Living Paycheck To Paycheck: Empirical Analysis And Macroeconomic Implications

>> John Taylor: So let's begin, we're very pleased to have Andrew Levin come here from Dartmouth to speak about living paycheck to paycheck, what a name. Empirical analysis and macroeconomic implications. I remember not too long ago, Andy and I wrote a paper together. It was published in a book by Orphanides and Bordo.

It was on stop, start, whatever it is, yeah, that's right. But thank you very much Andrew is a professor at Dartmouth. He got his PhD here at Stanford and worked at the Fed for how many years?

>> Andrew Levine: About 20 years.

>> John Taylor: 20 years, so he knows everything there is to know about the Fed and has mixed it up with other things as well.

So welcome Andy and you're welcome to come down to the big table if you like, bye bye.

>> Andrew Levine: Okay, well, great. Thank you so much, John, for inviting me to be here, not just today, but this month. So this is completely new work. The paper should be ready, actually, by the end of June.

But I'm eager to share this work with you today to get your feedback and advice, suggestions. We think it's an important piece of work. So Dwight with Arunima Sinha, who's at Fordham, and Carl Walsh, who's at UC Santa Cruz. Carl sends his regrets. He's actually with his granddaughters on a tour of Italy this week, and I was totally supportive of him doing that in lieu of.

Originally, when we scheduled this date, we thought he might be able to be here in person. But anyway, it's been great working on this project with Arunima and Carl. And obviously, even though I'm used to having the red ink at the bottom of the COVID page, not to interpret these views as representing other than the three of us.

And if I say anything silly today, then interpret that as solely my responsibility and not Arunima or Carl. So the first question you might be wondering about is this phrase, paycheck-to-paycheck, I did a search on EconLit. There's exactly one paper published in a referee journal that has used this phrase in the last 20 years, and it was about 20 years ago in a finance journal where they were looking at the consumption as a pricing model and made reference to this.

But by contrast, if you talk to ordinary people and you look at media articles, it comes up all the time. And so here's to kind of a little bit of background here, BankRate.com. Living paycheck-to-paycheck is the common term for those who don't have enough money to pay for future expenses until their next paycheck arrives.

And by the way, it's paycheck-to-paycheck. But I think the phrase would also be used for someone who's retired, living on Social Security, where it's not really a paycheck, but it's a check. And they basically just arrange their expenses so that their money is essentially spent at the end of the month when the next check arrives.

57% of Americans don't have enough savings to cover $1,000 emergency expense, okay.

>> Bob: That number simply cannot be believed, it's like one of those numbers that one third of Americans go to bed hungry.

>> Andrew Levine: Okay, all right, great. So Bob, this is exactly one of the reasons why we started the project, it's like, okay, these are opinion polls, and so maybe we should just take them with a grain of salt.

But before we get into this question of what are the facts? Okay, just to finish the what's the common usage? And an important point that's made in Investopedia is that this idea of living to paycheck-to-paycheck is not just for poor families, for low income families in poverty, but it can occur at all different income levels.

And they say many Americans live paycheck-to-paycheck because the cost of living has not increased in proportion to salaries. Or in other words, this project we're working on right now is actually connected to the earlier work that John and I did ten years ago about the horrific costs of inflation and the extent to which many ordinary families, their labor income doesn't keep pace with inflation, their real income drops.

That's what this Investopedia quote is talking about. So here's a just kind of scrapbook, I guess you call it electronic scrapbook of kind of recent articles. So, CNBC, your money, with inflation stubbornly high, 58% of Americans, there's that number, again, are living paycheck-to-paycheck. Okay, that was published April 11th, 2023.

Now, they obviously didn't talk to Bob, I guess, in that story, maybe that's why they got it wrong. This is not a new idea. If you look at the bottom of the screen, there's a quote from NPR, in fact, a whole story they did in December 2020, Paycheck-to-Paycheck Nation.

Why even Americans with higher incomes are struggling with paying their bills. And you can do a Google search, and you'll find this phrase coming up and surveys and all sorts of things probably over the last couple decades, at least,. Again, across age groups and income groups in many different media journals.

 

>> Bob: So, as I say, it's a kind of common phrase. Are you gonna talk about Greg Kaplan's academic work?

>> Andrew Levine: Yes, we are, of course.

>> Bob: I'm speaking nervous today.

>> Andrew Levine: The funny thing is, Bob, I remember last year when I was giving a paper at the very beginning, you said, well, this can't possibly be right.

And at some point we kind of reached some common views. So I'm hoping that will happen again today. Okay, so now coming to macroeconomics, there's been a long standing consensus that structural macro models should have good micro foundations, especially when they're used for analyzing alternative policies. Okay, that's the Lucas critique.

The first-generation of new Keynesian models that were really developed in the 90s, and 2000s were formulated using a representative agent. I assumption it was convenient analytically very convenient. In the simplest case, you get a single new Keynesian`s equation that's essentially a log linearized version of the Euler equation.

And you get Keynesian Phillips curve, and you have a policy rule like the Taylor rule, and you can analyze the economy in that sort of simple setting. These are now often called RANK models, representative agent, new Keynesian models. So I'll use that phrase few times today. Very useful for analytical purposes, but they obviously abstract from some key dimensions of reality.

In fact, if you talk to anyone who's not a macroeconomist and you say, would it be reasonable for us to kind of assume that everyone in the United States, that all families are essentially the same. They want to shake their heads like, you must be an economist. Okay, so the good news is that over the last 20 years, and especially over the last ten, there's been a burgeoning literature of formulating macro models with heterogeneous agents.

So starting to recognize that not everyone's alike, that there's some intrinsic differences. The seminal work was Bilbiie (2004) in his PhD dissertation, and Gali, Lopez-Salido and Valles. It's worth mentioning that both of those papers, the dissertation was published in JT and the Gali, Lopez-Salido, Valles in the journal of the European Economic Association.

They were not top journals. And before the Global Financial Crisis, before the Great Recession, this work was still kind of pretty far into the dark corners of macro, not really front and center. But over the past decade, it started to get published in some very visible places. And another one of the seminal works is by Kaplan, Violante and Wiedner in the Brookings Papers in 2014.

So a key contribution of the Kaplan and all paper was that, sorry, Gali, Lopez-Salido Valles coined this phrase of hand-to-mouth households. And the idea was kind of like in the old-style Keynesian world, consumers who just spend whatever they earn, they called it hand-to-mouth. And Kaplan, Violante and Wiedner said hand-to-mouth doesn't necessarily have to just be poor people.

