Social Security And Trends In Wealth Inequality

>> John Taylor: Okay, we're about ready to begin. Thank you so much for being, appreciate it. And I have a title, I don't know if the same title you're going to use. Social Security and Trends in Inequalities.

>> Sylvain Catherine: Yes, that's the title, yes.

>> John Taylor: Yes, co-author's Max and Natasha.

>> Sylvain Catherine: Exactly.

>> John Taylor: Thank you for being here.

>> Sylvain Catherine: Okay, well, thanks again for having me. So this is, as you said, joint work with Max Miller and Natasha Sarin. So let me start. Yes, so I have a relatively short slide deck, I have more slides like to answer questions that come up often in this talk, but maybe we can focus on clarifying question for the first alpha and then I'm happy to to answer deeper questions, comments and protests.

Okay, so the motivation for this paper is this graph. So what this graph shows you is the share of wealth in the United States that goes to the top 1%, and its evolution has been basically over the last 120 years. So this is coming from a work by Gabriel Zucman on Emmanuel says.

So if you look at this graph, it looks like wealth inequality is getting as bad as it was in the 1920s. So is that actually the case? I think to answer this question, you have to first be reminded of why is it in the first place that people accumulate wealth?

Traditionally in economics we think in terms of precautionary savings, so people accumulate waste to be able to maintain consumption in bad times, whether they lose a job or whether they face some unexpected expenditures such as illness. And the second reason they accumulate wealth is to smooth consumption over time, so they will accumulate wealth during their working life to be able to consume during retirement.

The big difference, of course, between 1920 and 2020 is that between those two periods, the government has started taking care of a lot of these. So during the New Deal, the US got Social Security and unemployment insurance, during the Great Society got Medicare, Medicaid, and in the late 1970s there was an expansion of the Social Security program.

 

>> Michael: So before you go any further, you did mention bequests.

>> Sylvain Catherine: Yes, so I felt one would be bequests.

>> Michael: Academic research on request motives and what share of wealth accumulations due to life cycle motivations versus bequest.

>> Sylvain Catherine: That's a contentious issue.

>> Speaker 4: That's also the sort of looking at a part of the stock market here.

And there's this, if the stock market goes up mark to market, we're wealthier, but of course, the cost of consumption-

>> Sylvain Catherine: I'm going to talk about that, I promise.

>> Speaker 4: Sorry.

>> Speaker 5: I thought the answer to Michael is that, that's for sure true. But in the US, I don't know.

If you look at SEF type of data, the survey consumer finances is.

>> Speaker 6: It's absolutely true if you believe the Scandinavian countries, for which you can really trace intergenerationally the persistence of wealth bequests are super important. In the US, if you look at the survey consumer finances, the numbers are so small, but it does seem to matter.

 

>> Sylvain Catherine: Yeah, because people die with a lot of wealth.

>> Speaker 6: Yeah.

>> Sylvain Catherine: Yeah, okay, so what are we going to-

>> Speaker 7: How much of them are planned and how much are unplanned because of lack of annuitization and dying early and all that sort of stuff? Hard to figure it out, but there's maybe 20 papers going back and forth between Franco Mariani and Larry Kotlikoff, Alan Auerbach, Mary Summers about them.

 

>> Sylvain Catherine: Okay, so what are we going to do in this paper? So we are gonna try to take into account Social Security in measures of wealth concentration. So we are going to compute aggregate Social Security wealth, which we are going to define in different ways, either in an NPV way, which is like the present value of future benefit, net of future taxes.

Or just the present value of benefits, but only the shares that you have already accrued, even the taxes that you have already paid relative to your lifetime taxes. For retirees, this will be quite simple, because in the survey of consumer finances, we observe how much benefit they are getting.

And so we can basically value Social Security like a life annuity. For workers, it's a bit more tricky, and I will come back to this. We are gonna rely on Monte Carlo simulations to compute their expected benefit on their present value. Then we will distribute that Social Security wealth between basically the bottom 90% of the wealth distribution and the top ten of the top 1%.

And by doing this, we will be able to recompute the evolution of top well shares between 1989, which is the first year of the survey of consumer finance, and 2019. And just to give you a preview of the result, this is what we find. We find that once we incorporate the value of Social Security accrued benefits, we see no significant increase in the share going to the top 1% since the late 1980s.

 

>> Speaker 8: Your original graph was pretax wealth also, right? If you sell all that, you're going to pay capital gains tax on it, you're not going to get as much.

>> Sylvain Catherine: So in the paper, we are gonna focus on whatever is the measure of net worth in the SEF.

For this graph, we just take the value of sales on Zucman, I don't exactly remember what they are doing. This is based on IRS data, they capitalize IRS capital income-

>> Speaker 8: Pretax, okay?

>> Speaker 7: When I used to fill out my financial disclosure forms in the federal government, they couldn't handle the fact that I would make that distinction.

Stuff that was taxable, don't want me to add the numbers, even if some were taxable and some weren't, too much information for them.

>> Speaker 9: I wanna ask a somewhat impertinent question, which you may wanna postpone, which is, why are we looking at wealth and not consumption? One answer might be we can't measure consumption very well.

Another answer might be we think wealth actually belongs in the utility function separately from consumption. Third answer might be that we think wealth at the very top end at least creates a threat that there's too much political influence in a small number of people. But do you have a view about that?

 

>> Speaker 10: Because a bunch of other people looked at wealth.

>> Speaker 11: The French revolution here, because wealth inequality has spiked so much, right? So that's sort of a policy.

>> Speaker 9: That's what they say. They say that as if that's-

>> Speaker 11: So it's a policy thing not in your framework.

 

>> Speaker 10: Yeah, so it's a bigger number,

>> Speaker 9: I'd like to hear the motivation. Other than other people are doing it and making noise about it, what's the motivation for looking at wealth rather than consumption?

>> Sylvain Catherine: So first of all, the first thing that was mentioned, the fact that this is an important part of the political debate, means that we should politicize We try to measure it as accurately as possible.

Now, from the point of view of economic theory, I agree. What matters at the end is lifetime consumption. There may be concern, if you were to ask, PGTR says about this, there may be concern about political power. But I think what we are trying to do here is just to focus on one particular aspect of the India temporal budget constraint and try to measure it very seriously.

The reason it's gonna be related to the rest of the wealth is something that John mentioned previously, which is that the evolution of the market is closely tied to the evolution of interest rates. And this will also be true for Social Security. And so in a sense, the evolution of the market value of Social Security is going to exactly offset the evolution of the market value of traditional wealth.

And so that's why it's interesting to measure them.

>> Speaker 10: That sounds like another reason to look at consumption. There are already several reasons on the table why wealth is a hard thing to discern. There's the life cycle versus bequest, there's the fact that people gets taxed away and so on.

Just the connection to actual utility is pretty,

>> Speaker 9: Warm defenders some politicized people are making big hay out of wealth. And if their numbers are all wrong

>> Speaker 10: I'm fine with that. But then we also ought to say that they're looking at the wrong thing.

>> Speaker 9: But they don't show up when you

>> Speaker 14: So what's your favorite well measured consumption?

 

>> Speaker 10: Well, you know, consumption's hard to measure.

