just getting started. And they were doing tons of business registrations in those locations, and we were able to capture that rise in those particular office builders. And so then let me just kind of kind of finish off and then we'll take a little break. So this is like, okay, all right, we were all impressed by this earlier. And so now the question is, how would you actually see that? And then what could you do with it? And so this is the same thing. And so here is right that exact areas the area obviously adjacent to MIT, right, here's our campus. Here, of course, are the pizza shops and the bars and the accountants in central in Central Square. This of course, over here started to capture the growth of Kendall Square. And if you'd like looked really carefully, it actually started over here, right where like, right, like my whole list is now right, like one Capital Square, but then after the rents cut, they're so high because Biogen and occupy expanded. Then all city went and all of a sudden it was over at the Cambridge Innovation Center. And then what you're able to do over time, is actually see the full evolution of an entrepreneurial ecosystem. So far, so good. Okay, why don't we take till night at work? Okay, yeah, that's it. I'm happy to you know, so why don't we take 10 minutes or so eight o'clock, and then we'll cover.
Thank you.
Thank you. Very good. Right,
right the fact that you're able to capture that full dynamic with specificity at the building level has two features one is cool little graphic, but the it's a super good proof of concept. Because basically the the ability to have a very good test of are you capturing something real using a predictive analytic is to take an example that know the answer is at a time and then prove out at some level of detail your answer your your predictions against the actual graduates. And so this was basically a ground truth that kind of worked and we did many others as well. Okay. So once you have a tool like that, then we can use our statistics to actually reduce it down the filing is and it was taken an hour if you're like, very efficient, right? To actually look at the state of American entrepreneurship. I think we gave this one out. And right and so this is now break. I'm trying to remember what the gaps here are. So this is right so this is rectified. So this is our net. This is our, our, our now cast, where we use a more limited number of variables, right and then this is the company. And what you can see here is that we capture, right so this is rec pi, that kind of quality adjusted quantity divided by GDP. Okay? And what you can see is we capture extremely well, this big boom in the 90s. So if you remember in the other graph, I probably should have done it where we compare the two more directly with but you can definitely see, you know, in the estate and these data, we absolutely see. Right the.com And then there's is in fact with Joe observed a decline but just to be clear, that decline is actually not relative to its min for relative to the peak of the.com. And then things kind of go along and then there really is a recession. But then things come out pretty quickly from the one year and then things kind of recover and achieve essentially this by 2014. They achieved essentially the second best year. And once again, an absolute level that's much higher because of GDP basically doubled in that time. Okay, now, there's a little hit there, you can already kind of see what must be the problem here. Not a problem, but just like the reality. So here, you can sort of see here's 1995 Here's 2000. Which year do you think it was better? So the more high quality firm founded in 2009. Which year do you think it was better? To fact why? Absolutely. It was like if you found it founded a high quality firm in 1985. You literally had, so this is now this reais number, okay? And what this is capturing is how many growth events we predict how many occurred and what you can see is that basically 1995 was was basically a Pete for a long time, you know, through the, through the when we're able to actually measure performance when we actually had the data and then actually went along your downward spiral. Right. And in particular, there's a real decline right after 2000 and then the kind of one of the things about the 2000s cohorts wasn't so much that we didn't have enough shots on goal. It's not enough when it should have just very different playbook and what we talked about before. Yep. So sorry, sorry. So this is react. So this vertical axis is one. And so imagine that one has number of growth events that occurs is exactly equal to the number that you think. Yes, it's just the ratio. So it's literally just the ratio and So 1.6 means you got 60% Boost. And point six means you got a report. And so, what's interesting is because the index is like a growth protect your type, it's like a concrete number because so many firms are supposed to see a group, the cohort X, you can literally say, how does that compare to that?
I spent too much time in this in our paper. We then use that to do one of the you know kind of test of whether or not quality goes up or down and recessions and we're able to sort of use you know, some time series to sort of look at that. I have to admit, you know, the referees made us do that. It's I'm not sure if some of you have had access, just so you know, there's, I'm gonna show you some examples of this. But basically, you can go and look at essentially, any data that you want
for any state for any county, you're happening to google street address, and we provide
a dataset that goes from making up through 2016. The zip code level, the county level,
and the state level, the state level and the country, and yet so many people and use that now, and I'll show you some examples. Okay.
