEHD data, after a woman abandons a startup, a failed startup, and returns to paid unemployment, she experiences more of a wage penalty than do men. And in results that aren't in the paper, but which were disclosed through the census process, I also find a wage penalty for women who return to the workforce after a period of unemployment, a larger wage penalty than for men who took a similar-length break from the workforce. So, harsher penalties for failure. Now, there's another component which is played up in the paper, but I think you can make the argument even without this. And that is, if you have read Marianne Bertrand's excellent 2011 summary of the extensive literature on gender differences in risk aversion, both experimental and observational studies have found that women are generally more risk-averse, which may contribute to not wanting to start a company if there's a higher risk of being sued. And I'll show you evidence consistent with that, though I don't think you have to appeal to risk aversion alone to make this argument. But, taken together, given that non-competes increase the risk and the costs of failure, non-competes can discourage entrepreneurship among women and exacerbate the gender gap. So, how do we test this? Well, we're going to use the census data, and look at workers in not all states, because I'm not a census employee with access to all states, but I have 25 states and the District of Columbia for about a 20-year period. So I'm looking at all of the workers, except I'm going to reduce them to a subset, and the reason for that is that I need to track the workers' entire career without gaps. Many workers change states. They start out in one of the 25 states I have, such as Maryland, but then they move to a state where I don't have access, like Massachusetts. And if I include that worker in my data set, I may have missed them starting a business, or I may have missed them returning to paid employment after doing a startup. And so, what I do is, I throw out any worker who left – who – where there's a gap in employment, and I can't verify that they're in the – still in the LEHD. That reduces the data set drastically to about 5.7 million workers, but where I can track them entirely during this period. And I don't miss a thing. And this is thanks to a – the 2014 edition of the LEHD. Now, my dependent variable is starting a company, and starting a company in the same industry, and by that, I'm going to use six-digits NAICS indicators, because we want to know whether you're utilizing your expertise, and not just moving from, say, a manufacturing to services, which might be entirely unrelated. And in terms of determining who's the founder, the LEHD doesn't have records of who the founder is, and so, I'm going to use the Kerr and Kerr heuristic of the top three earners in the first quarter that company is in business. And then, what I'll do is, I'll assemble a state-level, year-by-year panel of non-compete policies, essentially building on work that I had done with Mike Ewens, where we had done a one-zero indicator of state-level changes, but this time building on Norman Bishara's index of non-compete enforcement, the strictness of that, and doing that throughout the 20-year period. That index is also available in the – at the link, if you wanted to use it. Now, there's one big threat to identification, which is this. We don't know from the census records who has signed a non-compete and who hasn't. That information simply isn't in the employer-employee match data. So what we're going to do is compare what happens to employees in industries where non-competes are not used very often, such as – such as education, or construction, or other industries like that, versus industries where they are used. And that, we know from Evan Starr's survey, where he found that in architecture, engineering, computers, et cetera, non-competes are used much more often. And so, I'm going to split the sample that way, and build a regression model. So again, here are the 20, 25 states and District of Columbia that I'm going to exam, and again, only the people whose entire careers are within those 25 states. It doesn't have to be that they were in Iowa their entire life. They could've moved from Iowa, to Colorado, to Texas, to Maine, but if they moved to Montana, then I take them out. So I have to have seen them their entire career. But here's the regression equation. In the interests of time, I won't dwell on this, but our – but we're identifying off of the non-compete policy, the gender, and that they're – and whether they're in a high versus a low-use industry. The first result is that women are less likely to start companies than men. That's fairly well-known, and if you just interact women with tighter non-compete state-level regime, it looks like there's less entrepreneurial activity, but we don't want to conclude too much, because we don't know who has signed. We're more confident here, interacting non-compete policy with being in a high-use non-compete field, and there, we also see a negative coefficient. We find that we interact that with gender, which is the result that we are trying to test. Now, there's three coefficients there. The one that I trust the most is in column five, because here, we add prior