>> Speaker 10: It's hard to measure at the top end, especially. But that would be a motivation I wanna hear at the outset, okay? Consumption is very hard to measure. So is wealth, by the way, that's already become quite clear in this conversation.

It's not like we got great measures of wealth. So I'll stop on this theme. But

>> Sylvain Catherine: I agree, consumption purely is a better measure of welfare.

>> Speaker 15: We haven't even touched on the maintenance defect, which is this is not total wealth, this is only financial wealth,

>> Speaker 14: human capital.

 

>> Speaker 15: Human capital is much, much more. So if you're trying to relate this in any way that consumption, you'd have to bring in

>> Sylvain Catherine: Human capital, which is like, I think it's good to go step by step. Even measuring Social Security is quite complicated. Discounting human capital and taking into account the fact that you have to work to get it is philosophically complicated.

 

>> Speaker 9: I just want to defend him. Those two make a lot of hay about wealth, they don't use the word own, they use the word control.

>> Speaker 9: So, let's point out that their numbers are empty.

>> Speaker 14: This is a great Graph. So I'm in support of exactly what you're doing

>> Sylvain Catherine: But I think that there is room for a lot of papers.

And people have also written consumption in equations.

>> Speaker 7: I don't know which of the many papers these people produced, but in the early work, they left out housing, the largest asset for majority of American families. And also I don't know if you're what you're doing about. And Bob's point is quite correct, but any of us who have tried to capitalize earnings and into human wealth, if you want something sensitive to interest rates, try that.

Okay,

>> Speaker 14: But you do have durables in the SEF, like how?

>> Speaker 7: Absolutely.

>> Speaker 14: So, a part of the consumption is measurable.

>> Sylvain Catherine: Okay. How it impacts wealth shares. But I agree there are plenty of people you can invite for consumption. I've seen, like recent papers. There's a recent paper in the JPE, actually, that shows that there's been no increase recently in consumption inequality.

Okay, so first let me tell you, in case you don't know the details of it, how Social Security actually works. So you have to pay taxes, that's part of life. So, part of those taxes are going to finance Social Security. Specifically, you're going to pay 12.4% in payroll taxes, of which 10.6% goes towards the old age program.

This only applies up to a cap. That cap has been around 2.5 times the average wage in the economy since the 1980s. And for 2019, that would be around $133,000. So the tax itself, in a way, is regressive, because above that cap, you don't pay any more tax.

But then, overall, the program is going to be progressive because of the benefit formula. So, on the day that you retire, the Social Security Administration will do the following. The first thing they will do is that they will look at your earnings record and they will adjust your past earnings, both for inflation and also for the growth in the real wage index.

Then they will take the best 35 years on your record, and they will compute an average. So this is what we call the average index, yearly earnings. And then your yearly benefit will be a concave function of this average. So, on the first $11,000, you will get a replacement rate of 90%.

Then the marginal replacement rate will fall to 32% between 11,000-67,000 and above this, you will only receive a replacement rate of 15%. Of course, only the part of your earnings that was taxed will count toward your Social Security benefits. And so as a result of this concave function, basically, people that are at the top of the earnings distribution will get a lower internal rate of return or lower replacement rate on their contribution, which makes the program, in principle, progressive.

 

>> Speaker 17: That's partly ameliorated by differential life expectancies, but only partly.

>> Sylvain Catherine: Yes. So there's been like some discussion about that, whether, like

>> Speaker 18: Don Fullerton had an excellent paper on this about, I don't know, 15 years ago going down these demographic elements. For example, high-income earners tend to live longer,

 

>> Sylvain Catherine: yes.

>> Speaker 9: It's also taxed, though, so that tends to make it more progressive.

>> Sylvain Catherine: One of the things that goes against this household on the spousal benefits. So, for example, if you're a household where you have only one member that was working, the second member would receive at least 50% of the benefit of the main earner.

And so this would be also a progressive aspect of Social Security okay.

>> Speaker 19: The timing of claiming a majority of people claim at 62 and are penalized.

>> Sylvain Catherine: Yes. So basically, if you claim one year earlier, you get, I think, an 8% reduction in your yearly benefits, you will also receive those benefits for one more year.

And so whether this is optimal from an NV point of view actually depends on on your life expectancy. For the Average life expectancy, those two effects, to a first order are going to offset each other from a present value point of view. Then, of course, if you expect to live very long, then there is a benefit.

It's a net benefit of delaying the claim of benefits, so how do we define Social Security wealth? So we are going to focus on one measure, which is our measure of accrued Social Security wealth. So what you see in blue on these slides is the present value of all your expected benefits.

But of course, if you're in the middle of your career, you are not fully entitled to those benefits yet because you still need to pay some taxes to be entitled. And so we are gonna say that at any point in your life cycle or for your working life.

The share of the benefits that you have already accrued is basically the present value of expected benefit multiplied by the right part of the equation, which represents the share of the taxes that you have already paid divided by the total expected lifetime taxes to be paid to Social Security.

And so that will be our measure of accrued Social Security benefits.

>> Speaker 20: So why are you including accrued taxes already paid?

>> Sylvain Catherine: So it's not included. So the first part of the equation just represents the taxes already paid divided by.

>> Speaker 20: But you're doing it based on the tax on the taxes.

 

>> Sylvain Catherine: Yes, so basically what we are saying is that if you have already paid half of the taxes that you expect to pay over your lifetime. We are saying that your accrued Social Security benefit is half the benefits that you will receive over your lifetime.

>> Speaker 20: The taxes accurately reflect that given grossing up with real earnings, as opposed to.

 

>> Sylvain Catherine: So the expectation here takes into account the lifecycle profile of your earnings and the cap on everything. So all those numbers will come from simulations.

>> Speaker 21: It's a little surprising cuz you multiply rather than subtract taxes.

>> Sylvain Catherine: Yes, so the alternative method, which I think is the things that you prefer, which was actually the first thing that we did, is a net present value method.

Where basically you would just subtract from expected benefits of present value of the taxes that you still need to pay. And so we also have that in the paper, but we focus on the accrued measure.

>> Speaker 22: What do you do for the unfunded liabilities that are immense? So how you're treating people say under the age of 45, that number is gonna be negative.

 

>> Sylvain Catherine: Can I keep that for the Q&A? Because I have a slides to show you for that.

>> Speaker 9: Yes, tremendous net wealth for the whole economy cuz Social Security benefits didn`t come from Mars.

>> Sylvain Catherine: This is more like a philosophical discussion of what it means, that in aggregate there is positive Social Security wealth.

So it's a claim on future generations. To the extent that this claim exists, I think it should be counted even if it's at the expense of people that are yet to become adults.

>> Speaker 9: It's gonna be a claim on. You're not a future generation, you're gonna be paying this stuff cuz we're running out in about five years.

 

>> Sylvain Catherine: So this will be counted in the NPV, the parts of-.

>> Speaker 9: Income taxes, or almost certainly they're gonna blow through the upper limit on the Social Security tax.

>> Sylvain Catherine: So the part that will be missing. So me, to the extent that I'm already in the labor force, I will have my Social Security wealth from which the future taxes will be deducted in the NPV approach.

What is really the residual is the taxes of people who are yet to become adult or who are yet to be born.

>> Speaker 23: I'm positing the tax rates have to change because-.