It could be someone who has a higher income, who even owns a house, and maybe some other illiquid assets. But still acts like a hand-to-mouth household because they don't have any liquid assets that can be used as a shock absorber for fluctuations in their current income. So that was very important, I think help people recognize that this isn't just a small group of poor people, that these kind of issues might be much more relevant.

Nonetheless, as Arunima and Carl and I have looked through this and thought about it, what we noticed is that most of the literature on what are called HANK models. So heterogeneous agent and New Keynesian models, has focused on heterogeneity and assets, and the corresponding implications for consumer spending.

But those models have continued to maintain an assumption of homogeneity in labor markets. And again, if you talk to people in the world, I'll say the real world, that would also be a pretty objectionable assumption. But there's differences, some people work, some people don't, some people retired. Some people are stay-at-home parents or staying home taking care of an elder.

Some people are just retired or disabled for all kinds of reasons, okay? Another thing to say is that in those models, wages matter for inflation, the aggregate supply block of the model. But wages don't matter for aggregate demand, so going back to this phrase here.

>> Bob: What is aggregate demand mean?

 

>> Andrew Levine: Well, in a New Keynesian model, it's the highest equation, okay? That's aggregate demand, so the relationship between the path of real interest rates and the level of spending. So this kind of thing that CNBC says with inflation stubbornly high doesn't come into these, not just the RANK models, it doesn't come into the current generation of HANK models.

And that was a kind of key motivation why Carl and Arunima started working on this project is trying to think about, well, would it be the case that if inflation picks up and nominal wages are sticky? And by the way, this builds on John Taylor's work, 1970/1980 on staggered nominal wage contracts, and the idea that sticky nominal wages, important part of how the economy works.

And of course, I did work later with Chris Ersig and Dale Henderson. That model has become commonly used in the new Keynesian literature with sticky nominal wages. And so in a world with sticky nominal wages, the idea that inflation picks up people's real purchasing power drops. And that might matter for their real spending behavior, and seems like something important that we wanna be able to consider.

And we haven't been able to do that yet because it's not in the models. Okay, so what do we do? So there's really two parts of this, it's in the title. The first part is Empirical Analysis, we're gonna look very closely, very closely at the PSID. And by the way, I'm old enough to remember when the PSID was all on magnetic tapes.

It was just a heroic effort to do anything with it. The great thing now is it's all available electronically through University of Michigan in a really user-friendly way. And you can download the whole thing, of course, now the amount of space the whole PSID takes is trivial. Probably Johnny could fit on your iPhone or something back.

We do a detailed analysis of PSID to try to look at heterogeneity in sources of income and employment, as well as asset holdings. So, this is one way in which we help to build on the existing literature, is to think about these other types of heterogeneity. And then the second part of the paper, is to use the results that we're gonna find to build a new, next-generation HANK model.

And our goal, if you know Carl, you wouldn't be surprised for me to say this. It was a mandatory imperative for our model to be understandable. Okay, Carl does not like black box models. Okay, I don't think he's ever written a black box model. He likes ones where you can kinda look at it and understand the different channels there.

And so that meant that we weren't trying to get a HANK model that had every element of reality at every possible type of heterogeneity. We were hoping to be able to come up with a Hank model that was more realistic and some key dimensions. And that would capture some issues that real people seem to care about that have been missing from our macro models.

Okay, so again, two parts Empirical Analysis, macro modeling, trying to bring them together. Okay, so the first part of the Empirical Analysis, let me just briefly summarize what we do here, and then I'll get into details in a minute. For today, we're focused on the 2017 and 2019 waves of the PSID, it's a series done very carefully every two years.

They've done the 2021 wave, but it's not yet publicly available. I'm hoping, was absolutely with the optimal pandemic, but that's where it would be interesting to sort of see how these issues come up. And in another couple of years, we'll get the 2023 wave even write this, it's an ongoing.

PSIDs just had their 50th anniversary recently, ongoing longitudinal survey.

>> Bob: Become a COVID specialist.

>> Andrew Levine: I hear you, Bob, and so I understand. So it will be interesting to look at, but as you say, it will have some difficulties as well. Yeah, go ahead.

>> John Cochrane: A difficulty called a wonderful opportunity because there's all these hand-to-mouth people and then just, saying, people haven't built generational wealth or whatever.

They got given a $20,000 check I said, great, they're gonna save it. And that's the end of the hand amount. No, they quit their jobs, spent it all down, and it was all gone in about six months, just amazing-

>> Bob: Quit their jobs.

>> John Taylor: Experiment, yeah.

>> Bob: No, that's not-.

 

>> Andrew Levine: Well, anyway, okay, given Bob's caution, why don't we continue this? Cuz what Marie Christine said is we're gonna try to save 15 or 20 minutes at the end for Q&A. So this is a great question, I think, how we should think about the COVID pandemic and its aftermath.

I will say there's other sources, we looked at some of them, the consumer expenditure survey. There's the survey of consumer finance that's done by the Federal Reserve and the CPS, which I love cuz you can access it with IPUMS. And so at some point we're trying to build up a whole set of knowledge of these micro facts that are relevant for the macro modeling.

And the PSID was just a useful starting point because it looks at families over time and lots of details about their assets, their income, their employment, their spending. Okay, it's a really great resource. We identify several distinct types of households, the first type is what we're gonna call fixed-income households.

These are households where at least 90% of their total income comes from things that are essentially fixed on a month-to-month basis. So you could think of this as kinda like the paycheck-to-paycheck, either transfer payments, which could be social security or disability insurance payments from Social Security or welfare payments or alimony and child support.

Okay, those are all classified as transfer payments or interest from longer-term bonds or a trust. Okay, that's again, fixed and predictable. Okay, so those are the fixed-income families, how many are there? Okay, what do they look like? How old are they? How much income do they have? All those things, I think, we think are actually important and I don't know.

Please help me fill up our bibliography. Okay, if you know of papers that have looked carefully at fixed-income families and how they matter for the macro economy, we haven't seen it much, even in the new generation of HANK models.

>> John Cochrane: Who are not working out of labor force, really?

 

>> Andrew Levine: Okay, great question. And let's flag that question cuz I'm gonna come back and look at this with our microscope in just a couple minutes. It's okay, but who are these people? Okay, we can tell from PSID that 90% of their income is transfer payments and interest from these longer-term investments, okay?

