Big Data's Lens Into the U.S. Economy

from surveys of households and businesses. Now, these surveys tend to be several times bigger than say, political polls, but they're still samples of the broader population. But what if we could measure consumer spending by looking at every single purchase that Americans make, or look at every small business's bank account to analyze cash flow. One think tank is trying to do that and potentially reshaping how we look

at the economy. Welcome to Benchmark. I'm Start Landman, economics editor with Bloomberg used Washington, and I'm Daniel Moss, columnist at Bloomberg Opinion in New York. Don't get us wrong. US economic data may be based on surveys, but it's the gold standard for surveys. Yet they're still subject to regular revisions and often face questions over whether they've properly adjusted for seasonal events such as holidays, and they often don't fully capture what's happening on a week to week,

let alone day to day basis. Now we have with us in our DC studio a person who's using a set of big data to explain trends in the economy and also to find new ones. Diana Farrell is the founding president and chief executive Officer of the JP Morgan Chase Institute, a position she's held since Previously, she was a senior partner at McKinsey and Company, where she was the global head of the McKinsey Center for Government and

the McKinsey Global Institute. She served in the Obama administration as Deputy Director of the National Economic Council and Deputy Assistant to the President on Economic Policy. Dianna, thanks for being with us on Benchmark. Thank you so much for having me so. First of all, can you tell us what exactly is the JP Morgan Chase Institute, why did the company started, and what we're in are your goals.

I'd love to UM. The JP Morgan Chase Institute is a relatively new initiative of the Bank UM and the best way to think about it is it's a think tank and it's trying to do original economics research, but with a twist, which is much as you started to introduce it. Instead of relying on the typical surveys that are designed to answer specific questions on the economy and then aggregate up to a view of the economy, we start with the actual choices that people make as evidence

through the financial transactions. So think about the credit card, debit, card, loans, UM, any number of finacial transactions that come to the bank as the window that we have, and it's a pretty extraordinary window. Consider that the JP Morgan chases two and a half trillion dollars balance sheet worth of financial transactions. That's you know, over half of US household UH, two and a half small businesses UM and a very large

number of institutional investors. If you think about that, is the window on economic activity not based on what people say they are doing or are going to do, but

on what they are actually doing. That is the basis for us to try to explain some of the things that are transpiring in the economy, especially as the economy evolves UH compared to the way some of the original statistics were developed, so that we can inform what is the economic and financial well being of households of small businesses? What new developments do we see in labor markets, in

out of pocket healthcare spend at the city level? Do we see, you know, with our very high frequency and granular data, patterns of economic vibrancy that cannot be seen through some of the other data. UM. Small businesses themselves as a sector are really important, and increasingly we're turning not just to the Chase platform to understand the economy, but the JP Morgan platform, which are more the finacial

transactions of institutional investors UM. So short view is to say, can we take this extraordinary window, which are the financial transactions of JP Morgan Chase to inform important discussions that the traditional data are not doing a good job of it. And I think three years in the answers, Yes, Diana, I lived in the Washington area for ten years, and there's no shortage of think tanks working modeling the US economy. It sounds like your approach is for one of a

better term disrupting big think tank. Well, that's an interesting question, and I would say that it is disrupting certain areas of economic research. I would venture to say that to have a full understanding of the economy, we're gonna need multiple lenses, and no one lens is going to give that to us. So even though I think in some ways we are informing important economic questions better than say

the government statistics are. Take for example, the work that we've done on the so called gig economy, the online platform economy. I think we have better data as as even these statisticians at BLS at the Bureau of Labor Statistics would would acknowledge. But where it's not a substitute for that, it's a compliment to many of the other data series that have a long shelf life that been around for a while. And I would say the same thing about the think tanks, that we are disrupting the

way economic research is being done. I think we're doing that hand in hand with academia. But there's plenty of room for multiple windows and we get a much richer um tapestry of what is in fact a very rich economy. With those various lenses. Well, well, that was actually one of the areas I wanted to ask you about Theanna the research on the online platform economy. It really is a buzzy issue that you've been working on taking a look at how people are using and generating income through

platforms like Uber selling things leasing. Can you tell us a little bit about what you discovered in this research how profitable it is for Americans to work this way?