>> Sylvain Catherine: So again, I have a slide on this. So either you will get the benefit or you will increase the taxes we have post hypothesis.

 

>> Speaker 14: I guess I had a question about the captee meaning if you have in mind a sufficiently long horizon where you're thinking about balancing the total receipts that are taken into account in computing the Social Security benefits.

>> Sylvain Catherine: So we're going to do that as a robustness. So as a robustness, we're gonna either account only the benefits that are payable given future revenues, or we would increase the taxes, such as the benefits that have been promised can be paid.

Yes,

>> Speaker 24: I'm thinking you're underestimating the accrued Social Security, present value of accrued Social Security because of income growth over the lifetime, right? So, I'm thinking about somebody who's 30. They have probably paid a relatively small share of taxes relative to the benefit that they've earned, or maybe they have to have been in the labor force for ten years.

You see, I'm saying, what justifies the share of taxes already paid? Yes, a measure of the accrued benefit. Can you directly calculate the accrued benefit?

>> Sylvain Catherine: There are two ways to think about accrued Social Security benefit. One is to just say, what would I receive if I stopped working tomorrow?

And the other one is the approach that we propose. So our approach, as you pointed out, is conservative because of the concavity of the formula. Because of the concavity of the formula is basically you stop working at 35, you will have a lot of zeros in your earnings record, which will make it look like you were a low earner and therefore you will get a very high replacement rate.

So we take a more conservative approach where we basically simulate the rest of your lifetime earning, so that we take into account the concavity.

>> Speaker 24: But it's wrong, I mean, its-.

>> Sylvain Catherine: It depends, I'm not sure it's wrong. I would say it's also wrong to assume that everybody's gonna stop working tomorrow.

 

>> Speaker 24: But that's the definition of accrued benefit, I mean, you can't-.

>> Sylvain Catherine: That's just semantic, I guess, then as you said, this is going to be the more conservative approach.

>> Speaker 24: Anybody could stop working tomorrow, just decide that they're gonna take the benefit that they've accrued.

>> Michael: There are court cases on who owns what.

So Supreme Court has ruled that you don't own your Social Security benefits, so Congress could eliminate it tomorrow. But actually, this is obviously a political, demographic and economic series of interrelated assumptions that have to go in here to figure any of this out. But I was trying to make this point earlier that the concavity of the real income growth which is built in is something I have to have.

I wanna raise another issue, so are you thinking, we're thinking about this for each individual, before you sum it up over everybody? Are you thinking they're already kind of optimally annuitized and they're risk neutral, therefore, and we just take the expected present value and we don't need to worry?

 

>> Sylvain Catherine: Can I keep that to the Q7A? So I will have a slide where I think about individuals that are financially constrained and that might value this differently from the market.

>> Michael: Yeah, so they are paid as inflation indexed annuities, maybe slightly over indexed.

>> Sylvain Catherine: Okay, yes?

>> Speaker 5: Michael is asking another very good question but you're doing what those guys did.

 

>> Sylvain Catherine: I'm sorry.

>> Speaker 5: They do some accounting, I think Zwick and others, they do some accounting for some transfers and they use this kind of approach. So, as Zwick and Zidar and there's a various measure.

>> Sylvain Catherine: So no, so in the Zidar and Zwick paper, there's no Social Security, actually they have a graph, but That's coming from our paper.

 

>> Speaker 5: This is another conversation, but there is, in this photo group, there are a couple of papers that were presented, I think, here by Splinter and others from the treasuries. And they show that you can refine their estimates, and they're not too dissimilar if you account for some Social Security transfers.

 

>> Sylvain Catherine: For the income inequality.

>> Speaker 5: For the income inequality, and the treasury group did it also for measures of top wealth. Represented here, I think-

>> Sylvain Catherine: It's the votes paper, the Sable house, there's another paper.

>> Michael: And Splinter, etc. There are a variety of things that Sayes and Zucman and Piketty do.

And as near as I can tell, every one of them, bias is the result of the direction of their conclusion. Now, maybe they're all right, but that's. Yeah, one of them is they make a lot of assumptions about stuff that doesn't show up in the surveys they claim is underreported.

And so Auden and Splinter went to audits, and when they correct for what's in audits, they find that the top income, I forgot it was income or wealth share, had grown by one-fifth of what Sayes and Piketty claimed.

>> Speaker 5: Yeah, very nice years of paper. They tried to do the same with wealth, but it comes and going, depending on the version of the working paper.

They take a similar approach, saying that this is the way even Piketty, the World Inequality database, when people are trying to do something like this, they've not accounted for. And you wanna interject within the debate and saying that if you take their measurement, their measurement is, in this sense, not totally accurate.

 

>> Sylvain Catherine: So, just to be clear, so Auden and Splinter paper is about income inequality. So there is one other paper that tries to look at trend in wealth inequality with Social Security. The big difference in that paper by John Sebel, Hoss and Vos is that they use the same discount rate in 1989 and in 2019, and therefore, they find very different results because they miss the price effect coming from the yield curve.

 

>> Michael: Auden and Splinter is about income, that's right. However, most of the underreporting that they claim is from income from assets.

>> Speaker 5: Exactly.

>> Michael: So when you capitalize them, you get something very similar.

>> Speaker 5: In fact, you can do a back of the envelope, and that's what they informally have done.

 

>> Speaker 27: Well, that stuff is also a big deal. But what survey is showing us is that Social Security is a big deal.

>> Michael: Absolutely. Interesting.

>> Speaker 5: It also depends many non-tech center count preferences, and so you're doing.

>> Speaker 27: Whatever makes sense, yeah.

>> Speaker 27: Yeah.

>> Sylvain Catherine: Okay, so for retirees, this computation is very easy, as I mentioned before, because you actually observe the benefits in the data, and you can use mortality tables to compute the probability of still being alive in a given number of years.

The only assumptions that you have to make is about the evolution of the price index, because Social Security benefits inflation. And then you can just discount as a government nominal yield curve. So this is what we do for retirees, which is pretty straightforward.

>> Bob: And what are you discounting at?

 

>> Sylvain Catherine: So we are discounting everything for retirees using the government yield curve, the nominal one, because we can't go back to 1989 with the real yield curve because treasury inflation-protected securities were only introduced in the 2000s.

>> Bob: You made it sound like there was indexation to the price level.

The indexation is actually to the real wage, which is-

>> Sylvain Catherine: No, that is for workers, once you are retired, the indexation is only on the price index.

>> Michael: The initial benefits, Bob, are adjusted up by real wage growth. They're adjusted up by the change in the CPI.

>> Bob: Okay.

 

>> Sylvain Catherine: Okay, so for workers, it's a little bit more tricky. So what we're going to do is that for each year of the SCF, we are going to simulate earnings trajectory for millions of workers, and we will basically apply the formula for taxes and benefits to compute Social Security cashflows.

 

>> Bob: I'm still stuck on the nominal discount. So these are inflation-protected benefits. So I understand there's an issue with getting an inflation-linked yield curve before 2000, before 1998 or something, but people have tried to construct those series. In fact, I mean, recently the OMB tried to construct a bunch of those series when they were doing their revision of the cost benefit analysis stuff.

 

>> Sylvain Catherine: I think whether you actually subtract the inflation rate from the yield curve, or you use the nominal yield curve and you let the cashflow grow at the inflation rate should keep more or less the same thing.