Who are they? Okay, second group, what we're gonna call liquid families. Okay, not fixed-income families. We're now talking about the rest, okay, non-fixed-income families. Liquid means they have more than one month of liquid assets. And illiquid means they have less than one month of liquid assets net of their credit card debt.

Okay, pretty simple. We think this actually corresponds reasonably well to this idea of paycheck to paycheck. Especially when you think about the idea that if you pay rent or mortgage, you get paid, say, on the seventh of a month, and your mortgage payments due on the 27th of the month.

Okay, there's gonna be a balance in your checking account or your transaction account for several weeks cuz you gotta make sure the money's there when the mortgage payment gets paid. Likewise with your health insurance, likewise with your electric utility bills. Okay, so ordinary families are used to all this, they gotta manage it.

They don't want the electricity to cut off. Okay, then they gotta manage a small amount of balances. Okay, but they still consider themselves paycheck to paycheck because depending on the precise timing of the mortgage payment compared to your paycheck.

>> Bob: Liquid, if fixed income overlap, right?

>> Andrew Levine: No, they would overlap.

And actually, we have information about the fixed-income families that are liquid versus the ones that aren't. But what I'm trying to just do here, and maybe the slide needs to be more clear. I tried to think about it, is when we say liquid and illiquid, we've already taken out the families that are on fixed incomes, okay?

So again, we can retrace this in a minute. Here, I'll show you.

>> Bob: So liquid should be qualified-

>> Andrew Levine: Non-fixed income who have more than one month of liquid assets. And illiquid is non-fixed income who have less than. I think it's actually on a slide, a few slides from now.

But it should have probably been on this one. Okay, and then we're gonna further disaggregate the illiquid, non-fixed income, illiquid households into based on how they get paid. And this is a source in the PSID, I think it hasn't been really utilized, but it can be and should be.

PSID asks people, how do you get paid? Do you get paid by the hour? Hourly-wage plus tips. Okay, do you get a salary or a salary plus commissions, okay? And so we think it's interesting to actually think about this form of heterogeneity, too, does it matter? And it might matter for the family, but it might not matter for the macro economy.

That's the sort of question that we'd like to be able to look at. Okay, so what do we find? And this, Bob, is front and center, okay? You said we're wrong you said those surveys are wrong. I think the PSID comes exactly the same conclusion. Okay, so what we find in the PSID is about 60%.

And it's plus or minus cuz they're standardized, all these things, it's a survey, okay? But about 60% of households are reasonably characterized as living paycheck to paycheck. Okay, how do you get to 60%? First, about 20%, it's really 18 or 19%. So nearly 20% of adults live on fixed incomes.

Remember, fixed income here means 90% of their total income and PSID is pretty careful about this. They don't just ask the household, they get the tax forms and the bank statements. It's okay to verify the information that they're getting from the survey, okay, so give PSID some credit on this.

90% of their total income is fixed income and 20% of us adults live in families where 90% of the family income is fixed. Okay, now you can break it down. Two-thirds of those are older adults and the fixed income is essentially social security and pension income. And it shouldn't be surprising to you because there are a lot of American families that are essentially retired and where they're essentially living on Social Security and a pension.

 

>> Bob: Well, they have assets, too, you're leaving the assets.

>> Andrew Levine: No, those are not in this 20%, there are other people who are retired, you're absolutely right. Actually, of the families-

>> John Taylor: You didn't say that, you didn't exclude holdings of assets. You talked about interest even.

>> Andrew Levine: Right, so that's true.

 

>> John Taylor: Go back to your breakdown here is now more interesting.

>> Andrew Levine: Okay, so it's transfer payments and interest from bonds and trusts. So you're right that if they. And annuities, I should have said here again, this slide isn't perfect. Annuities, bonds, and trusts. And we can quibble, by the way, we have all the data in the PSID afterwards.

This afternoon or when you get back from Puerto Rico, I'd be happy to sit down with you. And we can say, well, how does it change if you take out the annuities and trust? I thought you said Puerto Rico.

>> Andrew Levine: All right, this is a little bit longer trip, but okay, Portugal sounds good still.

All right, so, we're gonna get to the details in just a minute here, this was supposed to be an overview, okay? So let me just say one more thing, okay, which is that not all older adults live on fixed incomes, okay? About half of older adult families, a lot of their income is from owning a business or a farm, or some other kind of assets, okay?

If they have a lot of shares of stock and they're getting dividends from it, okay? That's not counted as fixed income. And what we're really trying to do here is to separate out the families that are, what we would think of as kind of acyclical, okay? Their income and their spending patterns are gonna be acyclical.

It's not that they're poor, I'll show you that in a second, okay? Not at all that they're poor. Okay, but the other third, and so now we're talking about 6%, roughly 6% of adults live in families that have a fixed income because mostly because they're permanently disabled, they're getting SSDI.

And if you know this, in order to get SSDI, you have to get letters from two different doctors saying you're permanently disabled, you're never able to work again, okay? And it's unfortunate, I think it's a controversial question David Autor and others have worked on. There's been an increasing number of people who are permanently disabled living on SSDI.

And maybe those rules should be changed or maybe incentives should be made to help people get back into the workforce. Rather than incentives for people to become registered as permanently disabled and get SSDI. But a substantial fraction of these younger families, close to half of their income is disability insurance, okay?

But many of them are also getting welfare and other transfers, as I said, alimony, child support, aid for needy families, okay? And some of these might be also ones that could come back into the workforce. So, go ahead.

>> John Taylor: I think I understand now, now you're making sense.

 

>> Andrew Levine: Okay.

>> Bob: There's a V literature which deals with the question of suppose that there's a, say a medical expense shock, then does the household have the flexibility to deal with that from all the different things they can do? There, the emphasis is very much on just what their liquid assets are, because you could have a completely fixed income, but you could deal with this need for extra spending just by drawing down your assets.

But that's not the question you're asking.

>> Andrew Levine: Not precisely.

>> Bob: You seem to be asking your question more like, how exposed are they to-.

>> Andrew Levine: Just like all income shocks. Yeah, that's right.

>> Bob: Great, great.

>> Andrew Levine: Yeah, that's right.

>> Bob: Of course, you're looking at a time when-

 

>> Andrew Levine: The great thing about PSID, Bob, is that it's such a rich resource for these kinds of questions, we should be asking more. For example, okay, there are a lot of these older adults who have no wealth, nothing, they don't have anything, okay? And so, fortunately, with Medicare and other things, they can probably manage.