I would love to um, As you say, it is an area that we've done a lot of work in and um what I would say about it that is important is that we got into a view of what was happening in the online platform economy by understanding that if you observe households on a high frequency basis, you know, monthly, weekly, what we observe is that most hustles are facing high levels of income and spending volatility. That's not something you

would capture with annual surveys, for example. And the conventional wisdom of three years ago or so, as this economy was getting started these gig jobs and otherwise is that's why we have so much income volatility because people are doing multiple gigs and and that explains it all, and we thought, well, what kind of data exists on that, And you know, there's the Contingent Worker Survey, but that's

only done every three years. There's the pieces of this, some of those were discontinued as a matter of fact, and we said, we can answer that question because we can go and understand who is participating on these platforms by observing who's receiving income from them and then linking those to their overall economic and financial kind of outcomes.

And so what we learned is that counter to that conventional whiz them, this is not really the future of work as as cast glibly, because in fact, even now that we've updated these numbers, at you know, at best, it's one point six percent of the population is participating in these platforms, and we can document that with very large samples, and most of them are not working on more than one platform, although that is increasingly a phenomenon.

The only link we had therefore to the income volatilities that we noticed that for some of the participants, say those that were participating in what we call the labor platforms, think of the transportation services in their full kind, and uh, there are now many many of those, but also all the other services that walka dogs and the shop for people and all that. There's a labeled platforms as distinct from capital platforms like the airbnb s or other places

you can rent an asset or buy a good. That for those participating in labor platforms, this was a way of offsetting the income ball is tility from their traditional job. But most of them actually had traditional jobs, so we could observe that for most people this is supplemental income

to the tune of um, not the whole reality. In any given month where they're participating, say driving or or renting an Airbnb room or otherwise, it can be significant for that month, but over the course of the year it's really just supplemental income, often designed to with offset in effect, the volatility from the traditional source of income. Diana. The official unemployment rate in the US is three point seven percent, and it's been below four for some months now,

and yet it isn't generating a surge in inflation. C p I is more or less around the Federal reserves target of two percent. What do those figures miss? Well? I think there are many aspects of the economy that are new and interact with inflation in wages in ways

that we are only beginning to understand. So if we come back to these online platform economy businesses, particularly those that are involving assets, whether it's driving a car or renting a house or an apartment or a room, one of the effects of that, if you think about it, is that you are delivering a service or a good without requiring more capital in the system. So before the car would sit in a parking lot, now it's being used at a much higher rate by those who are

driving folks around or delivering things for them. So to that house that used to sit empty is now being used without having to build a new house, and so that in effect it has a deflationary impact on the economy that we're not entirely measuring it because the way we think about GDP growth or even unemployment UM is through the category worries of employment and GDP that we used to think we're the norm. Uh, new forms are

being introduced that are sort of challenging that. So I think that that's one of the reasons we're not seeing UM.

The pressure on inflation or pressure on wages is that assets are being utilized better and that keeps costs down in other ways, and your data tells us about that car and the garage and that house that's being used very clevely, very clearly, and we really are note seeing that it's it's assets sort of being utilized that would otherwise be laying fallow, and so you're not creating new production to make those available and therefore not driving the

inflation that would that might go with that. That's really interesting, Diana. Let's talk a little bit more about the ways that you measure cash flow in Americans bank accounts and their

day to day spending. One of the most interesting things that I thought you put out was looking at out of pockets spending on healthcare and and the financial burden, uh it places on Americans where we have this kind of crazy health care system in this country that people have to kind of really manage their expenses and be careful about. What have you found about about Americans healthcare spending and in something you know, really interesting things that

stand out about that. Um. Yes, So let me um zoom out a little bit to understand the out of pocket healthcare send because it's important. I mentioned earlier that we observe through this high frequency lens at the individual household level, very high levels of income and spending volatility and um when you double click on what is driving that volatility on the spending side, not surprisingly, one of the big buckets is healthcare. So the three are healthcare,

auto repairs, and tax payments. Those three events will be an extraordinary expense for nearly one out of four Americans every year. But healthcare is important because even for those who have insurance, if they have a high deductible plan, which most of us have, UH, that will require a cash outflow. Now, some of that cash outflow may eventually be reimbursed, and by insurance, some of it will not be reimbursed, but it creates a real cash flow event um.

