That Annuity Show
Returning guest Barbara Matthews, Founder and CEO of BCMStrategy, Inc joins us for a timely discussion of the public financial policy and the ability of machine learning to separate signal from noise. We cover crypto-currency intermediation, SVB, interest rates, COVID subsidies and the early insight that her machine learning model provides.
Links mentioned in the show:
https://www.linkedin.com/in/barbaracmatthews/
Listen
Watch
Receive Updates
Show Sponsors
Episode Transcript
The discussion is not meant to provide any legal, tax, or investment advice with respect to the purchase of an insurance product. A comprehensive evaluation of a consumer’s needs and financial situation should always occur in order to help determine if an insurance product may be appropriate for each unique situation.
Paul Tyler:
Hi, this is Paul Tyler and welcome to another episode of That Annuity Show. And Ramsey, good to see you.
Ramsey D Smith:
Always good to be back.
Paul Tyler:
Yeah, well, we had a great guest on, and I’m looking at the date, this was December 28th, 2020. I’m thinking that was the deep,
Barbara C. Matthews:
Pandemic.
Ramsey D Smith:
Is that, is that
Paul Tyler:
dark,
Ramsey D Smith:
really the last time?
Paul Tyler:
that was the last time we were in the, like, throes of the pandemic. And we were talking about… policy, monetary policy, how to read it better, how to under interpret it better, and oh this thing called machine learning and like how to take un-text and turn it unstructured data into data. That isn’t the news at all, but fast-forward to the day we actually do have incredible policy issues we’re all kind of dealing with in the financial and economic issues that we’re dealing lot of different issues today and machine learning is suddenly on the tips of everybody’s tongue. So Ramsey, do you want to do the intro and set up the discussion?
Ramsey D Smith:
Sure, first of all, I’m just blown away that it has been that long. And again, it’s another sign of sort of the pandemic time warp. I can say that as we were thinking about just what a crazy macro environment we’re in right now, Barbara Matthews, our guest, came to mind very quickly as one of the first people we wanted to talk to to help us decode everything that’s going on. So she joins us today again. So great to have you back, Barbara. founder and CEO of BCM Strategy, Inc. And we look forward to having a great conversation about any number of things that have been going on, certainly this year, and importantly, helping us sort of better understand what we’re being told. How should we interpret the messaging and the words that we’re getting? So welcome back, Barbara.
Barbara C. Matthews:
Well, thanks for having me back. You guys ask great questions and this podcast is just fabulous. I’m delighted to be back.
Ramsey D Smith:
All right, so let’s get into it. So we were ahead of the show, we were chatting a little bit about some of the many things that have been going on and one of them was really around Fed policy and bank solvency. Those are two separate issues, but obviously they’ve also been pretty closely related in recent months as well. I remember sort of as there was a lot of messaging coming out of the government handle each of those issues, trying to figure out, well, everyone’s trying to figure out what are they really going to do, and now we know. Help us understand how you looked at that whole process and what were the things you saw as things were unfolding.
Barbara C. Matthews:
Absolutely. Thanks for the opportunity. It might be helpful for your listeners to explain what we do. Because
Ramsey D Smith:
Sure.
Barbara C. Matthews:
people are accustomed to thinking about looking at any policy suite, and they expect to hear a human provide analysis, or do their equivalent of the crystal ball. And we do something very different. And so it might be helpful to level set for a moment.
Ramsey D Smith:
Absolutely.