So does it have the underlying data? Because it's that data because many states will have once you start getting started downloading millions of records, people were like crop and so we have a data agreement with the state. Some of those states allow us to distribute the data, some help some, you know, we are sometimes we're like, oh, we talk to someone who will talk to you to hear oh, you here are eight states that are easy to work with to find if you want to use that micro data, just to kind of recap.
Yeah, I was just curious. When you are describing the Harvard Business Review article for foreigners are doing decision making goals, like so I was wondering if the distribution of beta hat has a similar shape, what they will just want us to shape that like distribution about before. And then I was thinking about comparing that with
oh, so um, so I don't think they have today one thing where there's more products and things like that, for most of their manufacturing are more big, but but actually, it's the effort around the use of natural Raphaela. That is, for bigger firms in general, but it's a very powerful it's a very similar thing. Like, I don't think they even believe that their individual practices are a bundle of practices that have a better quality measure. You can still use that. But absolutely. Two, which I would say we are probably much more skewed because Are they really looking at productivity or percentage growth of the firm? We're really looking at this very skewed distribution among the firms and their usual data set 00 to 100.
Okay, one last thing, because I don't want to get too detailed with this, but it does, but there is a sense sometimes, you know, we were able to use this so that when we kind of move past this, but like put, but there's I think I'm saying this for right, the nature of these business registration records is massively under add to your question. Massively. So for example, during the COVID there was this like incredible, right, so I remember March 2020, bad month. No one was happy. Right? Definitely not my favorite month. Right. And, like we thought and certainly you know, John, hold the way we like everyone who studies entrepreneurship to heart we all love that's it. Entrepreneurship is going in the toilet because we're about to hit this huge economic recession. Right. Doors are fragile, and even what seems to be this emerging legislation that's coming out is going to protect existing firms rather than new. So like, whatever your metric is, it's all going downhill. Right? And then there's just been this amazing fact both in the business registration records, right. Right, that there's just been this massive increase in the rate of entrepreneurship in the United States since the pandemic and started like in April 2005. Okay, so that's this is from halter why your state this is from the right, you know, so on this point, like there are small portraits and we have some disagreements on the quality versus quantity debate in the broader context, but this we all agree on that point. You know, we're all looking at quantity here. We looked at our state level records, they look at their federal records, and it's all the same. There's a big increase right. Now, interestingly, right. And they right. Yeah, I'll say great.
Right, so one thing you can do with this and just I think it's indicative, because I think no one's really taken advantage of this kind of data. We were able to do it very quickly and tough continue, but I think there's a lot to it. So for example, during the pandemic we were able to capture in real time, like, what kind of things were not going to open during the pandemic. Restaurants. Right, right. Great, but what was going to open up tutoring camp nanny, right. There was a lot for that. Because if you said I'm sorry, right. severally there was a huge burst this in New York City, in online virtual digit, right. So we were able to use for names for easily downloadable data that we were getting on a daily basis, basically, to basically track out, you know, kind of what businesses we're going into, which is once again as many consequences and one is sort of recording that people somehow relatively quickly bounce back in terms of their entrepreneurship right. But also gave you a sense of how the value was shifted. You said upside, and in particular, right? We were able to track us geographically, just kind of changing demography. So so everyone talks about this big increase now in entrepreneurship. After COVID. Right. What was interesting about that, this way, right? Is this is New York City, for those of you don't know, here, red is good, blue is bad. And what you can see is it's all in the outer boroughs. In fact, Manhattan never recovered in terms of his entrepreneurship rate, right. And we were able to then do some additional work in that. That basically documented that there. The largest increase in entrepreneurship was basically a
historically lacking with us. Whether or not that was some curious
interplay between the where those places where civil rights, you know, kind of right kind of the better and deployed some things and we're sort of working on that right now. But sort of right we were able to but the point being that there by using business registration records, where you're really able to get at who owns the firm, right? You can look them up on the internet, your ethnicity, by names, gender, and as well as their names as a firm, what industry you can sort of write it, those are not perfect, but I think those are underutilized resources, or search be very retail. I think innovation is awesome. Alright, so far, so good. Of
okay
so now what I'm going to try to do so that
so why don't you one more layer of how to then kind of think about not just these data, but more generally data, sort of think about how we're characterizing these place based sort of realizing the truth a little bit about the twin twins, but that's okay. Okay. So for example, this came out basically from this course and people are here. So this is when the son of someone who's at Georgia Tech, and which, you know, we're talking about all these types of things. Right. And she became really fascinated with sort of right to think you know, when you see that math of it, and you know, like, I'm learning from habits that people were, that was kind of what was in the air, you know, years ago right? was kind of like well, how is it that that local, that very local industry actually influences the universities themselves? And then potentially both the direction of research and the entrepreneurial activity by people within universities and so she, and then she did that within and so there's a, there's a bit more of an animation paper, but it's kind of let's go with it for just a bit. You guys haven't seen the script peered into
just a bit more of an innovation paper, but I think it kind of gets at this idea that these places matter. You're trying to look at where ideas come from, and ultimately what their consequences are. So she very cleverly, I think there's got a really interesting, so I think it's useful to think about what the challenge is. I think she did a really good job of thinking about that challenge. And it's kind of I'm not sure many other papers that sort of stood up to that challenge quite as as squared. Okay. So on the one hand, right, I showed you that picture, right of Silicon Valley, I showed you where Stanford was, and it was very natural to think of that Stanford University is an engineer. Right? You think that the President of Stanford, well, now he's got I'm not sure he's president sandwiches anymore, but, you know, President Stanford get
fired because of some stupid journalist you
have committed some Trump. I think the scientific he was himself with eminent scientists, who was not sufficiently attentive on high profile publications.