>> Michael: Is a nominal cashflow.

>> Bob: Nominal and nominal, okay, nominal and nominal, okay.

 

>> Michael: Yeah, absolutely.

>> Speaker 9: Does the Social Security administration make this calculation? I mean, maybe they're not gonna show it.

>> Sylvain Catherine: I'm going to talk about that in three or four minutes, depending on the number of questions.

>> Speaker 30: Good luck.

>> Sylvain Catherine: Okay, so we simulate both the past and future earnings trajectory.

So to do that, I don't want to enter into detail of the stochastic process, but basically we take what's the most recent state-of-the-art method to simulate earnings. We take that from a paper by Fatih Guvnin, which basically matches many moments from both the dynamics and the cross-section of earnings in the Social Security administrative data on earnings.

We need to calibrate the life cycle deterministic trend of earnings as a function of age. So again, we are using another paper by Faith Guvnin, where basically the average wage for each court on each year is reported, and we basically extrapolate those for the future. And so our goal with that process is to basically emulate the Social Security administrative final data.

So once we have all that, we can compute Social Security cashflow, whether there are taxes or benefits, and we can discount them. So a naive approach would be to discount them using the, sorry, the nominal government yield curve on taking into account inflation. And another approach would be to take into account something that was mentioned, which is the wage indexation.

So remember, when you retire, your past earnings are indexed not only on inflation but also on the growth of the average wage in the economy. And so that is going to expose worker to some amount of systematic risk. So I'm not gonna go through that, I have it in the additional slides, if need be.

We have a small asset pricing model to basically compute the market beta of Social Security cashflow as a function of alpha in the future. They are, and so that adds some premium to the discount rates that we apply.

>> Speaker 9: Why the attention to the macroeconomic risk and ignoring all the idiosyncratic risk that people face and where they are in their life cycle, where they're borrowing or it seems-

 

>> Sylvain Catherine: I think it's most interesting to measure, in a sense, the market value from the point of view of a diversified investor. And it's also So interesting to measure the private valuation from the point of view of people. And the difference between the two actually tells you, about the optimal design of Social Security in a sense, and whether the fact that Social Security benefits are illiquid, is a good policy or not.

So, now to respond to your question about the implication of using the market value as opposed to the private value, I think there are two elements. The first one is the liquidity, and I'm going to address this during the Q and A, I can show you what we do to try to get a sense of the private value.

And then there is the idiosyncratic risk, related to your labor income, so this would actually increase the value of Social Security. And that is because of the concavity of the benefit function. The concavity of the benefit function means that basically, when you receive a negative income shock, you're going to get a higher replacement rate than if you didn't receive that shock.

And so the covariance between returns on Social Security, and your own income pass is actually negative. So there is an insurance component of Social Security just like there is an insurance component, to a progressive tax regime, it's the same.

>> Speaker 31: There's an insurance component, for sure but there's also an anti consumption smoothing over the life cycle component.

 

>> Sylvain Catherine: Yes, so that's the liquidity that I will discuss

>> Speaker 32: to defend you. Another way of putting the market question is it's, what is the present value to the government of what the government gives people if that's not worth it to the people cuz they'd rather sell it?

Well, it's still what it's costing us. And, a proposal to why don't we let people sell their Social Security benefits forward. Would go absolutely nowhere in Congress, even though all of us would testify at how much greater utility that gives people to get the cash and blow it all today.

 

>> Sylvain Catherine: Yeah.

>> Speaker 32: People had the cash and were blowing it off today.

>> Sylvain Catherine: This has been a little bit frustrating because one criticism of our paper is precisely that people value Social Security less than the market value. But the people that bring those criticism are in general not in favor of privatized, privatizing Social Security.

And there is some kind of a theoretical tension here, because if you actually think then people value it less to what the government values it. Then it would be some kind of a pie to improvement to actually privatize. So we don't want to make any statement on this, as that sounds dangerous.

But there is the difference between the private value and the market value is a very important object in the sense that. Again, it determines the optimal size and the optimal design of Social Security.

>> Michael: Could you explain to me a little more carefully what you mean by the risk free valuation is the real government yield curve.

I have one year of life expectancy last year. Using the one year rate. Is that what you're doing?

>> Sylvain Catherine: So for the cash flows that is one year away from you, we are using the one year rate. For the cash flows that is two years away, we are using the two.

So we are using the entire yield curve.

>> Michael: What do you do when those are negative?

>> Sylvain Catherine: So I don't think for nominal, the nominal.

>> Michael: Real, risk free, real government yield.

>> Sylvain Catherine: No, I'm sorry, this is a mistake. We are using the nominal yield curve and we are projecting cash flows, we are nominal.

 

>> Speaker 33: But you're not going to live forever.

>> Speaker 5: Ouch.

>> Speaker 33: I'm not going to live forever.

>> Speaker 5: Just very low level clarification. How do you measure taxes?

>> Sylvain Catherine: So taxes will be basically 10.6% of the part of your earnings that is below the gap, what we call the maximum taxable earning.

Remember, the cap is around 2.5 times the average rate.

>> Speaker 5: How does tax apply to individuals reported, SCF reported?

>> Sylvain Catherine: No, all of this is simulation.

>> Michael: So the implicit assumption is the workers bear the entire tax.

>> Speaker 5: Yeah, exactly.

>> Sylvain Catherine: Yes, exactly, so the thinking that we have about the tax incidence is that basically people would have higher wages if Social Security does not exist, higher growth rate.

So the incidence of the part that is paid by the employer is actually taken out.

>> Speaker 5: I guess the question is, why not doing something like this? I download all the tax sim I apply to. I don't have any profile in SCF. I see, but you could apply functional.

Heathcote, you could apply functional. That is straight out of tax sim.

>> Sylvain Catherine: So you need to predict the future here, and because.

>> Speaker 5: If you do that for all.

>> Sylvain Catherine: You can't just extrapolate because of the concavity of the benefit formula. So you need to take into account that it's not linear and so the variance actually matters.

 

>> Speaker 5: I see, but the way they do it, they talk about it in their QG. They do something about it. You didn't like it, okay.

>> Sylvain Catherine: What people, so, yeah, we simulate and we think it's important precisely because of the different discontinuities. There is a discontinuity in taxes at the cap, and there is a concavity of the benefit formula.

Okay, so how do we calibrate this? So for all the parameter of the benefit on tax formulas, we are gonna assume that they scale up with the average range in the economy. And this is what has been observed over the last 40 years. Now what we want to do is when we compute Social Security wealth in, let's say, 1992, we want to put ourselves in the shoes of someone who had a given information set back in 1992.

So all the assumptions that we are making about the gross rate of wages, inflation on the yield curve, we are taking the perspective of, again, someone in 1992. So for the yield curve, we can just look at the data on the nominal yield curve back then. So that's pretty straightforward for inflation on the growth rate of the economy.

What we are doing is that we are going back to annual reports published by the actuaries of Social Security and look at what their forecasts were for inflation and wages. We are using those forecasts in our analysis.

>> Michael: Through time people are looking at, or they have access to the same information and make the same assumptions as the SSA actuaries.

 

>> Sylvain Catherine: Again in a sense goes back to the private value versus the market value.