But some of these younger adults are permanently disabled, they have nothing. There are others, I mean, this is where I really come to this second group here, the bigger group, which is 40% of adults who live in households where labor is practically their only source of income, okay?

And we're calling those paycheck to paycheck because they have practically no liquid assets. Many of them actually have a home, and they have substantial home equity. And one of the things Arunima and Carl I've been debating with just in the last week or two, is how much does it help you if you have a lot of home equity.

And you had an emergency, like a medical emergency, could you get a home equity line of credit in an emergency, or if you lost your job suddenly, okay? Could you get a home equity line of credit? And the answer is, it's pretty tricky, because one way to get a home equity line of credit is with income verification.

That's all kind of been tightened up since the financial crisis for reasons you can imagine, okay? And so if you just lost your job or your spouse lost their job, okay, you may not be able to get that type of home equity like credit. There's another type which is called stated income, and many financial institutions won't do it, okay?

Again, they're discouraged after the financial crisis from doing this cuz you probably remember from the big short, right, the stated incoming. Just write down a number and no one checks it, okay? And the problem with now with a stated income HELOC, is that your house becomes collateral for it.

And that means if you miss a payment, they could start foreclosure proceedings, and many households don't wanna do that. So the bottom line here is, I actually think that even though these families have substantial assets, in some cases their home equity is a multiple of their family income.

So they have a lot of savings in their house, but it's not liquid. And in the circumstances we're trying to imagine here, where there's a downturn in the business cycle, or a pickup in inflation that erodes real wages. Getting the home equity line of credit's probably not something they're gonna want to do because they just don't wanna put the house at risk.

And they don't wanna go through the income verification process. So again, these are important questions, we're just scratching the surface here, trying to make some progress in really thinking hard about, okay.

>> John Taylor: So can I just question, am I supposed to add the 40 and 20 to get to your 60?

 

>> Andrew Levine: Yes, absolutely. So that's what Bob asked before, and I just apologize, it wasn't crystal clear. Okay, we take the fixed income families, remember, those are 90%, okay. Transfer income and some interest income, okay? And we're putting those aside and then we're looking at everyone else and we're dividing those up into liquid versus illiquid.

And so the 40% is about half of all the rest of the population, okay? That's not already in fixed income. And these half are people for whom labor is really the main source of income, 90% of their income, I'll show you in a minute. Okay, so just quick overview of the model, cuz I can see we're gonna run out of time.

The model is gonna have four types of households, so there's something called a tank model. Some of you have heard that phrase, tank means two agent, new Keynesian model. And recently someone from the ECB has written a paper with a frank model, which means three agent. Okay, so we're thinking maybe ours eventually will be called a frank model.

It's just too bad that none of the co authors is named Frank. But anyway, so it's a frank model, okay? The fixed income households, wage workers, salaried workers, and what we're calling executives, okay? And the point here about the executives is that practically all the businesses and all the shares of stock are held by liquid households.

It's not just that they have liquid assets, they own essentially all of the businesses in the economy in the PSID, okay. And so then we can get some pretty cool results from this Frank model, okay? It has first order effects on the macro dynamics, real wages enter directly into the new Keynesian equation.

So now we can start to think about that CNBC thing. Inflation, dampening people's purchasing power actually would affect aggregate demand in our model. Please John.

>> John Cochrane: Just the source of income, building up a buffer stock of assets, are they very impatient?

>> Andrew Levine: Okay, that's a great question and I hope we can get to directions for future research here.

Some of these TANK models and HANK models kind of assume that a household is a particular type. It's just fixed forever, okay? And that's kind of how we're building the model here, okay. So there's just a fixed income household, the truth is-.

>> John Cochrane: Fixed some household which consume a little of, you have a lot of assets.

 

>> Andrew Levine: I agree with that and I think that. So one thing we can look at in the data and the PSID, and we have started to do some of this is, what's the flow over time? And we've just done it right now for 2017 to 2019, but what's the flow of people who are doing what you're suggesting, John?

By the way, there's lots of financial advisors who are saying this to people to save 2%.

>> John Cochrane: I know lots of people who live like this but it has a modeling question, how did you write down your model?

>> Andrew Levine: The model is just fixed types, okay?

>> John Cochrane: Fixed type to solve an optimization problem and something in that optimization problem stops them from accumulating well.

 

>> Andrew Levine: Okay, so as I say, this is like beyond the scope of the current paper, okay? In the same way that it's beyond the scope of the current paper, why are they in the real world? Why do people ever retire? Why don't they keep working all the way through the end of their life.

And the answer is, there's things that aren't in the model, like people's health, disabilities, age and other kind of factors.

>> John Cochrane: Much simpler question, when you write a model, if say, agents of type a, max beta to the tu of ct, subject blah, blah.

>> Andrew Levine: Right.

>> John Cochrane: And if you do that with your income types, those people are gonna start building up buffer stock.

So you must have done something to stop them from doing that in the model you wrote down.

>> Bob: Give them a high discount rate.

>> Andrew Levine: You can do that, I wanna come back to this question, because it's gonna come up at the end, okay? But this probably, you could think of this as more similar to.

What's that? Yeah, you say, it's in the model. These types of households can't borrow and they can't save.

>> John Cochrane: You just wrote down consumption equals income.

>> Andrew Levine: Essentially, yes.

>> John Cochrane: You didn't solve the max.

>> Andrew Levine: Not for that, right?

>> Bob: There is a literature piece to area in particular who ask how much implicit insurance do people have based on how their consumption actually does change?

And there you get the impression that there's a substantial amount of implicit insurance, that he does most of his research with SSID. Have you looked at that?

>> Andrew Levine: There's a number of papers that have asked this question. Kind of Can we explain why some households have lots of savings and others have practically none?

In fact, the Kaplan, Violante, it's not the Wiedner paper, it's the other one, I think Ben mole. The AER paper actually looks at kind of a model where you can explain why there's some households that are they call them wealthy hand to mouth households. So, those are interesting questions, but we don't do it here.

Okay, so what are the things that are interesting about our modeling results? Just briefly, so real wages enter into the NK IS equation. Aggregate demand is affected symmetrically by the remitted wages of the hourly and salaried workers, maybe we can get into this a little bit later. But even though their labor contracts are different, the salary workers, the employment, the number of hours they work is determined by what we're thinking of as an efficient allocative wage.