employer fixed effects. You might be concerned that some employers are more likely to sue. Other employers let you go, and they don't care. Here, we're, you know, putting fixed effects on the prior employer to make sure that we're accounting for that sort of heterogeneity. When we do, we see that women subject to stricter non-competes in fields where non-competes are more often used are less likely to start businesses in the same industry. And that's an effect that doesn't occur when we look at moving to different industries. Okay, so why is this happening? In my remaining – oh, I need to move along. We look at five potential mechanisms. First, do women get sued more often by their employers? And for this, we're not going to use census data. We're not going to use Lexis/Nexis lawsuit data, because you only get the decided lawsuits. We're going to use the Courthouse News Service, which is an amazing and underappreciated data source that captures all the filed lawsuits. This is important with non-competes, because what often happens is, you get sued, and then the judge issues a preliminary injunction saying, “I can't hear your case until sometime next year, but maybe you are doing irreparable harm to your employer, and so, we're going to ban you from working at that startup – let me get the lights going again – until I can hear the case.” And so, what I did is, I gathered this data, coded the gender of about 12,000 defendants in around 17,000 lawsuits, and found that about a quarter of the defendants were female. Now, is that high, or is it low? It depends on what industry the lawsuits were in. If it were construction, where 90% of employees are male, you'd say, well, that's pretty low. If you were looking at hair salons, where 90% of the employees are female, you'd say that's really low, because you'd expect more. So, let's do that. And this is – sorry for the eye chart, but you can see here are the top five occupations where there are lawsuits, and actually hair salons are among the most-litigated non-compete fields. Because when hairstylists move, they tend to take their clients. They break their non-compete, and they get sued. So we've multiplied the number of defends times the share of employees in that field, ing to BLS estimates, that are female versus male, to compute an estimate of how many lawsuits there should have been, given those breakdowns, and given those estimates, we say that we should have seen roughly half of these lawsuits against women. But it's only 25%, so we conclude that there's actually not bias in the lawsuits. Women aren't over-represented. If anything, men are being targeted, or perhaps what is happening is that women are just not leaving their employers to start new businesses, because they fear the lawsuits, that there's a chilling effect. Okay. So not lawsuit targeting. Second point, relative costs of litigation. So what I'm doing here is a P represents paid employment. E represents entrepreneurship, and I'm going to calculate the earnings that each person in the LEHD had before they started a company. And what I'm going to find is that women have lower pre-entrepreneurship earnings, both because they have lower salaries, and because they have fewer years of work experience. They start companies earlier in their career. Men wait longer on average, and in the interests of time, I won't show you a regression table. But this holds even when you control for these six-digits NAICS sectors. That said – and so, even if you're no more likely to be sued, the defense is more costly, because presumably you haven't saved – you don't have as much earning power. Now, we can't see household wealth in this data, so that could affect us. But we think that the relative costs of litigation would play a role. Penalties for failure – we return to this setup, but this time, what we're going to do is compare what happens to men versus women whose startups fail, or more precisely, who abandon their startups and return to paid employment. So here, I'm putting an X through that second E. So they left their startup. Either it shut down, or maybe they got sued, and they said, “Sorry, I can't stay. You keep working on the startup, but I've been sued. I need to go find another job.” What I'm not including here is a startup that got acquired, which would be a non-failure event. But when this happens – and so, I'm going to compare the earnings they had when they returned to paid work with their pre-entrepreneurship earnings. Did they get paid more, or less than before they became an entrepreneur? And here, I will show you a quick regression result, and that is that women are penalized more once they return to paid employment after abandoning their startup. Again, not an acquisition, but if they abandon their startup, they have a greater penalty in returning to the paid workforce. Now, I've put this in yellow and not in green because I'm not sure ex-ante whether women are aware that this is a penalty. If so, then that would be a stronger mechanism, but it may contribute. Okay, fourth, there's a – even if you start a company, who will you hire? Because if you're in a state that enforces non-competes more strictly, you're going to have to deal with non-competes in being able to hire people. And so, what I'm going