
TWICV special with Matt Hendricks from Perspective Data Science
Gary D Stocker (00:01.219)
Welcome back to a very special episode of this week in college viability. Hi, Gary. Hi, everybody. Gary Stocker back with this special episode. If you follow college finances, my guest is becoming, is fast becoming, maybe the top resource for using data to tell the good and the bad about college financial health and viability. Please welcome Dr. Matt Hendricks. Matt, thanks for making time to join us.
Matt (00:30.264)
Thanks, Gary. Thanks for having me.
Gary D Stocker (00:32.397)
So Matt, economics is your field of academic expertise. Tell us about the journey that ended up with your PhD and that.
Matt (00:41.986)
Yeah, so I guess to, I initially decided that I wanted to do a PhD just kind of interacting with my faculty at my undergraduate institution, which was really small institution. And I was really curious about economic research, particularly trade policy, which was a hot topic back then and has now become a hot topic today. So yeah, sort of through the
Gary D Stocker (01:03.023)
Yeah.
Matt (01:09.39)
Small college feel that I had and relationships I built with my, uh, sort of mentors at my undergrad institution. They said, wow, I really like economics. I want to, uh, do an advanced degree in it and maybe teach someday. And so that's the path that I chose, um, ended up at the university of Minnesota doing a applied economics degree path, which is a bit different than a traditional econ path. Uh, it's more data focused. Um,
econometrics is like required econometrics just fancy word that we use for stats, right? A lot of stats that's sort of specialized in what we do in economics trying to a lot of it's trying to tease out the causal effect of, if you change one policy lever, how does that and hold everything else fixed? How does that impact the outcome? So my training was really heavily focused on that.
Gary D Stocker (01:59.415)
Interesting. Interesting.
Matt (02:04.878)
And I was a trade economist initially, but turns out trade data is not as interesting. There's not as many, I like using bigger data sets. And I sort of naturally moved into a different field, which is labor economics, where you have big data sets and sometimes really important questions, right? And to me, the most important question at the time was how
Gary D Stocker (02:11.319)
Hahaha
Gary D Stocker (02:25.636)
Yeah.
Matt (02:32.034)
How can we improve worker productivity in the education field? So I'm talking about teachers primarily. I don't really care so much about the productivity of somebody working at a plant or something. Although that's interesting, but I just think it's far more interesting if we can improve the productivity of a teacher and get better student outcomes out of that. And then this idea that maybe we can improve.
teacher productivity and student outcomes for free, right? That's the ultimate question that economists wants to ask. And so that's sort of how my research started. And my PhD was on performance pay programs for teachers. And that's really where my career started was all about teacher compensation and higher education policy, but mostly focused on K-12 kind of policies.
Gary D Stocker (02:59.661)
Yeah.
Gary D Stocker (03:06.85)
Interesting.
Gary D Stocker (03:25.439)
When you were still an academic, and I know you went to the University of Tulsa, when you were still teaching at Tulsa a few years ago, step us through the early phases of the financial challenges that were at that university.
Matt (03:38.56)
Yeah. so I guess the first I heard that we had financial issues was in an email, that our retirement contributions were going to zero immediately. And which was kind of a shocker because my, yeah, I'd actually joked with some of my classmates from back in my PhD program that my school's got.
like endless amounts of money. It's got a huge endowment. It's got all this oil money from years back. And I was going to conferences and our school was paying for all of my stuff and they had to keep receipts and all this and I didn't. So I was like, I got a pretty nice gig here. And I always joke that it's oil money to them. And it was a funny thing when I'd see them at conferences. But then all of a sudden our retirement went to zero. So was like, Holy cow, that's like an abrupt change, right? And
Gary D Stocker (04:22.851)
Hahaha
Matt (04:32.75)
Yeah, there's a lot of like a lot of confusing things happen. Like there were rumors that we were going to build. We were building buildings like crazy. And then there was rumors that we're going to build a new building for my department among others. And so to have our retirement cut to zero just out of nowhere, it didn't jive with those rumors at all. So
Gary D Stocker (04:51.652)
Yeah.