Barbara C. Matthews:
So what my company does, we have a system have patented technology that measures the momentum and the volatility in public policy. And so our machine reads every day more than any human could in a 24 hour period in general, but then it puts a number on it and it measures it without using sentiment analysis. So we’re measuring, if you will, if you think about the old where there’s smoke, there’s fire. So we’re finding the smoke. numbers from language, that’s what the patented process does, it turns words into numbers, because we’re doing that predominantly, but not exclusively, from the official sector. It means you can also see the difference between what is going on in the media, what the journalists are reporting. We do take in, we have data mining licenses with Thomson Reuters and Dow Jones, strategic partnership with Dow Jones. 95-90% of our inputs are what policymakers are actually saying and doing, because we think what a policymaker says matters. It matters a lot. And they will tell you what they want to do and what they’re going to do. You just have to learn how to listen. And because I have been in public policy, I have trained my machine to listen for the signals hiding in plain sight. policymaker. Our machine listens like a senior policymaker. And so that’s why we are able to identify a number of signals instead of just repeating back to everyone the echo chamber of what’s happening on Twitter and in the headlines. So your question specifically was about regulatory policy, monetary policy, and I can give you some examples from our own. We started a substack podcast this year in January. I’m inspired by you guys.
Ramsey D Smith:
Tell us the name of it, please.
Barbara C. Matthews:
It’s measuring policy volatility.substack.com. So Fridays, it’s once a week. So Fridays, it’s digital currency policy. Saturdays, it’s climate finance policy. Sundays, it’s monetary policy. And we use our data, we read our data, we point out to people what our data shows you. And so what that meant was, because we started in January, January, we identified right away, I think we were the first to identify, that the new regulatory stance of the federal banking regulators was going to dramatically decrease liquidity for cryptocurrency intermediaries. They issued a big statement January 3rd, our system caught it, I podcasted about it. They said effectively, they actually said it. You
Paul Tyler:
Cue,
Barbara C. Matthews:
don’t
Paul Tyler:
cue,
Barbara C. Matthews:
actually
Paul Tyler:
cue.
Barbara C. Matthews:
have to outlaw crypto intermediation. All you have to do is say, hey, heads up, bank examiners are gonna view this as an unsafe and unsound banking practice, and that just sends a chilling environment. And then they did it again in February.
Paul Tyler:
Okay, so maybe just stop there and
Barbara C. Matthews:
Yeah.
Paul Tyler:
we’ll just make sure everybody who is listening hears this, because what you said is really significant. Disintermediation, what was the word you used? How do you describe the
Barbara C. Matthews:
Chilling effect.
Paul Tyler:
chilling effect of the cryptocurrency intermediaries? And this is interesting, what makes our currency work? I mean,
Barbara C. Matthews:
Yep.
Paul Tyler:
I should ask you, but my definition is you’ve got a central bank, I’m in here that sets monetary policy and controls lending rates, sort of makes sure a currency is stable. You don’t have that in cryptocurrency. Now effectively some of those companies were, right? Coinbase, you know, go through the list. They effectively were market makers, Barbara. Am I misinterpreting this or
Barbara C. Matthews:
No,
Paul Tyler:
would you say it better than I did?
Barbara C. Matthews:
not at all. And that’s actually a great point about the market makers. Because although most people think of the crypto space, and it’s not just crypto, it’s also stable coins that maintain a one-to-one peg with the US dollar. But these are not hermetically sealed environments. There are on-ramps and off-ramps to the US dollar, and that goes through the banking system. In addition, They actually need US dollars or some hard currency to buy the computers, to pay the people. You know, people still need some kind of hard currency to pay for food. You can’t use crypto to pay your grocery bill, your utilities bill, your mortgage or your rent. So there are on and off ramps between the crypto space and the banking system where there are US dollars. you know, our system basically set up this big alert, said, heads up, you know, it’s an unsafe and sound banking practice. We’re going to see a constraint on access to credit for a range of crypto intermediaries and market makers, like you said, like, and, you know, and sure enough,
Paul Tyler:
Yeah.
Barbara C. Matthews:
you know, Coinbase has been in the news. Not a
Paul Tyler:
Right.
Barbara C. Matthews:
surprise. They’re not the only ones, but, you know, Binance is having trouble getting their bank account. I mean, it’s an issue.