to do now is, I'm going to take all the data, and take it down to the firm level, and look at the firms that were founded, and build a couple of new variables. Analyze the percentage of employees who had previous experience in the industry, people we want to have in our company, and also those who had previously worked with the founder, who were in their network. What we'll see here is when the startup is a rival, those founders know the value of that experience, and they try to recruit more employees with industry experience. That's that first coefficient, but they're unable to do that when they found a rival in a state with tighter non-competes, and when they're in an industry where non-competes are used more often. They're blocked from doing what they want to do. Now, there's no gender in that, but look at the final model. What I'm doing here is, I'm adding the wrinkle of employees with experience who you have worked with before. Women are less likely to recruit their former colleagues who have industry experience when they're subject to non-competes. Now, why is that? Is it because they're reluctant to put their former colleagues at risk of a lawsuit? I don't know the mechanism. We're not allowed to interview, of course, the people in the LEHD data set, but it – suspect it may be this chilling effect of knowing that a lawsuit is possible, and then being hesitant to move forward. The last mechanism – and then I see I'm coming up on my 30 minutes – is the screening of high-growth ventures. As I proposed, non-competes make the already-risky prospect of entrepreneurship even riskier, and so, if, as literature suggests, women tend to be more risk-averse, what kind of startups are going to get cut? Prior literature suggested that non-competes filter out weaker startups, because if you're going to found a start-up, don't do it unless it's a really great idea. But I'm going to suspect that the opposite may apply here, that actually, we may see this effect among the margin of the riskier and more high-potential ventures. So this will, again, be firm-level analysis, and here, my dependent variables will be the initial size of the start-up, the final – the – as big as the start-up ever grew, and then how long the start-up survived. And what you'll see is that at the initial founding of the start-up, there's no difference in size for women. However, the startups founded by women don't grow as large in column two, even if you're controlling for initial size in column three. However, they do survive longer. Now, this is a surprising result, because in prior work, when gender wasn't taken into effect, non-competes seemed to screen out weaker startups. They had larger initial size. They grew, and they – which is the opposite of what we see here. And so, what may be going on is that this risk tolerance is screening out actually some of the higher-risk, higher-reward businesses, and instead, when women decide to break through the non-compete barrier and found a company nonetheless in the same industry, using their experience, they're founding businesses that are maybe less risky, business that wouldn't require hiring their friends with industry experience, as we saw on the last table. And therefore, the ones that we see getting founded last longer, are less likely to fall apart, but also don't grow as large, so more conservative approach. Okay. So just to sum up, I think what this paper lets us see is that non-competes, even though they don't – (audio cut out) – institutionally discriminate against women. Because these contracts are sanctioned by the state – and by “state,” I don't mean the country. I mean individual states make these decisions, and they seem – they not only disproportionately discourage women from founding startups where they can utilize their expertise, they appear to screen out startups that would be higher-potential, which then might be higher-risk, higher-growth, and require hiring people with relevant experience. But there's also good news – which is pretty bad news. The good news is that, unlike tastes, and unlike individual discrimination, these things can change. In 2018, after seven or eight years of hearings, Massachusetts enacted some sweeping reforms, including the fact that if you sign – if you get a job offer, the offer has to mention that you will be asked to sign a non-compete. If it doesn't, then the non-compete is void. There are many other state-level reforms, including a notable reform in Chicago that took effect the first of this – sorry, in Illinois the first of this year, where it says that if you are below a certain wage threshold, you cannot even be asked to sign a non-compete. That alone is not allowed, and that's important. Because let's say you're in California. Let's say your employer says you have to sign a non-compete, and you don't know that California won't recognize that. You might sign the non-compete, and still think you are bound by it, but now in Illinois, under a certain wage threshold, you're no longer allowed to do that as a company. So I'm optimistic that we will see more state-level reforms, and hopefully this gender gap in entrepreneurship will close. Okay, so that's what I had to say about the paper. I was asked to also mentioned a couple of thoughts about working