And we know from our data, but also other data the Federal Reserve and others have put out that most Americans don't have a financial buffer to sort of withstand that extraordinary expense or withstand a drop and income of some sort, and so the result is large levels of out of pocket healthcare spend that can either translate into deferred care, people not seeking the care they need until they have that cash flow, say a tax refund or something to do it, or just a real impact on

their financial outcomes, meaning they go into debt through credit card or a need to harness other resources. UM. And we do find that to the tune of over a thousand dollars. You know, families are going to face that kind of out of pocket hit, which is extraordinary given that they don't have the financial buffer to withstand that. Deanna, you've also sat on the other side of the table. You are a policymaker in an Obama administration, governments and

central banks inching closer to your model. What are you hearing from them? Well, let me I'm glad you mentioned that, because one of the inspirations for the institute was this recognition that at times of crisis in particular, but maybe at other times too, policymakers aren't always equipped with the real time, granular, high frequency information that they need to

make the best possible decisions. And so part of what I think that the bank excited certainly got me excited, was imagine if we could begin to bring that into the decision making that policymakers have, that that other people who are making decisions have, and and so that is

a bit of the inspiration behind this UM. I think many are moving close to that direction, and one as an example, one of the data series that we have now put up for two and a half years is what we call local consumer Commerce, and this is a view of sales if you like, that transpire in a given city as a measure of the vibrancy of that city. The inspiration behind that was a set of conversations we'd had with the statistical agencies that said, we don't have

good enough data at the city level. If we take our national data and then bring it down to the city level, it would be very helpful to have a corroborating view of what's happening at the city level that will improve what they do. So I think that's a

good example. Another example that that we are very keen on is you recall and we may be entering a period like this, this very steep drop in gas prices that transpired from it's a very dramatic drop in gas prices and very vaccine question for traditional surveys and others is what are people doing with the savings at the gas them And if you went out to survey people, which many people did, they pretty much categorically said, well, we're saving it. Of course we are because that's what

we all intend to do. And and it's very hard to know what you're really doing with what amounts to five percent of total spend over the course of the year, even though for some people it's a lot and for

other people's small amount. And our data lent itself very very powerfully to saying, well, at the household level, what decreases do we see in gas spending and what corresponding meaning causal that we can really map to that dropping gas spending can we link to different kinds of spending and we find out in fact people were spending most of that they were spending it on groceries, on restaurants.

Now that matters a lot to central banks. At the time, the FED was and you know, it's constant deliberations as to whether it should start raising rates or not. And it was important to that conversation to know whether that gas savings was in effect still a buffer that could bring the economy forward or whether it was already built into the numbers as a way of understanding whether it was time to move or not. And um and I think this kind of microwork, of course, with many other

things complementing it. To my earlier point was important to consider that maybe they should wait at least one more quarter before moving forward, since in fact that was already built into the GDP numbers and we couldn't reasonably expect that it would be a wind in the sales of

the economy in the future. So, with the gasoline prices again declining in recent weeks, as the price of oils dropping, fuel prices at the pump are probably going to go down even further, has that earlier work inform the approach that policymakers should be taking now, given that they're already uh in this interest rate well into this interest rate

hiking cycle. Well, it's a good question. And frankly, after we concluded that work, we did it once as a snapshot peak to trough, and then we did it again over the course of the whole year. And you know, people do habituate. You know, once they're used to a thing, they're more likely to keep doing it. And so we thought, okay, we think we have an understanding of how people change their behavior when gas prices go down significantly. Um, we're going to wait and see what happens when prices go up.

Do you have a symmetrical effect the other way, Do people spend less because they now have to spend more on gas, you know, other things go down or do they not? And so we were eagerly awaiting. But now what we have is the probability of a pretty significant continued decreases at least as as summer predicting it and um, so the question is do we have the same level impact in the increase in spending other things or are

we at a different point in that curve? And you know, my guesses will turn back to that question when we have a few months of data to answer it. That that's an interesting thing. I think most of us know that that's not a linear curve, meaning that a ten percent increase in gas prices doesn't at at gas price two dollars is not the same as a ten percent increase at gas price four dollars um. And so we're

interested to see what that curve looks like. The microwork we're doing, I think over time will help inform that both on the way up and on the way down as gas prices go up or down. Thank you for mentioning cities. The term urban rural divide is a popular one and it relates mainly to politics. Are you seeing a divide in your bottoms up economic data? I would say that getting a much better understanding of what's happening at the city level is important, even before we start

informing the city to rural divide. I would argue, there's actually quite a variation in the experience of even what we would consider cities, and then that variation gets even wider when you include rural areas. The work that we've been doing on cities has been primarily focused on at this point fourteen larger cities than not. We have some that are larger, some of that are smaller, and that's

the what we are mostly getting a window on. Partially, it's it's that we have enough representation, enough sort of observation, so to speak, to feel very confident that we're saying something important about those cities. And partly, as you can imagine, is that the bank's footprint is much stronger in cities than it is in the rural area. So I suspect that just as we're seeing significant variation across the city samples that we have, we would just increase that variation