Matt (04:56.258)
I decided to, well, actually at first I didn't, just kind of ignored it because I was really busy with my own research and last thing I want to do is spend any time. I wasn't tenured. I don't think when they first cut that. so a bit of time passed and then the administration came out with another plan, which was to basically cut over half of our academic programs. And then it really didn't make sense to me. like, wow, that's a pretty dramatic going from.
seemingly building buildings at will to, we've had a $1.4 billion endowment, I believe at that moment and only 4,000 students. So you just do a little bit of math there. You realize that we've got a lot of money that's just free, right? We don't have to generate a lot of revenue to support the academic side of the institution. So that's when I really got perked up. And of course it was very controversial, the plan.
And I said, I'm going to start looking at iPads and all the other data sets that are available for higher ed, which I'd never analyzed until this moment. And, um, yeah, thankfully there's great data in higher ed. In fact, in many ways better than in K-12, um, which was amazing. So I just started looking at the data initially. thought I was going to see, you know,
Gary D Stocker (05:58.368)
Ha ha.
Matt (06:16.267)
huge deficits and, you know, big spending on instruction and lots of waste. And that's why they're wanting to cut some of these programs and all that. But it out it wasn't that at all. I found results that were pretty shocking about spend that has nothing really to do with our mission, in my opinion, and then a lot of other people's opinion. And so sort of flagging some of those high numbers, high spending, high staffing and
That's sort of what started this this work that I do in higher ed.
Gary D Stocker (06:52.601)
So those financial challenges that you experienced firsthand at Tulsa eventually led you to found your own company. Tell us then the rest of the story as Paul Harvey would say.
Matt (07:04.781)
Yeah. So like I said, I started flagging things at the University of Tulsa and, putting spreadsheets together, sending them around campus, showing our spending versus peer spending. Um, and so it was clear to me right away that I need to have like a rubric of finding peers that's objective and not just handpicking peers. I have to have a decent sample size to make sure I'm not having like some weird outlier driving the results. Um,
So long story short, there was a lot of refinement that happened just through my own experience. So starting with the peer group, I was like, wow, we got great, you know, we have great machine learning tools to find peer groups. Let's use those. Right. So I adopted that early on. and then even that wasn't, perfect, right? There's always schools in the peer set that one or more people might think shouldn't be in that peer set. should drop them out.
Gary D Stocker (07:46.137)
Ht
Matt (07:59.968)
So I would drop them out and send out a new spreadsheet. And that was a very inefficient process. So was like, wow, there's gotta be a way where I can just like modify, like send out a live document in the sense where people can modify the peer set and on their own, right. And convince themselves, right. so that's when I decided to make a dashboard out of these. and so I learned how to make a dashboard in Python.
Gary D Stocker (08:15.876)
Yeah.
Matt (08:26.558)
And so that was the birth of the dashboards that you see on the shows. And they just, it just got more and more complicated and bigger and bigger as, you know, people ask more questions like what is, where's all this extra money going? One of the, one big advancement was, you know, we could see that there was a lot of money still being spent on stuff, but we don't know what it is. It's not staffing, doesn't appear to be staffing, but it's, we can see the spend.
Right. And so the big advance was to also bring in the nine nineties. And then you can see a lot more about the sort of the black box of for us. was all their money going to these vendors, right? These are huge vendor contracts and there's a lot of them. So that was another advancement that was built into the dashboard. So the dashboard just sort of like grew organically from there. And then, you know, by the end of this whole thing, I had a tool, right? I mean,
Gary D Stocker (09:23.555)
Yeah.
Matt (09:24.31)
Eventually the University of Tulsa got it together and quietly abandoned their plan to kill half of the school's programs. We got a new administration. They understood how to fix the place based on the work that I had done and other consulting firms that were brought in. And they did it, right? And it didn't require cutting half of our programs. What a shock, right?