But the remitted wage is sticky and for a salaried worker who's got liquid assets, that doesn't really matter. If their salary over time catches up to where it's supposed to be, it can fall behind or catch up or go a little bit above. They can use their liquid assets, it's a shock absorber.

But where this matters in our model is that some of these salary workers are illiquid. And that means that if there's a pickup in inflation and their nominal wage doesn't rise, then their spending power goes down. And by the way, my university hasn't given us a real cost of living increase, they're still giving me 3%.

I keep thinking I should get a letter from John Taylor to our new president saying about a nominal illusion or something. But anyway, so this is what's in the model. One more thing to say about the model is that we have a reasonably realistic marginal propensity to consume at the aggregate level, about 0.5 or 0.6, depending on how you calibrate it.

That's consistent with what I think of as one of the best papers on this question, which is the AER paper by Parker and Souleles and co authors. Okay, so with that overview, why don't we dig into some details here. Okay, just a few things about the PSID that are worth keeping in mind.

They're a little bit different than census, so the terminology here is a little bit different than people might be used to. PSID works with families of related people and so a household would potentially include someone who's renting, who's living in a room in the house, okay? And PSID would they know about that person, but they don't collect the same detailed information.

Then the information about the family doesn't include someone who was renting a room or if they're two families that might be sharing a house, then PSID would treat them separately. Okay, another thing is, I'm used to thinking about head of household but in PSID they've switched the terminology.

Now they call it the primary respondent and the spouse or partner. Then they also collect information about other adults and children in the family. And you imagine here, we've looked through it, there's so many different situations. The parents are in their fifties and the daughter, who's in her twenties, decides to move back home for a while and she suddenly shows up in one wave of the PSID.

She wasn't there in the previous one or then she becomes independent again or in some cases, an older adulthood moves in with the family. And there's all kinds of things here, changes in the composition of the family that happen over time. And those actually could be itching to think about for some of the macro dimensions of a heterogeneity.

For example, the extent to which married people might have more of a buffer than single parents or divorced people. These are all questions we think of as much micro questions. But I think the HANK Literature has helped us kind of awaken to the possibility that those things actually could be relevant in some cases, at least for macro modeling.

Okay, so now we're just getting a little more detail here. Here's how we categorize the source of income. So labor income is pretty obvious here, except that we're gonna add unemployment compensation to that category. So we don't wanna think of someone as fixed income. If just because they happened to be unemployed that year and they were getting a regular unemployment compensation check, we're gonna call that labor income.

So that would not be a fixed income household. Hopefully they're gonna get a job and go back to work. Okay, pharma business income, which we're including ownership of shares of stock and dividends. Another thing I should say, by the way, the PSD is great in many ways, but it does not have good coverage of the very top wealthiest households.

And PSID is acknowledged this many times. If you really wanna know about the top 10th of a percent, you got to go to the survey of consumer finance because they work with the IRS to get complete coverage.

>> Bob: Except.

>> Andrew Levine: What's that?

>> Bob: They don't include the 500 most wealthy individuals.

 

>> Andrew Levine: Okay, I think in some of the aggregated things that SCF publishes, even those are included. But you can't get the-.

>> Bob: You can't get microdata.

>> Andrew Levine: You can't get the microdata. So in any way, SCF has some other limitations. So each of these has strengths and limitations.

I just wanna point out here that for the kinds of questions we're focused on today, PSID is great because we're not so much focused on the measures of wealth of the top 10th of a percent of the population. What we're really thinking about here is the source of income of most families, okay.

Taxable fixed income, which would include the trusts, rental properties and interest on CDs and bonds, okay? And then transfer income, which I think we've already talked about before. Okay, so we classify family as fixed income if the transfers, which is what PSID calls transfers, including Social Security or SSID, plus what we're calling the taxable fixed income, is at least 90% of the total.

And we've done plenty of fiddling with this, looking at different thresholds, like 80 or 85, or including or excluding some of these detailed forms of income. It doesn't really matter very much if you want to get to 17% instead of 19%. You can slice things a little bit more finely, okay, but close to 20% of the families are fixed income.

And most of the families in this gallery don't have any other income. They don't have any farm or business income, they don't have any labor income. I would say a complication is a retired person who might be mid to late sixties, who in one year of the PSFD wasn't working at all, had no labor income, but two years later, say, in 2019, decided to get a part time job.

They don't go back to work full time, but a part time job, kind of interesting. And so suddenly their fixed income goes down to 60% or 70% of their total income, and they're not in this group anymore. Okay, so those are, again, the sort of complications working with the real data from PSID.

Okay, let me just show you now these are pie charts. These are for families where the primary respondent is at least 62 years old. And what you see is, on average, 60% of the family income is from Social Security, 31% from retirement pensions, and then little dribs and drops from the rest.

A little bit of welfare, which could be if a child is living with that family and the child is eligible for some public program and a few other things. Really, but mostly this is social security and retirement pensions. Okay, as you'd expect for people over 62 who don't own a business, okay.

Pharma business income here is trivial for the younger families. Almost half of their income is from disability insurance, 19% from welfare, 15% is from retirement pensions, 14% from alimony, child support and gifts. Okay, and by the way, this just reflects the extent to which someone in the family might be an older person who's living with their younger relative, okay.

Might be getting some social security income or a retirement pension. Okay, these are all the complications you have in the real world. But these families have essentially no labor income, no pharma business income, okay. Most of them, many of them are getting disability insurance. Okay, what you notice.

 

>> John Cochrane: You're not telling us anything about how big these groups are.

>> Andrew Levine: I already told you.

>> Bob: But we don't remember.

>> Andrew Levine: Okay, that's all right. I'm gonna show you again in a minute. So, but I'll remind you now, okay 19% of us adults live in these fixed income families.

And of that 19%, I'm gonna say 13% are the older ones and about 6% are the younger ones.

>> Bob: You showed us it's really pretty small 6%.

>> Andrew Levine: Yeah, and it depends on how cold-hearted you are. Ignore them or not, in fact, what we're gonna do for this paper is just add them together with the older ones.

But I think when we start to think about the question John raised earlier, like the COVID stimulus, okay. And how did that affect a single mom with two kids, okay, who was just living on welfare? How did it affect someone with a permanent disability? How did it affect someone who was 68 years old who might have been thinking about retiring and now goes ahead and retires?

I mean, these are all complex questions. And so I think even within this fixed income group there's some richness here, and we can look at it in the data. And we should be starting to look at it, not just rush it away and say, well, everyone's essentially the same like we did in rank models.