Paul Tyler:
So Ramsey is a client of mine, I’m Coinbase or competitor. Ramsey says, Paul, I wanna cash out my Bitcoin, one of my currencies, but I don’t have somebody on the other end of the transaction. I’m gonna have to borrow money to pay him, correct? And
Barbara C. Matthews:
That’s
Paul Tyler:
where do
Barbara C. Matthews:
what
Paul Tyler:
I borrow
Barbara C. Matthews:
you could, yeah.
Paul Tyler:
the money? So that’s the point where if I’m Coinbase and I want to go get money, you’re saying policy makers said, maybe charge Paul with more money. how it happened.
Barbara C. Matthews:
Yeah, well, so it was kind of banking regulation is this. It’s a lot about nudge. And there’s a whole field of economics associated with nudging. But basically, the the policymakers said was really, we’re going to view with suspicion and we’re going to think it’s an unsafe banking practice if you provide intermediation services. So that’s deposit act. Intermediation is broad. It’s deposits. You you brought up the lending use case. of some kind of a loan, you know, it’s a big deal. And then they did it again in February.
Paul Tyler:
February.
Barbara C. Matthews:
They did it again in February, the second
Paul Tyler:
So you
Barbara C. Matthews:
time.
Paul Tyler:
saw it in January, you saw it in February,
Barbara C. Matthews:
Yeah.
Paul Tyler:
and one of those blow
Barbara C. Matthews:
Yeah,
Paul Tyler:
up.
Barbara C. Matthews:
so fast forward then to March, a crypto intermediary, Silvergate, declares bankruptcy. They’re like, we can’t. We’re done. We’re done. And it’s not like the crypto industry was doing well. I mean, last year was legendary in terms of the number of implosions. And that’s when the monetary policy problem kicked in. Because when they went down, the Silicon Valley Bank And then on top of it, you know, the crypto space has been financed in large part by a lot of venture capital companies and venture capital individuals, all who banked at SVB. And they’re like, oh, we want our money. We want our deposits back. And the reality is, is that no bank can actually withstand a run period. Doesn’t really matter. I mean, it wasn’t quite a run yet. SVB was like, okay, fine. We’re going to liquidate our secure our treasury securities. because those treasury securities weren’t worth as much as they were when they originally bought them. Why were they not worth as much? Because interest rates started going up. So the value of a sovereign fixed income instrument, the value of it decreases over time when interest rates go up, because someone can buy a new bond with a higher interest rate. So the value of your old bond with low interest rate kind of goes to the… Then this is super important in the annuities business. this, of course, I don’t have to tell you that. So there’s SCB. They start liquidating their Treasury securities at a loss, but this only spooks the market more, accelerates into a full-on deposit run. And then for good measure, the FDIC closed Signature Bank at the same time, which was the other major intermediary in the crypto space. There’s a massive implosion in crypto, banking system. And then the question became, because this happened right, like two weeks before the next monetary policy meeting, the question became, will policymakers raise interest rates again in the middle of a financial stability situation? So we looked at the data. We looked at the language data. We looked at what policymakers were saying. We also looked at actual data. So this is something an analyst could do. And part of what we’re doing is we’re people how to, you know, it’s a leap for people to think of words as data, words as numbers. So we’re teaching people in Substack how to do this. And so we were like, well, look, you know, you guys, our audience is a lot of portfolio managers, a lot of advocates, portfolio managers, they’re going to do some research. I was like, oh, well, let’s see. Let’s see what the utilization rate is for that support structure that they created, the bank term funding program. Let’s look at the support structure they created to provide dollar liquidity to the international financial system. Well, utilization rates were really high. The market had calmed down. So we put the actual data together with the language data, and we told people on the Sunday, yes, the Fed is going to raise interest rates. And not just them. I mean, it was all the central banks, actually. They all did the same thing. But we used the data to tell them to say, look, you have to listen to what they’re telling you. And what they’re telling you is they want to hold firm. inflation, they’re still worried about a hot labor market. If you only looked at Twitter, if you only looked at social media, I don’t want to single out just one company, if you only looked at the headlines, the echo chamber, none of this really made it into the news. Reporters can only report so much. and illuminate a signal. Because
Ramsey D Smith:
So.