So anyway, the at the end of all this thing I had my school was in good shape. Finally, I had a tool that I'd built and I was really enjoying working in higher ed and I was I felt really good about the data sets and the value that it could bring. So I started building more dashboards for other schools and then started pitching the dashboards to other consulting firms and was able to get a lot of interest just in those cold pitches.
and was able to get a contract that gave me enough resources to do this full time independently. of like a, really am in some sense feel like I'm in like a research position that's fully funded by a donor, right? In some sense, but I'm, you know, it's not exactly that I don't have tenure anymore, but it is very rewarding and I'm able to do what I really love.
Although I did love teaching as well, but there was a point where I was like, man, every time I get interrupted to go teaching, I feel like irritated about it in a sense, right? Cause I was like having so much fun working on, you know, this dashboard, developing that and, you know, working with schools, trying to help them out and just try to make everything more clear to everybody. I mean, it's really, that was the big need that I saw is just, really bad information at a lot of schools. I mean, there was.
And just stuff that seemed to just be made up, right? To justify the policies that the administration wanted to do. it just turns out it was totally not true, right? It's like, sounds reasonable, but a lot of it was just, know, guessing and assumptions that were never tested, which is not a good way to develop policy, right? Turns out.
Gary D Stocker (11:20.751)
You
Gary D Stocker (11:42.413)
Yeah. So Matt, Perspective Data Science, PDS, Perspective Data Science, is a company you founded in 2021. You just talked a little bit about what it does. What are your goals for Perspective Data Science this year and in the coming years?
Matt (11:57.036)
Yeah. So, it's a little different. I mean, it's a, it's a business, but it's really, it's a mission driven business. My mission has always been to try to, mean, one of things that kind of irritated at me at the university of Tulsa, like I said, as I did all this analysis, right. And I showed our administration what, you know, a path to get, financially sustainable. and then they hired a really, really expensive consulting firm to come in and basically say the same thing, which.
irritated me because like we could have saved hundreds of thousands of dollars, you know, and basically got the same result that already showed them. So I know it's a bit more nuanced than that. But at the end of the day, I was irritated because the cost of to get this information was so high, our administration didn't have it when they could have had it from these free data sets that I already had. So I made it my mission to just help schools get better information to make better
policy decisions and to do it in a way that they can afford it, right? Not every school can afford to pay $500,000 or a million dollars for, you know, somebody to analyze iPads and some other data sets for them, right? I want to do it for, you know, a price that any school can afford. And we do that right now. Right? mean, Gary, you know, the prices that we charge and it's totally reasonable. Yeah.
Gary D Stocker (13:02.627)
Yeah, yeah.
Gary D Stocker (13:16.687)
Right, So I want to talk in particular about some of the technology that you used, Matt. And you mentioned the Form 990s, you mentioned iPads, but a big part of what you've done in the last couple of years has gone out to get the audited financial statements. Talk a little bit about the process and a little bit about the technology and the value it brings.
Matt (13:33.87)
Mm-hmm.
Matt (13:38.934)
Yeah. So, yeah, going back to my time at the university of Tulsa, working with the finance people at the school, it was clear that iPads had some serious shortcomings. In fact, that's one of the ways that they tried to sort of, you know, push back against my research, which was that iPads is old. This one that you probably heard Gary, right? I've had this old and it doesn't have the variables that we, we need, right? Like we need operating revenue. We need.
Gary D Stocker (13:55.971)
Yeah, a times.
Matt (14:06.414)
We need to distinguish between restricted and unrestricted revenues. We need to have cash flow. Net cash from operating activities was a key one in the debate at the University of Tulsa. A very important variable to get a sense of like how healthy is the institution. Most of these institutions die just because they run out of cash, right? So you need to have a good sense of cash flow. So that was a bit annoying to me that that iPads was one kind of old, which was true.
and that was missing these variables. So what I did initially was just fill these in manually. I would go through the audits and fill in the cash flow numbers, fill in operating revenue so I can get net income and did that by hand. And I was like, man, this is a pain in the butt. So then of course, I got this partnership with a big consulting firm.