Yeah, okay. Like, don't worry, he's been.
He does not he is like, totally not. Okay. Right. So then people do that and they build things like Cornell Tech in New York City. Right. hoping against hope that that's going to spawn some entrepreneur revolution. Right and once again, Cornell Tech kind of problem is good on some terms, but as an attacker, it's called the scraper revolution, right? And so the other thing you might say is Well, actually, right. If you actually talk to people at Stanford, it might be equally possible maybe even more possible that what happens is that Google and HP which are just a notable neighborhood, are themselves influencing what's happening. So it's not so much this direction, to say I'm saying as this other direction until the question becomes, right how right so we would want to write so can we actually see whether the activities of the local environment, shape the choices, the research choices, and the real choices of
of, of the research university? Okay.
And so, what he does, and so she says she does that in a very clever setup, which is the there there's an industry that started 25 years, 30 years at this point, agricultural biotechnology industry. And the great thing about ag biotech is that you're not in that industry, unless you know it. And what I mean by that is there's only a certain number of industrial labs a certain number of startup a certain number of researchers who study your papers will be like, on a study in this field, you're not studying plants and biotech. You're not in a very narrow kind of stuff, I guess, calories or something like that, but so it goes right and so then what she does, right is she says okay, how could I kind of look at this kind of bilateral relationship, right. And what she does is she takes advantage, right? Of the fact that the existing so that so this was an industry that went through a revolution. Basically, there were a small number of firms that historically were actually not embodied. They were kind of related but like in this world, that basically these are people who solve herbicides. Okay. And then what happened was people discovered doesn't really matter where they discovered that there was this whole new opportunity called agricultural, biotech, and the very natural place to conduct that research if you were an industry was within your own lives. So for example, Monsanto, which is located around Scipio St. Louis. Basically, I took a several billion dollar investment, like per year, like massive, massive r&d into ag biotech in St. Louis. And then what major universities in St. Louis, Washington University of St. Louis, and so though, you could say is, once again, Washington University of St. Louis wasn't in the biotech industry, either in any meaningful sense and then you could ask, how much did the activities of Washington St. Louis change over time, in response to a change in local industrial complex
which was the rise of ag biotech?
Yes, that is absolutely correct. Okay, and so what you want and so she does this kind of multiple levels, but you could imagine, right that there are universities that are similar in everything, except whether or not they're co located like, like they have a biotech program. They have an exporter, but they're not co located or not within within investment. Similarly, she has researchers who are either co located right with Monsanto or not. And then what she's able to do is she's able to then document what was the impact of the change in industry research on the incidence rate of publications within this domain. And then I think, once again, I think that adherence work right. And in particular, what she's able to show is that that number that that impact is much higher for those universities, which themselves had a very strong technology transfer office. Kind of makes sense. Okay. Okay, why? Sorry, I'm just I realized. I mean, just like one result is the energy additionally shows that that led to more startup like it was it that there's kind of more startup extra numbers outwards with more startup activity, a month that comes out of that activity, so she's able to sort of show right last right. In fact, the emergence of a much more clustered dynamic, as opposed to just the industrial lab. From that two way interaction or rental portfolio? Okay.