Barbara C. Matthews:
you can see it mathematically, you can see it jump, you can see the volume go up. And you’re like, oh, well, gotta pay attention to that.
Ramsey D Smith:
So where are all the various sources that you’re drawing your verbal data from?
Barbara C. Matthews:
Yes.
Ramsey D Smith:
Is it just direct policymaker language? Is it also Twitter? Is it also the mainstream media? What is the, yeah,
Barbara C. Matthews:
Great questions.
Ramsey D Smith:
where do you start with to filter down to what you think is the true signal? Because if you’re looking at everything, you’re gonna pick up some of the bias signals as well, right?
Barbara C. Matthews:
Yeah,
Ramsey D Smith:
filter this out.
Barbara C. Matthews:
exactly. Well, we start from the proposition, like I said, that what policymakers say matters, and people don’t
Ramsey D Smith:
Yeah.
Barbara C. Matthews:
hear it enough. And we are very committed to not having bias. On occasion, people do ask us to interpret or provide some normative analysis, and I resist that. I’m one of the few startups, I think, that actually turns down business, because I don’t want to corrupt the data. So we just take the language from policymakers, it’s publicly available. And I’d say that’s easily 85, 90% of our inputs. And that’s global. So every day we take a measure of what policymakers are saying around the world on the same issue. We do have data mining licenses with Thao Jones, who’s a strategic partner for us, and Thomson Reuters. And so what that means is that we’re not just generating one number. And for the quants in your portfolio managers and your listener base, what we actually generate a multivariate time series that permits the user to compare what policymakers are actually talking about with what major media is reporting. And the delta, the difference between the two, is a measure of your informational advantage when rhetoric media coverage is low, but policymaker activity is high.
Paul Tyler:
So it’s a little bit like Google Trends meets arbitrage.
Barbara C. Matthews:
I guess so. I guess
Ramsey D Smith:
I mean,
Barbara C. Matthews:
so, yeah.
Ramsey D Smith:
I guess my.
Barbara C. Matthews:
Yeah, but I don’t know exactly how they do Google Trends, so I don’t want
Ramsey D Smith:
Yeah.
Barbara C. Matthews:
to overdo
Paul Tyler:
Yeah, okay,
Barbara C. Matthews:
- But it’s
Paul Tyler:
okay.
Barbara C. Matthews:
a principle you get. You can compare different activity streams as it works.
Ramsey D Smith:
So my question is that, you know, how much does that delta vary over time? And
Barbara C. Matthews:
It’s
Ramsey D Smith:
do
Barbara C. Matthews:
really
Ramsey D Smith:
they ever meet?
Barbara C. Matthews:
interesting.
Ramsey D Smith:
Are they ever on top of each other? Is the delta
Barbara C. Matthews:
Yeah,
Ramsey D Smith:
ever
Barbara C. Matthews:
you know
Ramsey D Smith:
zero?
Barbara C. Matthews:
they
Ramsey D Smith:
Okay.
Barbara C. Matthews:
cross they cross I have come to the conclusion that when they when the when the lines cross We’re at an inflection point So for example if action levels are going up and they exceed rhetoric levels and it’s intuitive if you think about it But we patent we’re the first ones to do it and we patented the process to attach the numbers No one else can do it so I can tell you about it It was intuitive policymakers act and what happens next? So the dynamic, when something is really going on, action levels are gonna go up, and then it takes, depending on the issue, anywhere between a day to two weeks before the media coverage spikes. And conversely, but then there are some other patterns that are really interesting. So in digital currency, media coverage is always really high. you and persistently, I mean we’ve been generating this data since 2019, media hardly reports on central bank digital currencies. PS is a major competitor to the cryptos and they’re only just starting to get to a point where they’re actually going to compete in the market. So there’s this all this universe and we do that we do that and Ramsey I think you you’ve seen some of these charts. So The media coverage will be all over here for crypto. And in the meantime, policymakers are doing a lot of stuff to compete with cryptos by issuing, preparing to issue sovereign digital currency. And it’s like not in the news. Now, having said that, there are some crypto-specific news outlets that do a good job of covering this. But if you’re not a crypto fanatic, you have your pension and you have your annuity and you’re paying attention and you just want to know that you’re you’re and you want to engage in an intelligent way with your asset managers. You’re not going to be reading the crypto news coverage. You’re going to be reading the Wall Street Journal, Barron’s, you know. Anyway, so we’re measuring what everyone’s talking about.