Okay, here's this income distribution, I think is helpful just to have it in mind. Okay, the younger households, almost all of them, live below the poverty line. The poverty line for a single mom with two kids is about $25,000 per year, okay? And so almost all the younger households are below that, okay?

But among the older households, it's not true, okay? Most of the older households actually are above the poverty line, okay? So don't think of these as poor people, especially not the older ones, okay? They're not poor, some of them have family incomes of 75, $100,000 or $150,000. It just mostly social security, pension annuities, okay.

Other types of very predictable non cyclical income. Okay, now we got to turn to the liquid versus illiquid. So, Bob, we're gonna take away those fixed income people and just leave them aside now. Okay, we're looking at everyone else. Okay, so we categorize assets and liabilities. Again, it's amazing how much detail there is in PSID home equity, which is net of mortgages and HELOCs, personal vehicles, which is the value of the car, net of the car loans, and obviously not very much.

But for some families, that's their only asset is their car, what we call illiquid assets and liabilities, okay? You own a firm or a business and it has net worth, but it's not easy to borrow against it. So, if you have a medical emergency, it may not do you very much good.

The problem with the business is it can disappear tomorrow or next year, and so it's difficult to go to a bank and borrow against the value of your small business. I don't know so much about firms, so those are really tiny fraction here, so we won't worry about them, okay?

Some people own other real estate, they own some houses they rent out to people, for example. People also have IRAs that are definitely not very liquid, there's a big tax if you wanna cash in early. And then people have longer-term debt, education, school loans, legal debt, medical debt and other kind of debt, PSID asks about all that.

And then people have what would be the Kaplan and Violante and Moll and Wiedner kind of the liquid assets. And so these are transaction accounts, checking, savings and CDs. Marketable securities, which in some of the earlier work, they only included bonds, we're including stocks, it's a minor thing, so we can do it either way.

But we thought we should probably include them, cuz if you had an emergency, you could probably draw down your shares of stock and sell those, call your broker and sell it. And then there's the evolving debt, sorry, again.

>> Bob: Borrow again.

>> Andrew Levine: Okay, if you would know much more, okay?

So, that's how it seemed reasonable to us to add it in here, even though some of those other papers hadn't. So, and then revolving debt. So, basically, you gotta subtract off the credit card balances, okay? So for liquid families, I think I mentioned this before, means they have more than a month of liquid assets.

And illiquid families means they have less than a single month's worth of assets compared to their monthly spending, okay? And fortunately, PSID does very carefully. I don't know if any of you remember this, but 25 years ago, PSID only asked people about their spending on food. Unfortunately, in the late 90s, there was enough consensus among economists.

It would be really helpful that PSID started asking about all the different types of spending the household does. And so we have measures of total spending in PSID, and since we're doing 2017/19, we can use that here, okay? Again, we've looked at alternative specifications, both for the definition of liquid assets and for the specification of the threshold.

And the results are pretty robust, you can go from 19 to 21, sorry, in this case it would be from 40 to 35 or 40 to 45, right, a few percentage points, but generally pretty robust. And that's because most of the families that are illiquid don't have any other assets except for a primary residence or a personal vehicle.

They hold almost no stock, which many of you knew that already, right? Only a fraction of US households own stock other than in their retirement account, and they don't own any business or farm, okay? So the illiquid families are gonna be basically the labor only families, okay? So here's the same pie chart, but now we're looking at the illiquid families, not fixed income, okay?

But they're illiquid, 88% of their total income is labor income. There's a little bit of social security, a little bit of disability insurance, okay? You have to imagine here a family where someone in, like one of the kids might have an older grandparent or who designated them as survivor, and they get some survivor insurance.

So, and I don't know all those details, okay, but there's a little bit of social security here, and a little bit of welfare and a little bit of alimony and a little bit of. But the farm and business income is trivial, okay? And even the interest, rent and trust is really trivial, okay?

So, if you said these are families for whom their only cyclical source of income is employment, that would be a very, very accurate statement down to a percentage point or two.

>> Bob: Indulge me again by telling me what fraction of the total?

>> Andrew Levine: Okay, this is 40% of US adults live in this type of family, 20% of adults, little less, 19%, live in the fixed income families, 40% live in this type of family, that's how we got to 60, okay?

And the other 40% of families live in this type of family, this is the good family to live in if you can get there, okay? They have farmers and businesses, they have interest in rent and trusts, okay? They don't get any welfare, that little black slice of the pie is almost invisible, okay?

And a lot of it's labor income, but I should say here, by the way, another thing we haven't really grappled with much as micro economists is once you talk about executives. A lot of the labor income is things like bonuses and shares of stock and stock incentive plans and all those sorts of things.

It's not hourly wages and it's not a salary. So we really should be thinking of those managers and higher level executives is probably a different part of the labor market. So, this idea of a homogeneous labor market really doesn't make much sense.

>> Bob: There's one thing that Mister Ferry has investigated.

The other group, the young group let's say, this is the old group.

>> Andrew Levine: No, no, no, this has nothing with age anymore, those were for fixed assets.

>> Bob: But the people.

>> Andrew Levine: The illiquid ones?

>> Bob: The liquid ones are older, right?

>> Andrew Levine: I can tell you that in 30 seconds, but not this 30 seconds, I've got it on state, I can tell you very quickly.

 

>> Bob: I've seen other numbers.

>> Andrew Levine: Okay.

>> Bob: And it makes sense anyway.

>> Andrew Levine: And as I said, I think one of the cool things about longitudinal data is we already have started to do it and we can do more. I just tried to figure out how much should we include in one paper, okay?

But we can track families over time, and we can ask the question, what fraction of the families went from being illiquid in 2017 to liquid in 2019? And the answer is small, it's about 3% of these people moved up into liquid. And that's the thing about paycheck to paycheck again, and then we come back to John's question, why?

Why, aren't they saving more? Okay, and I think that's an interesting question we should come to in a few minutes. Well, let me,

>> Bob: I guess, I didn't get to point. It keeps changing, this group, liquid families, ensures the illiquid related families, okay? So, the liquid group, that I would think of as the parents ensure the younger group.

So, there's a contingent aspect that isn't captured by the work that you're doing, but is captured by beast of fairies work that you ought to keep in mind. But it doesn't, it won't fit this into the framework cuz it's contingent.

>> Andrew Levine: Yeah, okay, I mean, it's probably just write it down.