>> Michael: What happens when they turn out to be radically wrong?

>> Sylvain Catherine: So when the forecast change, the valuation is also gonna change. So the change in valuation is also going to reflect. The fact that the actuaries of the Social Security Administration change their views about what the growth rate of wages is going to be in the next rate return.

 

>> Michael: So getting back to John's original point about the stock market, the Dow is a under 80 0 in 1982. So a lot of this is just reflecting what actually happened in the real world Ex post.

>> Sylvain Catherine: That's the same again for the stock market people might be right.

Here we are, I think the implicit assumption is that the actuaries have some kind of a representative view of what the market view would be.

>> Speaker 9: It's also viewed as what's the cost to the government that they.

>> Sylvain Catherine: Exactly, that's why those reports exist. It's actually to kind of give some indication of what are the liabilities of the failure of the governance.

 

>> Michael: Or some non trivial differences between the CBO projections and the actuaries projections. You might wanna think about.

>> Sylvain Catherine: I don't think BCE Wages is going to be one or 1.5%, is not going to have a huge effect, because, like most of Social Security, waste first is going to accrue to people that are more than 50.

And so how much wages are going to evolve over. So for retirees, it doesn't matter at all and for people that are above 50, just like ten years of like so it's gonna change by a couple of percentage points. So from that we get a huge simulated data set and what we're gonna do is that each worker in the SCF, we are going to match with assimilated workers that has the same age, of the same gender, and has the same wage income in a given survey year.

And then to get an aggregate value of Social Security, we're going to use demographic weights of the SCF. So we want to validate that approach and so we have basically two ways. The first one is that there is a point where the simulation meets the data, which is the retirement age.

So we can see whether the benefits that we predict for each court is going to be close to what is actually observed. And this is what this graph is showing and as you can see, we do a relatively good job at matching the data, which again, is not surprising because the entire income process has been estimated in another paper to actually replicate the Social Security Administrative data on earnings.

The second thing that we can do is we can look at the aggregate value and compare it to whatever the actuaries of the Social Security Administration are finding. So again, remember, we have to measure our accrued benefit measure. I think Josh will be unhappy he's looking at this one it's perfect, because we are going to compare two definitions of accrued benefits, ours and the one from the Social Security Administration.

And we have the NPV approach. So this is what is reported in the reports, the historical reports of the SSA. So the problem, we can't compare that directly to our estimate because we are using the yield curve, the market yield curve, whereas the SSA, for some reason, have their own view on what the yield curve should be, and they are using that for their computation.

But what we can do is that we can use their view of what the yield curve was back then and redo our computation and check whether we find the same thing. And this is what we find so we are broadly consistent when we are using the same yield curve, we find, if anything, we find, like, lower increase so again, our result would be conservative.

Now, if we use the actual yield curve, we get so sorry. So this is what so a second type of assumptions that we are making that we want to validate is the one about inflation rates that we again take from the reports. So ideally, what we would like to have is to have the real yield curve and totally forget about inflation.

That again we can do for starting in the late nineties, where treasury inflation protected securities were introduced. When we do that, we get this result, and this is the valuation that we get, our main valuation, which is basically instead of using the real yield curve, we use the nominal yield curve and we use our inflation forecast that the SSA was making in any given year.

 

>> Speaker 5: Difference between the solid green line and the dotted is using tips, the tips here versus.

>> Sylvain Catherine: So the green line is our estimate with tips but unfortunately, we can't go back in time, back to 89. And our black line is using the nominally curve for which we have, like we can go back in 1989 and using SSA inflation forecast.

 

>> Speaker 5: They're surprisingly close only because the inflation inflation wasn't a big deal. They are closer than I would have expected, no.

>> Sylvain Catherine: Well, but that's good.

>> Speaker 5: Not yet.

>> Sylvain Catherine: That means the SSA was good at projecting inflation at least it was good in the sense that it was similar to the forecast of the market at the time.

 

>> Speaker 9: I can't do the sign fast enough my head so was the tips overstating or understating? Phenomenal.

>> Sylvain Catherine: So I think the biggest mistake we get is like, is for the SCF of 1999. So as you like, the green is much lower than the black. And so our interpretation is that the tips market was not like.

 

>> Michael: It was illiquid then.

>> Sylvain Catherine: Exactly.

>> Michael: The rate was too high.

>> Sylvain Catherine: Yes. And so, and then the two, the two lines converge.

>> Speaker 9: So the difference between the red, the lower lines and the both upper lines is that because the difference,

>> Sylvain Catherine: they mean colors,

>> Speaker 9: the top two lines and the bottom two lines.

 

>> Sylvain Catherine: Okay, so the basic difference is that because the bottom line two lines are using the yield curve that you find in the SSA reports and the top, both the top two lines are using market yield curves. So basically what we are saying is that in a sense, SSL reports are underestimating the liabilities of the government by not using the market yield curve.

 

>> Speaker 9: Because the market yield curve forecast more inflation than the.

>> Sylvain Catherine: So typically what the SSA is doing is that they always assume that real interest rates are going to converge back to 3%. And so all their yield curve is under that assumption. And so typically they have always a higher yield curve than we do for the late period.

 

>> Speaker 9: We observe a lower, lower than 3%. Okay.

>> Sylvain Catherine: Yes. Basically, the way they are thinking about it is that there is some, I think they have an air one type of model in mind, and they look at what is the historical average interest rate, that's 3%. And so we are converging back to that in the long run.

So they believe history more than the market. In a sense I guess that's their philosophy.

>> Speaker 9: That's also their assumption on the rate of return on government securities, which is that's very ambitious.

>> Sylvain Catherine: Okay, so what about our results? So then we can allocate that through the SEF, and this is what we get in a sense.

You already saw it. So the upper line is the case without Social Security, so that the increase in wealth inequalities that we observe in the SCF, this time not in sales on Zucman. And then the lower dotted line is under a risk adjusted valuation that takes into account the systematic risk and the black line is risk free valuation.

So basically, our basic result is that wealth inequality has not increased as much once you include Social Security. And so, of course, that raises the question of why we see this phenomenon so I'm just going to try to describe the economic intuition. We actually have another paper on that.

 

>> Speaker 9: For your next paper this is the top tip of the iceberg, because there's medicare, medicaid, food, strength, housing.

>> Sylvain Catherine: So that's also our next paper. We are working on that right now.

>> Speaker 9: You can say this is only the beginning of the present value of all sorts of stuff government gives people on a means test basis.

 

>> Sylvain Catherine: We are just being cautious.

>> Speaker 9: I'm sure that's obvious. As you notice, we're not cautious around the state.

>> Sylvain Catherine: I try to remain cautious.

>> Michael: So there's been a lot of work on that by Larry Kotlikoff and Alan Auerbach and Bill Gale. And you can argue with the assumptions and what will be permanent will be transitory.

But I think the big question you have to confront is why is your aggregate wealth in total not minus $18 trillion, which is the present value of the unfunded liabilities? And that's because you're ignoring future generations getting stuck with this or what?

>> Sylvain Catherine: Yeah, exactly, the idea is that here we are counting as security, that is a claim on the work of future generation.

We believe that to the extent that the security that claim exists, it should be counted because it allows current people to finance future consumption.