Sorry, the two way interaction somehow leads to the emergence of a cluster. Is that what you're
saying about this? So before ag biotech, there was a chemical industry that was probably closely tied to Wash U and otherwise you're now a new technology that emerges ag biotech. You have a big industrial lab, the big industrial lab affects quite a bit. Right. Right. So effects right you know, that's a big gap. Right? affects the amount of plant biotech that's being done at the University that itself engenders more startups in the region from that university. And that creates not just the industrial biotech now, but also now a little smattering from that.
Is it a generally right analogy for thinking University and science and industry as a tech? Like you emphasize the causal loop?
Sort of, but not completely right. You know, universities do engineering and science and they do you know, I'm saying, so I think the core here is that that boundary is a little bit more porous than we might think. There can be right. It is true that universities have different norms and industry. Right, open AI, they think they're doing computer science these days. Maybe they are maybe they are not, they certainly keep it lively. You said I'm saying, Rachel, I think for example, machine learning right now, a lot of most interesting foundational research has been done in companies. And then there are other areas where there was interesting applied research just on that right now with the kind of classic again, you should start at some point, right? Like few weeks ago with Pierre. Okay, okay. But if you think back to the idea of Dr. Rajan Stein model, that's a classic one where they sort of derived universities is doing a more scientific research industry doing more applied research, where the control rights in the university are held firm by the university, by the researcher and the control rights in the industry being held by everyone so yeah. Okay. So so once again, what we're still trying to do here and so just to be clear, what did he do that I really liked? She was able, you know what, I do think that you know, I think people your paper came out of an organization science, great paper, but I do think that kind of like, people still haven't kind of gotten COVID have leaked, it was was that what she was able to do was fine, a really nice experiment that allowed her to trace out the impact of the local environment on the trajectory of research, you know, fairly well identified. And what's the other other people who choose not to do that? The second thing you'd be kind of interested in is once again, there's like, we're interested in these innovation driven entrepreneurial ecosystems, right, people get very excited by that we're gonna spend more time on a little bit on the policy consequences and we want to be interested in kind of, okay with these ecosystems are important, but then also people do things within these ecosystems and I also met, okay, once again, Peter was at USC, right?
He has this great paper that is now published.
Right? So let's think about this. Right. So most of you, how many of you have heard pillpack Look at that, okay. Hilltopper shocked by Sloan MBA, a guy from the New England
College of Pharmacy, TJ Parker and Elliot Cohen,
and they built a company called pillpack. That was had a very simple premise was that when you have multiple medications, what you want to do is right, if any of you are very ultimately, many of you will be grandparents. They'll have like many pill bottles, there'll be lots of fights over which fill out should be taken right in what quantity, right? And so what they did is they're like no, the right answer is like, rather than have the grandparents kind of fighting over this. What they should do is you get a single path that tells you exactly what to do when it has all of your posts because you don't care which color says is the pill within the pack. It tells you at 8am do this. Okay, so they don't let's start with the most obvious point. Elliot Cohen took my course. Do you think I get any causal? Okay, so they went off with this idea that raised $58 million within 30 months. And one of the things that was true about T gay and Elliot was they were incredibly, very interested in using every single resource on this campus and in this ecosystem. So they went for example, they hung out at the voltage Cafe, which is now closed but used to be on Third Street, everyone. They spend time in my office and took my course they spent a lot of time with the Trust Center. They use the venture mentoring service at MIT. And in particular, they did TechStars which was a new liest at that time newly established accelerator in Boston in a relatively short amount of time. That's five years after founding. They were bought by Amazon for $1 billion. A lot of people happy. I was very happy for them, their investors were happy for themselves. Okay. Interestingly, when they just saw you now, when they got bought for $1 billion, the market cap of CVS two they got bought by Amazon for $1 billion. And the market cap of CVS and Walgreens went down 10 to a steady, like it's really drip, okay. And that became the team that then built out what we call today Amazon pharmacy, that sort of large arguments definitely brought pillpack into Amazon, and then that then TJ and Elliot led the transition that was cool. So if you heard that story, and you were Tech Stars, and you're trying to persuade people that Tech Stars or right was a good thing. You would tell that story and you would minimize the amount of time that they spent in my class when you said a lot of times plenty other curious about
2013 Okay,
is that is that like packing pills for over the counter medicine?