 

>> Bob: But the other group, the illiquid group, has something that has considerable value, and that's an insurance policy that their parents will help them out if they get into bad times. I think those of us who have older kids know about this, right? Knew about it when they were younger, too, when their parents filled them out.

 

>> Andrew Levine: Okay, and it would be interesting to look, I suspect from having looked at, I've spent a lot of time looking at this. It's really cool in the PSID because you can look at an individual family, you don't know their name. But you know where they live, how many kids they have, how old the kids are, what the different sources of income are, and all the rest.

And I would just say that my impression here was that within this category called alimony, child support and gifts. Let's back up here, this is the one for the illiquid families, okay? Alimony, child supporting gifts. That it's almost all alimony and child support. They have one for gifts from relatives, from others, okay?

And I have a feeling here that most of these illiquid families don't have parents with a lot of money. That probably what's actually going on here is that insurance is within this group. That there are people who, they've already got a house, they're starting to build up, some savings, maybe not quite very much, but two or three months of savings, six months of savings.

And they get into trouble, they lose a job, or they need to move to a bigger house, and the parents come to the rescue, but it's already within this group. I'd be surprised, again, I can check because we have the detail from the PSID, how many people within this group.

I can tell you afterwards, later today, how many of them got significant support from a family member that's in the darker green color. Okay, why don't we keep moving around time here. Just briefly, as I said before, PSID has really useful information about, are you paid on the hourly wage, hourly plus tips, salary, salary plus commission.

And here, what we've done is just because they also ask about your occupation. So we took all the PSID workers, no matter what family they're in. Okay, all the workers. And we just look at by occupation, what fraction of people in that occupation are hourly wage and what fraction are salary.

And pretty striking differences here, not surprising, okay, but interesting. So, for example, healthcare support, 96% of the people in that occupation get an hourly wage, and 4% are salaried, okay? In food prep, it's 88.5%, and 11% salaried, okay? Construction, it's 83%, okay? And then you go to the other end, in the legal profession and management and computer science and education, it's 70% or more are salaried workers.

Okay, why? I guess maybe someone could tell me, cuz I would assume there's papers written in labor economics about employers decisions and workers decisions about salary versus hourly wage. But has that been incorporated into micro models? Maybe in these HANK models we should be starting to think about this.

Okay, so this is the summary table, it's a lot of numbers, so I'll just basically say again, what you're seeing here is shares of the adult population in 2017 and 2019. They're pretty stable. So what that says here is there's some inflows and some outflows, but over that two year period, they were roughly pretty similar to each other, especially given it's a survey with some standard errors.

What is noticeable here is the spending shares are not the same as the population shares. And when we come to calibrating the model, the spending shares also matter. So the fact that the fixed-income households have a lower consumer spending than the average median household means they're gonna get less weight in the model.

Okay, so I think we're ready to go to the model now. So I already mentioned it's a frank model with four types of households. The fixed-income households are essentially just gonna be in the deterministic study state, they don't matter for the long linear dynamics, except for the structural parameters.

The wage workers, where we're gonna assume that they have Calvo nominal wage contracts, as conventional in much of this literature, building on John Taylor's earlier work. Salary workers, where their hours are determined by an allocated wage, but their remitted wage can deviate temporarily. And then the executives, and consistent with the PSID, the executives own all the businesses, they own all the shares of stock.

They get all the residual income net of the labor compensation paid to these groups of workers. Okay, here's some equations. I'm not gonna talk about the notation here, except to say that the second equation there is saying that there's inertia in the remitted real wage. And you can see it's a kinda partial adjustment process here in the real wage, which means the nominal wage actually has an indexation to inflation in it.

Which, again, I wish I had at Dartmouth, but that's what we have in the model. The hourly wage workers have the staggered nominal contracts with a cowboy ferry, and then you get a pretty standard Phillips curve that depends on the average allocated wage rate. Okay, what's cool and new about the model is here.

Okay, so this is the New Keynesian is equation. Again, I'm sorry, but I, I said I wasn't gonna use the pointer isn't, okay?

>> Bob: Yeah, yeah.

>> Andrew Levine: All right, since you're sitting here, if I can figure it out, let's see.

>> Bob: I think you're gonna use the pointer.

 

>> Andrew Levine: There we go, all right, so there's a script M here, okay, that's usually called the New Keynesian spending multiplier, and it's a function of lambda. And lambda is essentially the aggregate Marginal Propensity to Consume, and that's here, okay? And a simple point to make is that in the case where there's no heterogeneity, you get back to a HANK model.

The script M is equal to 1, okay? And so this term is completely ordinary. What's new here is that the real wage actually comes into the aggregate demand equation. So just to give you another glimpse of this, you can use recursion to express it in the following way.

So y hat is the deviation of output from potential. And here we have a sum of the real interest rate deviations from the natural rate, so that's like an r- r*, okay? And then you have this script M multiplying the omega tilde t, which is the real wage.

And what you might be able to see, if you think about it for a second here, is this term says that if there's a cyclical shock that pushes up real wages temporarily, then that will add to aggregate demand conditional on the real interest rate gap. And conversely, if there's an inflation shock and real wages drop temporarily, then that will depress aggregate demand.

Now, we haven't worked all this out, we're gonna put it into diner and do some simulations, and eventually this should probably be incorporated into a more realistic model that could be estimated with investment and all the rest. But it seems like this channel could be really important for understanding what's happened over the last couple of years.

We've seen in the data that inflation CPI, inflation picked up a lot more than nominal wage growth starting to close. That gap is starting to close now, but for a year or two, okay? And that's why a lot of these opinion polls and surveys of households and consumers, consumer sentiment dropped a lot because consumers were complaining that their pocketbooks were getting squeezed.

And that's exactly what we're seeing here in this term in this equation, so we think this is a promising direction. Now just to close and then we can have some Q&A, I guess, for ten minutes. What are some directions for further research? I mean, there's details, of course, we talked about like fiddling with the definition of liquid assets and with the definition of the fixed income work, you know, households and so on, okay?

But I think there's some bigger picture issues here that seem important to me. A lot of the work that's been trying to explain heterogeneity and household savings has been focused on heterogeneity and preferences like discount factors. And so that would say, well, if you're paycheck to paycheck over multiple years, that's because you're impatient.

And the wealthy people are patient and the poor people are impatient. Okay, and that's been around for a long time. But I'm not sure it's actually consistent with the data that we're seeing here. It seems like, again, a third of these paycheck to paycheck families are living on fixed incomes.