>> Speaker 35: There's a massive redistribution built into current law, massive future.

>> Sylvain Catherine: Yes, because the aggregate Social Security waste has to be at the expense of future generation.

 

>> Speaker 9: Again, taking it all out of the future, it's gonna come out of today's wealthy. This is gonna blow up all today's wealthy,

>> Michael: it will. I agree that unfortunately, which would cause a huge increase in marginal tax rates on the most productive people in this society. And very successful small businesses will be to raise the cap or eliminate it.

 

>> Speaker 9: First thing they're gonna do is raise the cap, and then the next thing they're gonna do is, what do you do in the debt crisis? You do a once upon once and never more wealth transfer.

>> Michael: Yeah, I think, I don't want to speak today. Hopefully we can resist it and get some more sensible reform, but there's definitely what the Democrats want to do.

The Democrats have already had bills that had 200 votes in the House over the years to do exactly that, eliminate the cap.

>> Speaker 9: Well, actually, to try to be constructive, there is a set of extra, I'm sure there's 150 extra graphs you could make, but one would be, we have this big discrepancy.

Suppose it all comes out of the top 1%, suppose it comes out of the 10%, suppose it comes out of the bottom, how do these things look? And what I think we're guessing is that where that discrepancy comes out of would make a big difference on the present value of people's wealth.

 

>> Sylvain Catherine: That's a potential paper we have been discussing, but we haven't written that yet,

>> Speaker 9: we have lots of papers for you to write.

>> Sylvain Catherine: That's good, that's good. But I'm French and I have only 35 hours per week.

>> Speaker 36: Speaking of incentive effects.

>> Speaker 37: But they're very productive hours.

 

>> Michael: Coffee breaks.

>> Sylvain Catherine: Yes, it does include the coffee breaks.

>> Speaker 9: Okay, enough jokes.

>> Speaker 38: And strikes.

>> Sylvain Catherine: Okay, so let me tell you a little bit about the role of interest rates. So I'm taking this graph from another paper by Dan Greenwald on his co-authors that plots basically the evolution of the top welsh shares against basically the inverse of the interest rate.

Which is the price of ten year bonds. And so, as you can see in the data, there is a striking correlation between the level of wealth inequality, market wealth inequality without Social Security, and the level of interest rates. The timing is nearly, nearly perfect. So why do we think that's the case?

So, in another paper, what we do is that we try to compute the interest rate sensitivity of people's wealth as a function of their age, their wealth and their earnings. So what this represents is basically by how much does one person in the ACF, wealth goes up when interest rates go down by one percentage point, so how many percentage points do you go?

Your wealth goes up when interest rate goes down by one percentage point. What you can see here is that, so on the first graph, you have it as a function of log wealth, we scale everything by the wage index, because we have data from like, so we need to scale things.

But the basic message of this graph is that when interest rate goes down by one percentage point. The wealth of the wealthy goes up by 12%, and at the bottom is more like six percentage points. And the reason is that the wealthy are much more invested in long duration asset like stocks, whereas the wealth at the bottom would be more like cars and cash on back accounts.

So whether you look at it in terms of wealth or earnings, you find that the richer have higher duration asset. And so when interest rates go down like they did over the last 40 years, not over the last two years, but over the last 40 years, you mechanically see an increase in wealth inequality.

Because the value of those assets, those long-term assets like stocks and startups, are going to increase. What we can do with our estimates is look at the same graph, once we introduce Social Security into the balance sheet of households, and it looks like this, and here's the reason.

So here we are looking at people who are 40, 45-years-old, and so for them, their Social Security benefits are going to be 2025 years away on average from now. And so the benefits are really a long duration asset whose market value is very sensitive to the level of interest rate.

But importantly, the relationship between wealth or earnings and the interest rate sensitivity of wealth is flipping once you include Social Security into the balance sheet. And so essentially, the capital gains that the very wealthy accrue on tradable assets are going to be offset by the capital gains that the rest of the population is going to get on their Social Security benefits.

 

>> Speaker 5: It's a little non monotone at the beginning, it does mean something, or. I'm sorry.

>> John Taylor: It's hard to hear.

>> Speaker 5: Monotone, but doesn't mean something.

>> Sylvain Catherine: For the red, for the blue, I am not quite sure, to be honest.

>> Speaker 5: Okay, cuz you would think that is exactly offsetting, but it doesn't.

 

>> Sylvain Catherine: So in the other paper, we actually, it's actually a portfolio choice model, and in the portfolio choice model, you would have an exact offsetting effect, because that's a result of the optimization. But in the data it's not exactly the case, but you have to remember that for people at the bottom, we are talking about very little wealth to begin with.

So I'm not sure whether like the duration of their wealth is actually very important to their lifetime consumption because they have very little wealth.

>> Speaker 9: I mean, I guess there is a complaint, which was, I forget, coming from the other side of the table, but I forget which one.

That the present value of labor income, the present value of the dividends you receive by living in a house are very similar to Social Security.

>> Sylvain Catherine: Yes. So we explore that in the other paper where we basically, I will present that as a finance seminar next week. Basically what you want to do is that at a given age, if people have identical preferences, they should all held the same duration portfolio, inclusive of human capital on Social Security.

And then you would have, like when interest rate change, you would have change in the wealth distribution, but no change in the distribution of welfare within a court. But here we are just looking at Social Security. We do not include human capital, but you would have the duration of human capital.

Buys a lot over the life cycle. It can be a long term asset at the very beginning. It very quickly becomes an asset that is of shorter duration than your consumption plan.

>> Speaker 9: And Ben, and also, once you put in all the other benefits, there's nothing so permanent as the temporary government program, as the saying goes.

But those things are like perpetuities.

>> Sylvain Catherine: Yes, we are working on that. We have many papers, but we are trying to Medicare, on Medicare, bit more difficult to model. And also the funding gap is much bigger issue for those programs than for Social Security. So like the certainty of the cash flow is less strong.

 

>> Michael: We also involve, we're going to involve a very serious attempt to adjust for risk because the biggest shock you have after retirement is to health, that you have Medicare as a fallback or your base or your full. It's very different just in the discounted present value. But the amount of it is.

And the amount of the unfunded liabilities are two or three times that of Social Security.

>> Speaker 9: Yeah, ideally what we want, I mean, since money is fungible, we pretend that Social Security taxes pay for Social Security. Ideally, what you want is the present value of all your taxes, net transfers that you can get from the federal government altogether.

We have a sense that this applies that would solve all of these unfunded problems.

>> Michael: And that's what Kolikoff an hourback try to do. You can quibble with how they do it and the notion that they tend to, in my view, exaggerate some things and minimize others. But they have kind of a machine that does that virtually every year, every time there's a policy change, etc.

 

>> Speaker 9: So they say there's like a 2300 trillion of liabilities. And what do they do about who's going to pay for those eventually?

>> Michael: About what?

>> Speaker 9: So they, I know they said there's like 300 trillion of liabilities when you add up all the healthcare and everything else present value.

And what do they do about the taxes and who's going to pay for it eventually?

>> Michael: Yeah, they have that.

>> Speaker 9: They have some assumptions on that.

>> John Taylor: You better make sure you have time to, they're basically trying to get life.

>> Sylvain Catherine: Okay.