It did no, it's packing pills for for primarily prescription drugs. It also is for over the counter but the real challenge once again, till packers had existed in the past. They focused on solution that was end to end that just really focused on user and customer. I can tell you a case like we were we wrote a piece on it and found where they go really job layout what's different about them that they're going to be in competition with CVS and Walgreens, like all their challenges are all laid out. They were very aware of them very shortly. With respect to your good point, you might say Oh, I mean, this is not like I could put an example in your journal up here, but that's a little bit like step one, be a genius step to go back to step one. This is like a, you know, any of us could shoot, we're smart, but there's nothing right? But TJ actually knew his dad was a pharmacist and he himself was part so he knew a lot about that industry and La really how to he just incredibly entrepreneurial, and they were able to think about what that meant, you know, kind of how you could actually drive sales using this model. Does that make sense? Yeah,
I was not understanding why that
they didn't have to have this isn't awkward to be able to say that they scored in the top 1/10 of 1% of our quality distribution that's on site. But in terms of there are many more success stories that look like this relatively modest innovations, but that really opened up a whole new market and sort of solve the customer problem in terms of concentrates relative to Maderna. Right, because by construction, right is like you know time that's like much
more fundamental, more deep tech
okay.
So, over the past 15 years, all these accelerators have grown up, right many of you I think you're right, you're participating. Right? Maybe not. But there's lots of these accelerators. There's TechStars, there's Y Combinator, but also tons of others. Right. And a good question to ask is when you see firms say you couldn't write so just to kind of think of now let's try to write a big question, because many times governments put money, right? Do they work? And then you might kind of get into theories of like, what you what or what's true, right? So you might think that what happens is that if you're in a good ecosystem, like MIT is the ecosystem, right? Relative to Cleveland. So then maybe we don't need accelerators. Companies don't need accelerators, because they already have everything. They have lots of mentors. They have lots of, you know, network and you said a time, that's 3132 is if you have if you're in a weak ecosystem, right? Then in fact, right, or you might have an alternative theory, right, but an alternative theory, where the very fact that you were in a strong ecosystem for like a rocket ship, right, every additional piece of power, this helps you go kind of farther along. Many distributions, right? Each of these chances are really, really low in a population. Right? So whatever, if you put pillpack in your data set, you've assigned to the control group with a treatment group that's largely determined by their identity program.
Quite frankly, it's not totally what you say I'm saying.
So but you might think okay, right, if I have a strong ecosystem, that actually amplifies the impact of the accelerator, because if people were able to use the returns from the accelerator more deeply within their endeavors, right, so in other words, if beta two minus beta one is greater than beta two minus beta zero, then right, we kind of have complements, rather, right. And in the reverse, you'd have the marginal return right? In a strong ecosystem. So if beta two minus beta one is greater than beta zero, right, then that's going to be equal to well, that's right. And then the reverse they do focus on everyone. So far, so good. Okay. And so then what they do or what did they or did was he took advantage of the existence of a program. Okay, so, now we have like, Okay, so there's program for an ecosystem. Now we're gonna look at how they interact in some kind of collaborative way. So there's a publicly funded program that they know happening, called mass challenge. One of the leading publicly funded salaries the United States people kind of pre big on it, but who knows but works well. Here's how we find out. There's 128 teams per year so it's nice is there's lots of teams in the last few years. I've had an infinite number, but we're starting to get to good numbers. There's a mentorship program. Right? And we already know that among the firms that after this program, they've raised you know, kind of even four or five years after founding, they have raised more than a million
what's the problem though? If you say, ah, compared to all entrepreneurs, these people raise a lot of money. It's because these are the firm's that got into the program, right. And what do you write and then select into the program based on like, you have a bad idea, we try to find the most promising for the improve. Right? So not necessarily when it comes to but that's what they do. Okay. So what they want to observe is over four year period from 2010 to 2013. If there's about 1000 firms, right, applying for the program, essentially didn't come do your math. Less than a few percent get in. Okay. And then for every startup, those those that apply, apply and get in and those apply that don't get in, he observes whether or not they are injured, the score that they get, and some demographics and kind of your family. That's very similar to kind of what Jorge and then we see the 128 startups, and then we can look over time through your survey and as well, our public data, the way he's able to then observe outcomes for both the admitted applicants and the non admitted applicants. Okay. There's, I think it's, I think it's what at this time, it was one rap, who's 120 and I'm not sure if it's time divided into like, too cold to too many covers that either. So what he then does is, is that then there are these judges that Judge ventures are there, you know, experienced judges, but they you know, they agree and disagree with each other that other papers, right? But the judges are drawn from a pool of experienced investors. Very early stage, it's probably hard, but the great thing is the judges all use the same standardized scoring sheet that and so here's kind of the core of the paper. Here's the distribution of scores kind of in a standardized way, relative to zero. And here's the probability to get the program. Right, so it's incredibly sharp, regression discontinuity. The scores are fairly continuous. The chance of beating the program is determined by whether or not okay, so of course, what that allows them to do is they compare the outcomes of these firms versus each firms everyone sees that okay. Increase light. Yep. So there are some below the threshold that also when there are Yeah, yeah. So there must be some group that I, oh, you know, must be like, this is the announced threshold and then somebody says, you know, maybe this firm decides they can't make it or they discover something and then you know, who knows that you know, maybe, but it's pretty tough, you know, time