Ramsey D Smith:
So then looking out sort of over the horizon a bit, right? So what are the some of the things that we should keep an eye out for? Where there’s that meaningful delta between what’s being talked about generally and kind of what we might expect? So central bank digital currencies, that’s interesting. I still remains to be seen whether or not they’ll be successful, I don’t know. But the first part is like, what’s the level of intentionality on the part of the central banks to actually try to make them successful? And then the next thing is inflation interest rates. Do you have some thoughts on any of those three?
Barbara C. Matthews:
All of the above and climate finance too, but anyway, all the above.
Ramsey D Smith:
All three, okay good. Oh,
Barbara C. Matthews:
Well, so you’re right, but all of the major reserve currencies now have very significant
Ramsey D Smith:
all right.
Barbara C. Matthews:
pilot programs underway. And they’re thinking very concretely, the ECB has promised they will have a decision in the autumn of this year about whether or not they’re going to try to issue a sovereign digital currency. I think the central banks are very serious. great job of exposing all of the faults and failings and vulnerabilities and frailties of the system. So yeah, pay attention to this because, you know, even if you’re invested in the FX market, you know, you wouldn’t price against it right now, but if you know it’s coming, there will come a moment when you’re going to want to pounce. You don’t want to miss that moment. You want to be in early enough where it’s smart but not so early that it’s, you know, risky. The monetary policy. I have taken in the podcast to putting together the language data related to financial stability and the language data related to inflation. And so right now we’re in the middle of the IMF World Bank spring meetings, G20 spring meetings. I will tell you, today’s a great week to be generating language data. I can’t exactly answer your question today because it’s only Tuesday. We’ve got a bunch more language data that has to come out. But the economic growth rates that the IMF released suggests strongly significant economic slowdowns in a lot of advanced economies. And that’s of course what the Fed wants. And what every central bank wants, they want to, ironically enough, they want to slow down the growth rates so that the pressure on prices comes down. And so I will be listening very carefully and more importantly, my system will be listening very carefully what they’re saying about whether financial stability issues or monetary, you know, the inflation rates are the driver. And then as a bonus, I’ll tell you that bank term funding program, it’s slated to end at the end of April, and the next monetary policy meeting is at the beginning of May. So I am personally, and this is just because I’m a geek, and I am at heart an analyst matter expert as well. Personally, I’m going to be watching like a hawk utilization rates. They’re published every week by the Fed. And because that’s what’s buying us financial stability. And so I will be looking at do they renew the program? Have utilization rates gone down? And then how are policymakers talking about financial stability? It’s enormously
Ramsey D Smith:
So do you think that there’s some likelihood that it will be extended? It’s just, I guess, for the audience’s
Barbara C. Matthews:
I don’t know
Ramsey D Smith:
benefit.
Barbara C. Matthews:
the short answer is I don’t know. So this is the thing about the language data.
Ramsey D Smith:
Yeah.
Barbara C. Matthews:
It’s not exactly a crystal ball.
Ramsey D Smith:
Sure.
Barbara C. Matthews:
The crystal ball comes from applying that data in machine learning artificial intelligence. We could have a really good conversation about that.
Paul Tyler:
We’re going to.
Ramsey D Smith:
Okay.
Barbara C. Matthews:
So I will tell you the narrow question about, will they raise rates or not? The last time I looked at the language data before Easter. Spoiler alert, Easter weekend, Christmas, New Year’s, not even worth it to look at it because nobody’s doing anything anyway.