with the RDC. So let me talk about that. I do see, oh, one other question. In difference in occupation levels between male and female – oh, yeah, I think I talked about that. Oh, but you asked – in terms of signing a non-compete, there are small differences, not huge differences. And let's see. I need to think more about this, but I will come back on that. And Gina [assumed spelling] asked, “Did you use SBA data on startup?” I did not. I just inferred it from the LEHD. So thoughts on working in the RDC – oh, there are more questions. Oh, hi, Christina [assumed spelling]. Question on initial size – let me take these questions, then I'll talk about work in the RDC. Okay, Christina asks, “Initial size of smaller – business.” So, actually, Christina, they're not smaller. So women – at initial size, the businesses founded by women under non-competes are no smaller or no bigger. There isn't this same effect of screening out weaker businesses that you see – your sample. What I think you are seeing is some of the – I think you're seeing some dreams dashed. You're seeing that women who have ambitious startups they want to start, that these companies aren't getting founded, and where you see that is not in the initial size, but in the growth of these startups. They don't grow as large. So maybe that was the question you were asking. Yeah, we don't see them grow as large. That's what I think is the most worrisome thing. Let's see. Wow, lots of questions. Wesley asks, “Survey weights to account for the survey method” – so the only place where the survey was used in – so I don't know if, Wesley, you're talking about the lawsuit data, or the – oh, you must be talking about the lawsuit data. Actually, I'm not sure – can you clarify, Wesley, what you're asking about? Is that about the lawsuit data? Actually, let's just talk about it offline – because I'm not quite sure what you're asking. And, David [assumed spelling], I haven't done a welfare estimation. It's a great question, but I've stopped short of trying to do that. So anyway, I just want to be respectful of the other speaker, so let me just offer a couple thoughts on working in the RDC. First thing I would say – this is good advice for anyone working on any type of code. Always write comments before you write the code, before your access may be interrupted. You may lose access, and then get back in six months later to work on an R&R extension. You may move, and not be able to get back to the RDC quickly. There may be another global pandemic. You will forget what you're working on, and it's not like having the code on your laptop. So that's just one thing I wish I had done better. It took me weeks to get back up to speed, and the code is huge. And so, that's one thing I would say. The second thing – because the data are so huge, at least I did not want to run everything from scratch every time, and so, I would do one – I would, you know, convert – I'm a Stata person. I would convert the SAS files to Stata files, convert the state-level files, and then batch them up into a national file, take the national file, and make it this file, and then I would run regressions on this file. And I got into a bit of a bad habit of just updating the final file, but you want to go back and re-run the whole pipeline periodically, even if the pipeline takes a week to run, and mine did. Mine took, like, five days to run end-to-end. Because if you only run the last step in the chain, you can get into this feeling of, like, oh, my, I wonder if the whole pipeline works. And then you might get scared running the pipeline, and you should just force yourself to re-run the whole pipeline – I don't know. Not every week, but once a month, periodically to make sure that there isn't cruft in the middle, such that it hasn't become replicable end-to-end for some reason. Otherwise, you might get into a habit of wanting to, like, save your data files with your disclosure, which is not where we want to be. We want to be replicable end-to-end. Another thing is, there are – whoops, there goes the light again. We're reminded often in the RDC to not ask for more memory than we need, and it's – they portray it as, like, do this for the good of everyone, but it's also a selfish thing. And so, I can think up a number of times that I asked for a 32-core machine with half a gig of memory, and went home for the night, and then came back the next day saying, “Ugh, my process hasn't even been allocated yet.” So I would say get in the habit of instrumenting your code, find the biggest merge or the biggest regression you have in an interactive session. Run top, see how much memory's actually being used for that, and then request that plus 25% when you submit to batch mode. It's not just nice to other people, it's smart, because your process will allocate faster. Also, merges don't run faster with more cores. Don't get a 32-core machine for a big merge. Estimates can, but is a 16-core machine really worth it, versus eight, or even versus four? How much faster is it going to be if you have to wait eight hours longer for that machine? And I learned that far too late in the process. Also, if you're running everything serially, it can be smarter to learn some shell scripting and just chop up your job into parallelizable