okay, and then what he's able to do is that look at right yeah, basically could run 25 regressions but so I'm gonna say he looks at the impact on significant financing this is not to this not that hard to refute. Right? But basically as a function, right of whether you're in the program or not, and what you can sort of see as well as increasing in the scores in general. It's right. There's discontinuous jump in the cloud. Right. And let's be clear, that, you know, I remember seeing this graph for the first time I was like, alright, you know, because like, I hope you don't, don't tell me that's made up because that's what kind of just be like, Okay, right. But then the interesting thing is, I'm sure I remember right. Here's the interesting thing is then you can go do that to look at interplay between the effect and the ecosystem. So how does has the ecosystem that we've been talking about, right, like we read, we started with this innovation ecosystem now, right? We want to kind of entrepreneurship ecosystem, right? And does that also matter in addition to the program, what you get here, right? Is to first approximation, you get that there is a and this is sort of like a meaningful size. The fact is that you get an additional bump and there's an additional ballpark, it's quite significant in the returns, right to the program, if you're in debt or if your home location is an event that isn't actually used as people who are in Boston, and the value of being an OT 139. You should I'm saying as opposed to like in Moulton is accessing
everyone so far so good.
So no, I mean, I think I think there's some theories of it, but basically, the theory behind this entire is it's kind of it's almost like a rocket ship theory of ecosystems. It's that every additional accelerate you can put on these firms makes it more likely to succeed. And so like being in an already strong ecosystem is even better to put on. Now, there's a good question as to whether that ultimately faces diminishing returns, and he has an equilibrium level for sure. Eventually rents will you know right, you get rises in rent supervisors wages, and that's right, but there's kind of like this idea was, is it the case that we should be introducing programs or places that are already happening? Okay, we're almost done and we'll probably finish up just a minute or two early. Get to now we're gonna turn back so so the reason I showed you he as particularly Dan's work was to show you Okay, so alongside work being done around, you know, kind of up your quality estimation by Jorge and gang, right. People like Dan, we're sort of kind of carving out this path of then how to how to take the basic fridges to just to be clear, there's nothing particularly revolutionary here. But you definitely need to ask the right kind of question for these ecosystems. And you have to get access and understand the data and the institutions. And then you end up with this very nice, theoretically motivated question that has nice policy consequences. So it's kind of beer is a good example, where very careful Applied Economics in toolkit can be applied to understanding this kind of area. So it's currently needed to economists and policymakers in terms of really
okay, okay. So
let's just finish up with a set of things. And so now and you know, there are just kind of a bunch of ways that you could go and now
shot now let's go back to using the startup photography, project data. Some of these are going to be papers by us. People on the opposite team is off people just use the data in all sorts of crazy trends to study this. So this is a paper by Elle Hochberg and co authors. And they just I will say, this is an idea that once I saw what they do with the data, I said, we should have done this like that was my definite thing. So there's a separate set of data. That is a little controversial, but basically gets at the idea that different places of the United States are different in terms of how we run it. Like literally like how, like, you know, are people living on the frontier. And it's literally these bands of places. That are how long that place was on the American frontier, during westward expansion. It's everywhere that people started, you know, obviously there were Native Americans living all over the United States, but settlers in Germany, that kind of siloing right occurred for the number of
people, right, so right so there's this westward expansion. And then what you can do, right is look at does that kind of historical legacy of being on the frontier of fat controlling for other factors, the rate lock, right. And so they have this really nice piece which is like using our data, right, which is a shown up both the total number of newborns and starts per capita and quality are basically increasing in the degree of distance from the basically whether it looks at the degree to which that zip code is a has historically been on frontier. So this is a fascinating to cultural consequences you get from that, but it's certainly better. Okay. Okay. And in some sense that goes you know, in some sense you like, I would say, this paper is the best paper I've ever seen. That sort of gets at the idea that like, there must be some people who choose to live in their work that live in cultures that are maybe a bit more entrepreneurial. Or more likely to become independent their own.