Ramsey D Smith:
Yeah.
Barbara C. Matthews:
Activity levels are low. So last time I looked at it was right before New Year’s. And I was at the IMF World Bank. I’m a member of the Bretton Woods Committee. I was at the Bretton Woods Committee session on climate finance yesterday. You know, I’ve been at these events, I’ve been at these meetings foot. It does not have that vibe at all. There is, if anything, and this is just totally reading between the lines, I have the impression that so far a gentle decrease in growth rates, most economists will view that with a bit of a sigh of relief and they’ll say, okay, maybe for good measure they’ll increase one more time just to solidify the trend. and then stop. Because the mood in these meetings so far, it’s really still in the week, but the meetings so far, there’s no panic. In fact, when the IMF managing director yesterday kicked off the climate finance sessions at the Bretton Woods Committee, and this was a public session, broadcast, when she kicked it off, she ended with kind of striking note, she said, we can survive inflation. We can survive a recession. So, head of the IMF, you’d be thinking that way, is stunning to me in general. And then the third part of that was, but we cannot survive climate change, very dour.
Ramsey D Smith:
Is that how it went? It went
Barbara C. Matthews:
That’s what she
Ramsey D Smith:
A,
Barbara C. Matthews:
said,
Ramsey D Smith:
B, and then C? Wow.
Barbara C. Matthews:
A, B, and then C. And that tells
Paul Tyler:
Interesting.
Barbara C. Matthews:
you, that sequence tells you a lot too,
Ramsey D Smith:
Sure does, yeah.
Barbara C. Matthews:
And then the last part of it was to make the case for informed policies that ensure we can both survive and thrive despite a shift in the climate. I thought it was enormously interesting, not just as a rhetorical device, but for those economists, and I’m sorry to stray into the substance here, but for those economists that believe the necessarily requires higher pricing, not just higher pricing for carbon, but just higher energy prices in total because it’s not as efficient. It’s, you know, it is more expensive. It may be renewable, but it’s more expensive and it’s distribution issues are not anyway, but either way it’s gonna be more expensive. To have the head of the IMF thinking in those ways, I thought was just illuminating.
Paul Tyler:
Well, maybe we just double click on that topic because climate is one of these issues that has massive ramifications, especially with the financial services market. I put climate with ESG trends. We’ve had a lot of noise, Barbara. Read the papers, states taking on companies for investment policies, controversy back and forth. If you looked at the policy statements probably the leading question based on what you just said. Are we serious about making these… are these changes going to happen or do you see noise in the policy in terms of how we may implement some of these climate mandates, climate directives that we’ve seen coming out in the last few years?
Barbara C. Matthews:
Well, that is a leading question. Well, you know, I worked in Congress, and I worked at the Treasury Department, and I worked at the State Department. I’m gonna tell you, there’s always noise in public policy. Always, always. But you are right to ask the trend question. One of the things that I like to do is I like to see where the money goes. It’s because I’m a Treasury person, what can I tell you? Where’s the money going? You know, all of the billions, if not trillions really, in subsidies under the Inflation Reduction Act and the COVID Relief Act in Europe are, as they deploy into the economy, yes, they are inflationary. So the title of the act is a complete misnomer. But they are going to change the landscape for energy and not just energy, but consumer vehicles and really all transportation. that that is permanent. I think there’s a bigger, within finance itself, we have much harder issues that the central banks have been grappling with. This is actually where we do generate data. It’s a very niche area though. If you want to calculate the net present value of an asset, and then you want to identify your risk around changes in that value, you need to have some assumptions about the future. And this is a tremendously difficult challenge that pits finance again. Even the finance people that want to be forward-leaning here, it pits them against a lot of activists. Because the science, math-based process for estimating risk of loss… is very different from a political promise, we will decrease temperature rise. As a former policymaker, I know it’s important to be ambitious, but that’s such a, you know.
Ramsey D Smith:
ideas versus
Barbara C. Matthews:
Policymakers
Ramsey D Smith:
execution.