batches that you can put over eight different processes that run in parallel in smaller footprint sizes. Okay, now, here's the thing that you can learn mostly from my bad experience. You're working with massive data files. You want to make them smaller. You may be inclined or tempted to try to compress some of the key identifiers, like PIKs or Firm IDs. Please do not do this. Oh, let me dissuade you from doing that, for I have done that, and I have lived – and I have paid the price. You might think everything converts back, and you'll be fine, but you will not be. You may not be fine. Your mileage may vary. I had very painful experiences with this. Stick with the original identifiers. … do not try to be too clever, because you will come to regret it when you try to merge with other data sets to that. Then the last thing is – this is more thinking – oh, and Shirley [assumed spelling], wonderful to see you. Just realize that more data is not always better in the eyes of referees. You know, it's amazing to have data that is a census, but, you know, it's – you have to make the case for why you've done something really interesting with this amazing data, and not just, “I have everyone.” Anyway, and with that, I'll stop talking, and hand off to our other presenters. Thank you so much for listening. >> Bryce Hannibal: Thanks so much, Matt. It looks like we have maybe one or two questions that kind of have some follow-up to their previous questions. So I think Dai [assumed spelling] is maybe how it's pronounced. They're saying, “I'm also wondering, since female founders are fewer than male ones, and maybe only the high-potential founders will go for startups, which increases the probability of being sued, and limited by non-competes.” Do you have any commentary on that? >> Matt Marx: I'm seeing this question – startups. That is a difficult question to answer, but an important question, because I can't observe intention to found. I really wish I could. I'm trying to think of data that would let you look at that. Let me think about it. I don't have a great answer for that. Part of what's going on here is that you are dealing with – in my research, I've found that the lawsuits are less important than the chilling effect that the non-competes have. There's only a few thousand – there's thousands of lawsuits per year, and if that were the whole story, no one would bother with non-competes, since it wouldn't have much of an effect. It's really about the – really about the expectation or the fear of the lawsuit. And so, that may play into it. So let me think more about that, I think. >> Bryce Hannibal: That's perfectly fair. Wesley, I think I'm probably going to direct you to an RDC administrator for your question about RDC data and methods generally. And I think, Matt, the last question that I see, unless one has popped up, is from Gina. And she's asking, “Do you identify that women take less risks because they are less likely to understand risk mitigation, and are not trained on risk assessment?” >> Matt Marx: Oh, Gina, I'm glad you asked that. So actually, I have no opinion on that matter, and I think that these findings can stand independent of risk aversion. All I'm saying is that there are dozens and dozens of studies that have been published documenting higher risk aversion among women. Marianne Bertrand has a JEP article in 2011 that summarizes this literature, and there's been even more articles since then. And you be the judge. There is a critique article in 2016 in a feminist economics journal, and I'm trying to remember the – Nelson [assumed spelling], Nelson wrote it. And Nelson's critique is that that article fixates on statistical significance, as opposed to magnitudes, and that we have to be careful about generalizing from the tails. I think it's a good critique. So that's why I try to emphasize the costs of failure, because I think those costs of failure are real. And you can make this whole argument without even appealing to risk aversion. I don't think you need that for the argument, and I don't personally have an opinion. Thank you for asking. >> Bryce Hannibal: Okay. Let's give it maybe just a few seconds to see if anyone else is typing a question into the chat. And in the meantime, Matt, thank you so much. This was fascinating. I was thinking in the middle of your presentation, my wife is a hairstylist, so that hit home. I had no idea Jimmy John's had non-competes, so – >> Matt Marx: Tell her watch out. >> Bryce Hannibal: – yeah. >> Matt Marx: I bet she signed a non-compete. That is one of the top five occupations for lawsuits. >> Bryce Hannibal: Yeah, that was intriguing. I'll make sure she reads your paper. How about that? But thank you so much. We really appreciate it. And I don't see any other questions, so I guess let's go ahead and transition into our RDC introduction, introduction to RDC presentation. And so, with that, I'm going to turn the time over to – oh, before I forget, if you have any additional questions, Matt has agreed and is happy to answer any questions offline that you can e-mail to him directly. His contact information is in the flyer that we sent out from your RDC administrator, and is available online if you Google him. I think we're all available online. So, thank you so much.