They're already retired, they don't have any labor income. They're permanently disabled or retired, okay? We shouldn't be surprised, it has nothing to do with patients or impatience, it's their circumstances.

>> Male Speaker 1: It could cut back on consumption a little bit, more acid.

>> Andrew Levine: They could, and some of you may even be over 70 soon, I guess I'm getting there.

Then you ask them, of course, the thing about being on fixed income is it's predictable. And so then the demand for insurance might not be so high. And maybe, I think there are issues, by the way, about financial literacy and financial planning, okay? That may be very relevant here, that even for someone who is 70 years old, having some kind of 1,000 or $2,000 or something for an emergency.

I'm figuring out how they can manage to do that. But another thing I just wanna say is that if you think about the last decade, real interest rates were mostly negative, which meant it wasn't that easy for an ordinary family to save, okay? Higher income families that know a stockbroker tell a different story.

If they can start putting money in the stock market could accumulate wealth really fast. But we're talking about the middle class and working class families, okay?

>> Male Speaker 1: Talking about building up one month's worth of savings, cash in the mattress works for one month.

>> Andrew Levine: Okay, so I guess that, again, we should talk some more and think some more about this.

I guess I have a feeling here that some of these things also connect to. Let me show you a picture here.

>> Andrew Levine: Okay, this is from some really cool work, by the way, that's been happening at the Federal Reserve Board, it's called the Distributional Financial Accounts, DFA. On their website, it draws on the survey of consumer finance and some other source of information.

This is asset holdings by education. And I think it's the education of the primary responder, head of household, okay? And what you see is very, very strong correlation with the education. And in particular, the holdings of corporate equities and mutual fund shares are almost all held by people with a college degree.

The private businesses are almost all owned by people with a college or at least some college degree, okay? And so then I think the question that we shouldn't be asking about patience and impatience as much. Maybe what we should be asking is, well, why didn't this person get a college degree?

Because maybe, in effect, this picture seems to suggest that once that's baked in the cake, and it pretty much is by age 25 or 28 or something, for most people, okay? Then their destiny in terms of whether they're gonna be paycheck to paycheck or not, seems like it's a really dominated by this.

Now, is that financial literacy, is it playing abilities? What is it exactly that makes it so much more likely that for someone with a college degree is able to own a business and build up wealth in the stock market and all these other things, okay? But that seems like the sort of question we're asking, they might be relevant for our macro modeling.

The other thing I wanted to say is that the kinds of issues we're looking at in this paper, heterogeneity in the fixed income households and the liquid versus illicit households. There's lots to do, looking over time and across countries about how those things matter for macro dynamics, so I'll just close with this pair of maps.

Okay, this is from PSID, and the scale here is red for the places with the most liquid adults. And it gets more and more orangey for the ones that have the most illiquid adults, okay? And there's a lot of variation here, I don't know much about Oklahoma, but Oklahoma is really at one end here.

There seems to be a pattern, a state level pattern, of course, we could look at the metropolitan fiscal areas or whatever, there's limitations of PSID. But then here is the latest map from the census household pulse survey, they do this every couple months now. This is the latest one that was released a few weeks ago.

And this is the fraction of households in different states that indicated that they had difficulty paying for their usual household. This is worse than paycheck to paycheck, this is someone who was paycheck to paycheck. But something went wrong, and now they're actually having trouble keeping up with the bills.

And you see it really pronounced patterns here. And it's not just, I mean, we might have thought there's California and Nevada there. Is this related to the high tech developments, but there's Texas and a lot of southern states? So I think we need to keep building HANK models and Frank models and Frank models and whatever the next one is, a flank model or something.

Because these seem like really interesting, important issues that are really relevant for real people. And we should be able to stop talking about hand to mouth and start talking in terms that we can engage with the real people to have a discussion about, including congressmen and senators and others.

So I'll stop there, is there any question?

>> John Taylor: The paycheck to paycheck is the key aspect that you focus on. So how do we change our thinking by thinking of paycheck to paycheck? What is the bottom line that you're trying to drive at?

>> Andrew Levine: I think that we were trying to understand why a pickup in inflation might squeeze consumers pocketbooks.

And the answer is, for a family with a lot of liquid assets, it's just a temporary shock. Under a permanent income office, it shouldn't make much difference, okay. It's for these families that are paycheck to paycheck. And by the way, that would include some of the fixed income ones, if they're maybe their pension's indexed, but it's not indexed 12 months later to the previous realized inflation rate.

That means for a year they're kinda struggling to keep up. And so I think these are the things that.

>> John Taylor: Mickey has a question, Mickey?

>> Mickey: Yeah, Andy, okay, so when we think about paycheck to paycheck, we have to think about what people spend their money on. And for, I'd say the largest portion of the people you've discussed in your PSID data, shelter, food and energy costs, are by far the biggest components of their spending.

And shelter, the cost of shelter is positively correlated with home values, which has increased way faster than the CPI over time. And so a larger portion of expenses are going to shelter. And I'd say a large portion of your survey, excuse me, your PSID are renters. And then there's food costs where it's very interesting there the production.

Productivity in production of food has gone way up, but food costs have kept pace with the CPI and then energy costs are variable. So I think part of your equation should be how the paycheck to paycheck affected by how people allocate their funds. The other point I would make is in your pie chart, there's been a dramatic increase in reliance on income from snap reed, food stamps and disability insurance.

And I would note snap is not only indexed to the CPI, it's indexed to the food at home component of the CPI. So that keeps people whole on the real purchasing power. But I think just the dramatic increase in the cost of shelter that's closely linked over time to home prices that have just persistently outpace the CPI.

Is an area that deserves a lot of attention and of course, is closely related to monetary policy.

>> John Taylor: Why don't you wrap up, Andy.

>> Andrew Levine: Okay, now are there any other questions? I guess, well, thanks, Mickey. I think we can talk offline about these issues, I will say PSID has information about snap.

It's not counted in financial income, but it's there in the PSID. And so among these lowest income groups, that would be part of their total equivalent income, but just not measured here, okay.

>> Bob: Medicaid, in terms of dollars, Medicaid is by far the big one. It's provided in kind, so it's.

 

>> Andrew Levine: Yeah, in kind, okay. So the point is that everything I've been showing you today is not including in kind benefits. So, okay, all right. Well, I think we can wrap up then. Thank you, John.