Barbara C. Matthews:
used
Paul Tyler:
you
Barbara C. Matthews:
to like make promises they could actually deliver on. This is a, you know, it is an ambitious thing to say and then work backwards from there. So anyway, this is going to be a jagged line. Policy is path dependent, but it is not a linear path. And we might want to talk about that another time, just because the climate issues are and even the process of disclosing to investors and the role that investment advisors have in either servicing the needs of their savers that seek to have a forward-leaning, creative positive incentive, the market for green bonds. There’s a universe in here that language. And so we measure when there’s momentum behind the language.
Ramsey D Smith:
So
Paul Tyler:
Play of… Yeah.
Ramsey D Smith:
Paul, I know you wanted to talk about chat GPT. And so I think we’re probably going to
Paul Tyler:
Yeah,
Ramsey D Smith:
do that
Paul Tyler:
we’re
Ramsey D Smith:
quickly
Paul Tyler:
probably
Ramsey D Smith:
before
Paul Tyler:
there.
Ramsey D Smith:
we run out of time. Yeah.
Barbara C. Matthews:
Yeah.
Paul Tyler:
Yeah, well, let’s open the hood
Barbara C. Matthews:
Yeah.
Paul Tyler:
a bit, and I’ll kind of set this up for machine learning. We’ve all been, it’s been around forever. I mean, forever since like the 80s. You know, it was, probably in my mind, machine learning was, gee, why do I see the orange button on the website versus the blue one? Oh, we’ve kind of trained the machine to figure out which people click more on. recommendations, product recommendations. Oh, I kinda like that. Netflix comes along, wow, that actually was a good movie. Ramsey must have watched it and it showed up in my
Barbara C. Matthews:
Thank
Paul Tyler:
feed to, oh
Barbara C. Matthews:
you.
Paul Tyler:
my God, what’s happening here? But machine learning is very basic, is about training. And I think of Google as a product, I’m trying to think what it’s called, now they’ve changed the name a couple times, but you go to a website and I wanna do a registration. Google has this great free service for me to put up This is not a bot coming in, and I have to go and check all these pictures and identify fire hydrants. Now,
Barbara C. Matthews:
Yeah.
Paul Tyler:
isn’t Google using that to basically train their driving service, Barbara? Isn’t that sort of the
Barbara C. Matthews:
Yeah,
Paul Tyler:
guts of this?
Barbara C. Matthews:
oh yeah, absolutely. And by the way, it’s not free. It’s not free. It is only free in the sense that you have not provided them cash or crypto. It’s not free. You’ve given them two things that are super valuable. One is your time and the second is your knowledge. Your knowledge. So you’re right, machine learning. Absolutely. It’s pattern recognition. It’s statistics. And so But the better the pattern recognition, which is intuitive actually. There’s a lot of very fancy language and very complicated computer architecture that does it. But at its core, it’s just pattern recognition. And it can be very personalized, so that’s the third thing that you’re giving up is your privacy. Because it’s not just Google. The, you know, you’re giving them a lot of information. The value proposition to use, you’re going to get back really good automatic recommendations. But you’re also letting them use your data in the pool for others. And so you can also see a center of gravity. The downside is if you are an outlier for whatever reason, you’re just going to be funneled into the medium and the median. You know, for some people that’s not optimal. The other issue with the training is bias, potential bias, and there’s a lot to talk about there. But when CHAT GPT kind of made a big at the start of this year, people started seeing that you could actually use it as a research assistant. I mean, that is tremendous. So it can go out and it can read everything. And this is why I said wait into this off of bias. Because you have to really know what the model was trained on. So people laugh that there are times that chat-chip-y-t will give a crazy, erroneous answer. Well, that’s not because there’s a problem with the machine necessarily. That’s because there’s a problem with the training data. It trained on the wrong thing. So it does really well, systems for over a decade have been doing really well on medical documents because medical documents don’t really have your medical research that’s not very normative it’s not very subjective it’s very