Gary D Stocker (14:39.597)
Ouch.
Gary D Stocker (14:55.309)
Hahaha!
Matt (15:01.982)
And the scale was becoming a problem, right? Cause I'm making, you know, dozens of these a year now, several dozen, and I can't be doing this by hand for a thousand schools and every time doing it every year. So I initially made a script in Python that would go through and look for keywords and then find the number next to that keyword and pull that number. that works kind of, although it's, not very
say robust, right? Any kind of changes to the keywords that are used or the formatting or the OCR, if it kind of fails for the document, it really doesn't work very well. So I had that, that was operational. But then I was like, man, I need something that is way smarter. And then here comes AI right out of the blue. And I'm like, okay, I got to learn how to do this because this could be applicable to what I did.
Gary D Stocker (15:51.439)
Ha ha ha ha ha
Matt (15:58.444)
because I didn't know it at the time, but eventually there's going to be an AI that can look at a picture, right? Look at a picture of a document and pull out all the stuff that you need. So it totally bypasses the, the need for OCR or anything like that. so I just started playing with that and, just to check if it would work and lo and behold it did. And I was like, wow. Okay. And then I made a big investment in, in hardware and I was off and running training these AI,
Gary D Stocker (16:11.673)
Yeah.
Matt (16:28.737)
to extract the information I need from audits and to teach itself essentially how to do that. I didn't have to say, oh, look for a keyword revenue, right? And look next to that. this doesn't say unrestricted or not. Like that, that was the kind of process before. Now I just say, here's, here's the page. These are the numbers that you should be extracting from this page. Go learn how to do it. So that's basically how the AI works now. And it's training right now. It's always training. have a training 24 seven. So
Gary D Stocker (16:53.612)
my.
Matt (16:58.464)
It's getting insanely good. As you've seen, I've made a database already, with most, almost all of, actually about all of the audit data I have now, I just haven't put it all into my database quite yet, but, that's coming probably in the next couple of weeks, but, but yeah, it's, it's pretty remarkable. It can just look at a picture of a page and I can say, pull out the, the asset section of the balance sheet and it will
They'll do it and pull it out in this perfect format and everything. It's great.
Gary D Stocker (17:31.097)
So I think remarkable is an understatement to describe what you did with that, man. But let's get to the products themselves. So you've got three basic products. One is a trial, and the other two are paid products. Talk about those.
Matt (17:44.674)
Yeah. So the, main products that I've been selling and that sustains my business is the, we call now the strategic compass. Gary helped me with that. was originally just called a benchmarking dashboard, just generic name, but we've got a brand now. so yeah, strategic compass. And that's the one I started developing probably back in 2018 or so out of necessity and has now morphed into this sort of giant tool that, you can ask it basically any question you want about a variable and compare yourself to your.
peer set and other schools and filter the peer set. There's a lot of different capabilities there. So we saw that tool. That's our most expensive tool, but very reasonable even for a tiny school of 500 students. You can afford it. It's definitely worth it. So you should check it out, but make that tool and a report and go through it with whoever purchased it. Purchases the tool, go through it and find detail with them. Show them the stuff that I see. Show them how to use it to find their own stuff.
they might want to focus on and generate dozens, usually several dozen policy recommendations out of that, right. There's, and things to check questions to ask, right. It's just, really sort of blows up from there. so that's one tool. the next is the advanced financial compass, which is basically a front end dashboard capability for the audit database that I just talked about. That's.
using the private school audits to build this huge database. The AI goes through pulls all this data for me and I put it in a database that has over a million values in it. Over 10,000 schools or school years, right? We have like nine years and 1300 schools or so. So it's become a really big database, that advanced compass allows the user to pick a school.
pick a bunch of peer set schools to compare to you and look at basically anything that's in the audit, compare yourself to the other schools over time. And it's a great way to identify areas at a high level financially that you need to focus on, You can look at things like, oh, how tuition reliant are we? Are we getting more tuition reliant or less? How's our endowment growing, right? Which is related to tuition reliance.