Instead, also the predictive analytics.
So the predictive analytics here is coming from this total frontier experience is just literally this map. It's like the zip codes tell you literally how long I think it was a county level or zip code. I can't remember where it is, but I guess it was the county, how long that county was on the American frontier. That's the x axis data normally. And then this is just our data. This is the quantity variable. And this is the this is this is the right this is cricket is the quality index of the firm's in that county and you know, in some years,
since they use your quality,
they took our data, okay. All right. So this and this are directly taken priority. And this and, and, like 100, slides back, and they have a general map of just the incidence of entrepreneur, the quantity of entrepreneurial activity by every county United States. Okay. When a good question then comes. Sure, you know, that might matter. But really where you want enough is really shows what's the particular way that policies and institutions matter. In terms of impacting the quantity and quality of entrepreneurship.
So there's Valentina Qatari, this is paper that she and I did together. Okay. And so the first thing you could do right is once you had this using our data, is you could then locate whether or not the first one went to one beat institution that's associated, presumably with the assistance everyone says. And so, what we do is we look here, at the quantity how quantity and quality are impacted, right by whether or not you're co located with a research university or not controlling for a number of things, right, as you can see, right being a research university matters. A lot. It actually matters more over time. So these are looking at places that have research university versus not. Does that sound like a good is causal relationship? No. Right. Right. But right so what you'd like to do is somehow control and vary over time the kind of amount of university is right the strength of university and then kind of trace out the impact of changes of university on this little block early versus their entry that's like, it'd be something like I don't ferret. And you know, there's kind of like a bunch of things that we do here, but I'm gonna just kind of, kind of get to the main point. So what we do is we take advantage of the fact and sort of don't think about it here. Think of like, universities across the US Okay, so that you know, much more, you, MIT and Stanford have a very scaled budget. So they don't have a lot of variation year to year because there's just so many Democrats, but most universities actually have significant grant variation. And so what we look at is
right, we look at specifically we look over time with any university so what we do is look at every single United States, we then divide we look at those that are zip codes that are adjacent within five miles of university or not. And then we look at the rate the kind of rolling stock of grants received by that university in the last few years. And then look at in a kind of, right three years ahead, what happens to the rate of entrepreneurship in those success. And so what's that as that's asking, is it saying how does a change in funding affect to the university affect the subsequent quantity and quality of entrepreneurship? And they're kind of two interesting kinds of results. The first is we look at kind of all sorts of cardio both r&d and non r&d funds to both universities and national labs. And it turns out that just giving people more money raises all kinds of entrepreneurship. Because you know, if you just give you know, there's a new sports stadium or something or new hospital independent. In fact, one other church going on there, you're going to need more services. But once you count for quality, it turns out all of the action is on r&d funds, not non r&d funds, and only to universities, not national laboratories, etc, are very big effect across a wide number of things. Just in the interest of keeping you know, not to be over what kind of come back to one or two of these other studies going forward. But what I want to emphasize here is that this is an approach, right? What you want to do is we now have these datasets, and hopefully we're just thinking of the kind of journey which we started with the idea that people have forgotten that we have gotten into the teaching of this course this year, about entrepreneurship, poor entrepreneurs trying to scrape away and all we care about is innovation. Then we said okay, let's care about entrepreneurs. But then we immediately ran into the fact that we weren't interested in small firms, but interested in the young firms. Even within young firms. We're only interested in some of those firms. And we needed a way of capturing that. And that led us down this whole route of kind of creating a whole edifice around the idea
of our firm that led to some fancy maps
and graphs. But once you now start to account for quality systematically, within populations of entrepreneurial firms, you're able then to then combine that with the traditional applied you tools applied econometrics. And then you can undertake in principle, right to bind predictive analytics and causal analysis, kind of causal studies of the impact of institutions and policies on
ecosystems.
We will return to those. So let me give you one last thing is that by tomorrow, I will send out your three or four readings for the syllabus for next week along with some supplementary readings. My apologies for that kind of finalize that there's a whole bunch of new papers that we've written and done and other things, and I've been sort of unclear in my head on exactly what to do for next week, but I think