significantly if we had rural observations. But we have so far not actually done a rural urban divide. To inform that question, well, Diana, your former White House colleague Larry Summers has popularized the term from the nineteen thirties Alvin Hansen's secular stagnation. What does your data tell us about

that term? Well, to try to mainstream the concept a little bit for those who may not be as familiar with it, I think the best way to phrase that kind of line of thinking is that we should be expecting a new normal, so to speak, that we're not likely to to see the kind of growth rates that we experienced it at peak levels. And that's not the

only aspect of that theory. But but I think as we look into that one in particular, UM, there are some things that are incontrovertible that would correspond to that, which are, for example, that we are aging society, and therefore, as there's some growth that comes UM strictly from population growth that we have less of, and if we go down the path of UM limiting immigration more, that's one other source of demographic sort of weight down. And so in that sense, I think most people would say that

stands to reason. I think in the other sense, we have seen even as you know, imperfectly measured as it is, GDP growth um pick up quite significantly in the last while, which suggests that there are some things that can be done to move in that direction, certainly some of the fiscal stimulus that was put in place or otherwise, although what's yet to be seen is how long lived any

of that is. Our data are not the best to inform that question, because really that is a macro economic question and the strength of our data are that we're taking a micro view and then taking it up to a macro view. So I would argue that that is probably not the best question we can inform. All Right, last question, Dianna, what can we expect from JP Morgan Institute in the rest in terms of big insights projects

or anything else exciting that you're planning. Well, thank you for asking, because this is my opportunity to say please follow us. And all our research is out in the open, available to all at our website. So um things that are coming up are will continue on some of these main themes, but with new twists. So I mentioned the

financial economic well being of household. We want to keep understanding this um notion of what makes households resilient, and you know, besides having a saving buffer, what do we know about the behavior of those households that thrive versus those that not That might be good lessons for folks

in the future. We are have been doing, but will continue to do, significant work on mortgages to understand how well does kind of that that portion of household debt interact with other financial outcomes, and that becomes important as we think about very significant changes in mortgage rate deductions and otherwise interest rate deductions on mortgages. We're starting and launching a new segment on student loans, which we think

is very important. Many of you will have seen how much of an increase in student loan debt burdens there has been in the last while, and we're going to try to understand what does that mean in terms of consumption patterns, other decisions that households with that debt are taking on or not. Will continue our work on healthcare and very proud of the work we just put out, but we'll continue to update that shows the out of pocket healthcare spending. We talked about um by county, by

demographic group. What are the levels of spend, what are the burdens as a share of income, and how can and how does that interface with the overall economic picture in the city. Escape Since you mentioned it, the first foray we had was on let's understand what's happening with purchases at local merchants. We've now done a companion view that will come out soon on what about local residents?

And the reason that's kind of an interesting view is that we know increasingly people are spending not just at local merchants, but online and at places other than their own UM city, and we really want to start mapping that out. There are some data series that commerce and others do on that, but what we're learning is that as the economy evolves in that way, we're not capturing that as well as we think we can because we really know where exactly that purchase took place and by whom.

So we'll do a lot of work on the online economy and and the health of residents in key cities. The small business area is one that we will continue to further understand, and we really want to bring um a demographic lens into that so that we understand better the performance of women and minority small businesses. Will have some of that view into households as well, which we're

excited about. And then I mentioned very briefly, we haven't talked about it, but for those of you who might be interested on the financial market side, we've done some interesting work to understand institutional investor behavior, So what actually happens minute by minute, hour by hour when big events like Brexit, the US election, uh, the Swiss Bank moving the floor on the Swiss franc Um And we're mapping those kinds of events out and we plan to do

much more of that next year. All right, well, it sounds like you have plenty to keep you busy for quite a while. Diana Farrell, President of the JP Morgan Chase Institute, Thank you so much for taking the time with us on Benchmark, and thank you for having me. Thanks for listening to Benchmark. You can find all our past episodes on the Bloomberg terminal, bloom work dot com, our Bloomberg app, as well as podcast destinations such as

Apple Podcasts, Spotify, or wherever you listen. We'd love it if you took the time to rate and review the show so more listeners can find us. And you can find us on Twitter, follow me at scott Landman Dan, you're at Moss Underscore Eco and our guest is at Barrel Underscore. D I A n N Benchmark is produced by Topor Foreheads. Francesca Levy is the head of Bloomberg Podcasts. Thank you for listening for the past three years. We wish everyone a happy holiday season and New Year.