science based the minute you start talking about other things that are more subjective it becomes a lot harder so ask Chad GPT to tell you about the Soviet Union depending on your political priorities and your perspective what you get back you may That’s a challenge when you think about public policy, which is very values-based. it becomes even harder. Which is why when we set out to do what we do, we deliberately did not include any normative filters at all. We are only measuring momentum. And we don’t tell, you know, we don’t tell the machine, well, this is a good thing or a bad thing. But there are people who will, they’ll use sentiment analysis to tell you, well, and then to see the challenge with training data. So there are many people who will just kind of say, okay, fine, we’re going to, policymakers say matters, so we’re gonna take sentiment and we’re gonna figure out are they feeling good, are they feeling positive, and monetary policy, are they feeling hawkish or doveish? And they kind of missed the boat, having written a lot of these speeches for myself and for various ambassadors, cabinet level people, chairmen of those committees on Congress. The formula, I love it, I love it, I love it, I love it, but I’m sorry, I’m going to do the opposite. Sentiment analysis is gonna say, oh, they loved it! Conversely, it’s a challenge, it’s a problem, there are risks, do it anyway. Sentiment analysis will tell you, oh, they’re negative on this. So we chose not to use any sentiment analysis. It’s a unique decision. Not a lot of people in the industry have taken that route. There’s a lot that we could talk about for other kinds of bias. Geolocation data, mortgage rates data.
Paul Tyler:
you
Barbara C. Matthews:
that you’ve got to be really, really, really, and then if you take the position, and many in finance do take the position, that for example, many types of mortgage lending have been biased for a long time. If you train on the market data, you’re just gonna perpetuate. And that’s the last thing about public policy that I think is really important for people to understand. The purpose of public policy is to make a change. It is to create a break in the time series to do things differently. And that’s why the signal matters. And that’s why when you’re training data in the mortgage market, you’ve got to really think about whether you want to stay with the trend, even though in finance, the trend is your friend. Sometimes it’s not going to be. And when the trend conflicts with the policy makers you’re saying, and you go with the trend, you are setting yourself up Big risks to be on the wrong side.
Paul Tyler:
Let’s see, Ramsey, how are we doing for time here?
Ramsey D Smith:
I think we’re actually over the allotted time, unfortunately.
Barbara C. Matthews:
So sorry.
Paul Tyler:
Yeah, no, Barbara, we’ve gotta talk more. A lot of questions on that topic for you, but you’ve got a very, if I were to net it out, tell me if I’m right or wrong, you’ve got a very unique set of data that has been trained by some people who really understand how to mark up or make sense of the words. And it must be unique. You must be in a unique position in the marketplace at this point.
Barbara C. Matthews:
think we’re pioneers. We’re on the innovation frontier and that’s an exciting place to be. We are what happens when policymakers understand how to use the technology. That’s how I think about it. And we don’t actually mark it up. The machine marks it up automatically. You know,
Paul Tyler:
Yeah.
Barbara C. Matthews:
there are companies who pretend they’re in AI, but what they really have is an army of graduate students in the back room marking up text. That’s not what we’re doing. Our system actually marks
Paul Tyler:
Interesting.
Ramsey D Smith:
to.
Paul Tyler:
Okay.
Barbara C. Matthews:
also ask great questions. I’m happy to come back whenever you like. And
Paul Tyler:
Thank you.
Barbara C. Matthews:
I’m a great I love your podcast. So I’m happy to listen in as well. Thanks for having me.
Paul Tyler:
Excellent. We’ll put
Ramsey D Smith:
sure.
Paul Tyler:
all the links to your show and your sub stack in the notes. Ramsey, thanks.
Ramsey D Smith:
Pleasure.
Paul Tyler:
It was great. Barbara, love to have you back and continue these discussions. So anyway, listen, give us feedback. We love it. And join us again next week for another episode of That Annuity Show. Thanks.
Barbara C. Matthews:
Thanks so much.
Recent Comments