>> John Taylor: Trying to get life cycle accounts by cohort.

 

>> Sylvain Catherine: Okay, let me. How much time do I have? 15 minutes. Okay. So let's speed up a little bit. So we have, again, I have like more robustness in the paper. Yeah, I'm going to discuss the main one. I'm actually not going to discuss life expectancy inequality because that's actually built already in our baseline specifications.

So we take into account that people that have lower earnings are going to have shorter life expectancy. That's part of our baseline specification now. But I want to address the funding gap. So this represents, basically, according to the Social Security Administration, the share of benefits that can be paid under different scenarios.

So the black scenario is like a very optimistic one. Everything can be paid medium scenario, which basically depends on macroeconomic and demographic assumption, is in green and in red, you have the worst case scenario, according to the actuaries,

>> Speaker 40: Partially colorblind.

>> Sylvain Catherine: Sorry, so it's, yes, so the IR line is the most optimistic one.

Obviously, everything can be paid. The medium one is some kind of a middle ground on. The worst case scenario is at the bottom. So on the left, you have what was described in the report from 1989. So, as you can see, the funding gap was many, many years away from people at that time.

But if we look at the report from 2019, the funding gap is coming up much sooner. So what we're going to do, we have basically two ways to discuss this in the paper. One is just to say we're going to cut benefits using those lines. So we're only going to pay the fractions that is payable under current.

Under projected revenues.

>> Michael: That is actually literally current law, even though it's not described that way. Because it's illegal to pay the benefits except out of the trust fund, which won't have the funds.

>> Sylvain Catherine: Exactly

>> Speaker 9: And do you do anything about the taxation of benefits?

>> Sylvain Catherine: So in, so we don't know, because, like, it's also true of, like, dividends and everything.

And so the other thing that we discussed in the paper, an alternative is just to raise taxes to be able to pay 100% of the benefit. So the most conservative exercise from our point of view, is to cut the benefits, because, again, if you raise the taxes, part of it is going to fall on people that are not in the data yet.

And so basically, you have some kind of a free meal out of future generations, whereas if you cut benefits, it's going to be much more concentrated on people that are already alive.

>> Speaker 41: Now, back in 83, we did both, but the benefit cut was basically delaying. No, it wasn't on recipients.

It was on delaying receiving money, so.

>> Sylvain Catherine: That's how it's been done in Europe all the time, so basically, you only cut the benefits on people that are not yet retired, but you basically have some kind of progressivity. The younger you are, the bigger the cuts. I guess that there is a political aspect which, for those young people, they don't realize how important it is.

And so they are less likely to vote

>> Speaker 41: on time to adjust. Whereas older people don't have time to judge.

>> Speaker 9: People have to retire at 62, they go out in the street and

>> Speaker 5: one more year. So it's increased for one year.

>> Sylvain Catherine: Yes, good at that.

Okay, so what happens if we cut the benefits to our estimates? So this happens. So, as you can see, it has an effect, so if you really cut, the benefit under the worst case scenario, which is basically the top line among the three, you now have a slight increase in wealth inequality, but not as stark as the ones that you see on the is curve, which is the case without Social Security at all.

So we think that qualitatively, our results are robust. Now, you,

>> Michael: if you take the Cochrane political forecast, the benefits will start with means testing, so the wealthy will start losing benefits first. So there's some question about. That's not my forecast, by the way. Although.

>> Sylvain Catherine: Yeah, so then our result would be even stronger because you're going to substract that wealth from the top.

 

>> Michael: I think it would be wise of you to sort of mention a few of these things and how they affect them because that's going to be a lot of people's minds when they look at this.

>> Speaker 9: Is it going to blow through the income cap on the taxes?

Right? So that's, that's going to really hit that.

>> Sylvain Catherine: What does that mean? That thing is a doughnut. Right? So you would pay taxes up to 150k, then you would pay taxes anymore, and then you would pay taxes again.

>> Michael: Sort of like they did on the investment taxes to fund Obamacare.

 

>> Speaker 42: Because of the number of voters between 120 and 40

>> Speaker 43: I knew there was.

>> Michael: But the one thing. The thing Biden claims he's lived up to is he won't tax anybody under $400,000. That's precisely what you just said, if he gets his way, and if they do anything.

But he'll be gone before they do anything though, the most likely thing is they'll wait until the evil of having to do something, which is what they've done in the past, unfortunately.

>> Sylvain Catherine: So, one concern that you might have is that the wealth data in the SEF is not that good.

And so what we can do is that we can take wealth inequality estimate from other papers, like, for example, say, more recent work by Smithy Downswick, and we can basically adjust their estimates to include Social Security. And as you can see on the panel B, peace doesn't really matter for our conclusion, and then, so John's preferred concept was the NPV approach.

So interesting things happen in the NPV approach, as you can see, qualitatively the results are the same. But under the NPV approach, if you look at the year 2019, there is not much of a difference, but there is a big difference in 1989. And that comes from the fact that under the NPV approach, Social Security wealth can be negative.

So when you enter the labor market, if the expected rate of return on your Social Security contribution is below the discount rate, then the NPV of Social Security would be negative. And so that's why in 1989, the effect of Social Security is relatively small under the NPV approach, even though it's much larger under the accrued benefit approaches, because the accrued benefit by consumption is going to have a positive value.

But overall, the message remains the same, so to conclude, so in their paper says on Zucman, they argue that Social Security should not be taken into account because it would call for the inclusion of other programs that reduce private saving. So discuss all those other programs, and then when, if you start including Social Security, then it wouldn't be clear where to start.

 

>> Speaker 44: Op-ed or paper?

>> Sylvain Catherine: No, that's in the QGE paper.

>> Speaker 9: You shouldn't stop.

>> Sylvain Catherine: So, yes, so we argue that now redefined market wealth is not the right place to stop. So according to our estimate in 2019, Social Security accrued benefit is 49% of the wealth of the bottom 90%.

The issue, of course, is that in the background, there is probably a substitution effect between Social Security and market wealth, the fact that Social Security exists reduces the optimal saving rate for private wealth. And so this could have perverse effect, if you think in terms of policy evaluation, because it would mean that as you increase the amount of progressive programs in the economy market wealth inequality would look worse and worse.

And so that would call for yet more social programs, and so if you want to evaluate whether those social programs are efficient at distributing wealth, obviously you need to include them in your definition of wealth. So we think our extended wealth concept is more relevant for policy evaluation.

 

>> Speaker 5: This is really important motivation, by the way, responding to what Steve had said at the very beginning.

>> Sylvain Catherine: Good point, I didn't think about it.

>> Speaker 9: Incredibly important in the debate, what's going on, says in Piketty and Zachman said, huge increase in inequality, Austin, Smitter, Graham and Erlan, others say, no, taxes and transfers is all flat.

To which the lefty response evil capitalism would have made things except for the benevolent government, to which the response is, no, it was the benevolent government that caused them to stop working and get so poor in the first place. Well, that's a harder that is in fact the right answer, but you can't just leave that off the table and you're doing a beautiful job if not, I'm just pounding my fist even harder on that table.

 

>> Sylvain Catherine: I'm glad.