Gary D Stocker (20:09.615)
Right.
Matt (20:09.774)
If our endowment is not growing and everybody else is growing to like 70, 80 % over the last eight years, that's a huge problem for us because they will be able to undercut us on that tuition and we cannot compete. Right. You can look at what are some of the big ones? How much of your budget is going to salary and wages? Do we look normal? Are we spending way more on salary and wages than what looks appropriate compared to our peer set? Things like that.
Gary D Stocker (20:37.807)
Yeah, yeah. And then, sorry, excuse me. So here's kind of a trick question. Let's have some fun with this. So Matt, you view yourself as a higher education consultant or a data analyst?
Matt (20:53.614)
Yeah, that's interesting. I think initially I thought I was just gonna be a data person, right? Like I'm highly trained in econometrics. That's what I've done for over a decade now, or I don't know, 20 years probably at least. And, you know, that's what I do. I just do this all day long. I code, I look at data.
Uh, but sort of what's happened is I'm also an economist and, and, and part of being a good economist is understanding what the data is saying, what the policy implications are. So I've been training that a lot more than I had even realized, right. And I wrote papers on policy, right. When you, when, when the data tells you a clear story, it's pretty easy to talk about policy. Right. And so in almost all the papers that I wrote, I tried to push it to the limit all the time. Right. Like, what is this data actually telling us? Oh, it's telling us we should change the teacher's salary schedule. Right.
And for free, we can get better teachers into K-12, right? And keep them longer. That's what it says. So I tried to pull that into everything that I did. And yeah, it's interesting when I made this partnership with a big consulting firm, I was thinking I was just going to sell them the apps and then they've got really smart people that will use the app, right? And I don't need to really talk to them. And very early on, they're like, can you also do your reports and talk to our teams?
Gary D Stocker (22:07.78)
Yeah.
Gary D Stocker (22:14.638)
Yeah.
Matt (22:16.966)
And so that's what clearly then they viewed me as a consultant. so I've embraced that and it's helped me, right? It helps me also learn the story behind and what's possible on the policy side and motivate those kinds of policy changes with the data.
Gary D Stocker (22:37.965)
And then the final question, is for those who don't know, you and I have a strong working relationship. Tell the audience how that started and then about the college financial health show that you and I have been doing for about eight or nine months now.
Matt (22:51.598)
Yeah, so somehow I started seeing your posts on LinkedIn and I was like, oh crap, I got this competitor out there, right? His name is Gary Stocker and he's got an app too. I was like, I thought I was the only one doing this and and I really liked your posts like a lot almost everything that you said I would agree with and like I believe. And so we just kind of got connected through LinkedIn as far as I could remember and.
eventually one day some posts I think triggered both of us to say we need to meet and we met and and the rest is history right we decided to do a podcast right and we did one podcast or yeah one show and and then I can't remember what it was about but I think then you were like we should do one per week and I was like sure I'll do one per week and and then sort of got it yeah we've been doing it
Gary D Stocker (23:34.381)
yeah, the show, yeah.
Matt (23:50.282)
Ever since, right. And now we're really, I think on the same page, same mission, essentially to help folks use data, make it more accessible to everybody, not just the administration board members, but also, you know, students, parents, families, that kind of thing.
Gary D Stocker (24:10.457)
So as a wrap, my very special guest has been Dr. Matt Hendricks, who has founded the firm Perspective Data Science, and he's talked about all the work that he has done before. And I'll get on the soapbox for a moment on this podcast. And for those of you that follow, know that I do that on occasion. If you need data, if you need analytical data, if you need comparisons, Matt and I talk about comparisons all the time. You want to look at the products Matt talked about, and even the ones that you see I post online as well.
It's the moneyball era. We talk about that, on the show all the time. So Matt, again, I am grateful for your time. Those of you who made time to listen to the podcast, thank you very much. And I'll be back next Monday with a regular edition of This Week in College Viability.