>> Speaker 45: I like this idea of sort of the doom loop on social programs, and, the work that Bruce Meyer and James Sullivan did on this, adjusting the income for the various benefits suggests, if people don't do that, then they say, we need more kinds of benefits, so this is great.

 

>> Michael: They, they do a lot about adjusting things for like the insurance value of Medicaid and things of that sort that cause big, big differences. But let me just make one, I hope, constructive suggestions I said some of this before, I think you have a way you cut benefits and a way you raise taxes.

But I think it would be really useful to take a look at some of the archetypical reforms that are out there. I think the place to start might be Sean Kogan was on the Social Security Reform Commission some years ago, and they had switching from wage to price indexing and some other things that you did.

 

>> Speaker 46: Thoroughly rejected.

>> Michael: Yeah.

>> Michael: But that's likely to be kind of a republican way to look at at least some republicans to look at it. So if you had like five archetypes that have been discussed in the policy community, it might add some interest.

>> Sylvain Catherine: We have been thinking about writing this as a separate paper, so, because then you could not only look at, whether this solves the problem and for how long, but you could also see who benefits is helped by different.

 

>> Michael: That's what I mean, doing the kind of thing you're doing, but with some of these archetypes.

>> Sylvain Catherine: Yes, in a sense, that paper really started because I was not really interested in wealth inequality as a research topic, and I always assume that Social Security was included in those estimates, because for me it kind of seems like natural to do.

And so we approach this computation with the goal of computing top well shares but then once you have like estimate of Social Security present value, there's a lot of other interesting questions that can be addressed.

>> John Taylor: So your last point is the main point, right?

>> Sylvain Catherine: Yes, that's the main conclusion of that paper.

 

>> John Taylor: How do you support that? What's the thing that we should take away that establishes that for the third point?

>> Sylvain Catherine: That, yes, that's the main takeaway, I guess, of the paper

>> Speaker 9: Graph two.

>> Sylvain Catherine: graph two, yes.

>> John Taylor: So graph two, we already know graph two.

>> Michael: What is the explanation for graph two?

 

>> John Taylor: The explanation, sorry.

>> Michael: Graph two is right here, that's graph two,

>> Sylvain Catherine: Yes.

>> Michael: Give us the explanation why it's the first.

>> Sylvain Catherine: So, in a sense that's like the next paper we are writing, which is more like a theoretical paper. So if you think about the optimal portfolio allocation of an agent, what the agent would want, ideally, is to have a portfolio that it doesn't need to trade when interest rates are changing.

And the form of that portfolio is basically buying in advance all the consumptions that you're going to need in the future, a natural way to do that is to buy a house, for example. Then you don't need to trade, then prices change, and then you just receive the dividend when you need it, now, this.

Is what Social Security does to some extent. It's taking away money when your income is greater than what your consumption should be, and it's giving it back later when the inverse is true. But it does it much more for people that are in the bottom 90% of the distribution than people at the top.

And so people at the top needs to do that privately. And when they do that, they need to buy long-term assets. And those long-term assets are going to appreciate in value a lot when interest rates go down. Whereas for the rest of the population, it's the value of the annuity that Social Security represents that is going to increase a lot when interest rate goes down.

 

>> Speaker 9: I mean, a simpler answer might be, Marty Feldstein was 100% right. That when you give people Social Security, they cut their savings by exactly the amount to preserve the retirement income they would have anyway. And so things are unchanged. Would that be fair?

>> Sylvain Catherine: I think there is a distinction between-

 

>> Michael: It doesn't hold up. It doesn't hold up in peer reviewed trials.

>> Sylvain Catherine: So again, we describe that. We come to the seminar, we describe that in the other papers. There are two effects. One is a substitution effect in terms of the saving rate, and there is a substitution effect within the optimal portfolio in terms of the duration.

And that is very important. Because if you have Social Security, your incentive to buy long-term asset is much reduced because it's already a long-term asset.

>> Speaker 47: And can I, sorry.

>> Speaker 45: Okay, related to this discussion, a stylized fact is that the return on assets for lower income people is less than for higher income people.

 

>> Sylvain Catherine: Yes.

>> Speaker 45: It would be interesting to take your exercise and see what happens when you start considering the return on Social Security taxes.

>> Sylvain Catherine: So in essence, that's what this paper and the other paper is doing. So what-

>> Speaker 9: They choose to own bonds rather than stocks, right?

Actually, the after-tax rate of return to lower income people is much higher than for higher income people, if they would just put it into Vanguard Total Market.

>> Speaker 45: It's true.

>> Speaker 5: But they don't do it because maybe they don't.

>> Speaker 9: Well, they don't.

>> Speaker 45: So the idea is the de facto difference in returns may not be as great once you take into account the Social Security.

 

>> Sylvain Catherine: Yeah, especially if returns are driven by a drop in interest rates. So if the reason stocks performed so well over the last 40 years is because interest rate went down, then this is offset by the fact that the value of long-term bonds, and you can think of Social Security as a long-term bond, has also increased considerably in value due to the drop in interest rates.

 

>> Speaker 9: I'll write Valerie's intro for her.

>> Speaker 9: Why do the poor not build generational wealth? Because we gave them a long-term asset, so they invest in short-term assets that don't build generational wealth.

>> Speaker 5: Can I ask this? I have a quick question following up on this. ] If I were to do the back of the envelope, the bottom quartile of the wealth distribution are in this.

And the debate in the labor literature is that these people are horrendously wealth poor. How much are we underestimating the wealth at the bottom?

>> Sylvain Catherine: So actually, that's an interesting question. Obviously, they are showing inequality is focused on the top. And again, we have not explored like this.

Actually, I have a graph on the appendix slide, if you can show the appendix slide. I have a graph that shows the distribution for black and white, because that's a question that has been asked a lot. And I have the entire histogram, so I could show that.

>> Speaker 9: But your question is actually very good.

Rather than just framing in terms of shares at the top, how much wealth is there at the bottom is actually a much more interesting, to everyone around this table, question. Because we don't suffer from ending.

>> Speaker 5: There is a lot. If I think about all Bruce Meyer's calculation, then a lot of programs that are phased out at the median, actually below the median wealth, are prominently present below.

And so I think that the bottom quartile is completely misunderstood and mismeasured in the first place.

>> John Taylor: Why don't you tell us about this, then we gotta stop.

>> Sylvain Catherine: So this graph just shows the distribution of total wealth, inclusive of Social Security, for black and white in '89 and in more recent years.

And so as you can see in '89, you had a lot of Black Americans with basically zero wealth, even accounting for Social Security. Whereas in recent years, the distribution, at least in its bell, has kind of converged. But, of course, if you look at mean wealth gaps, they remain large.

Because something that has happened is that at the very top of the distribution, you have a lot of white households and not so many blacks. And to some extent, that is related to the question of whether we look at mean or medians, for example. So what we show in another paper is that if we look at median wealth as between different racial grouping in the US is kind of converged.

So here on panel C, you have the median black wealth inclusive of Social Security, or Hispanic wealth as a fraction of the median white wealth. And you see this convergence, but if you look at mean, you don't see it. And that is really telling you that what is actually happening in the bell of the distribution is very different from what is actually happening in the tails.

And when you look at means, obviously tails are going to be very important. And so, yes, not just looking at top portions, but that the entire distribution is also important.

>> John Taylor: Thank you.