Patrick Adams: When Stock Crashes Matter for Long-Term Investors | Rational Reminder 403

0:08 This is the Rational Reminder podcast, a weekly reality check on sensible

0:11 investing and financial decision-making from two Canadians.

0:14 We are hosted by me, Benjamin Felix, chief investment officer,

0:16 and Cameron Pasmore, chief executive officer at PWL Capital.

0:21 Welcome to episode 403.

0:25 Another super interesting episode.

0:28 And you know, you said a line at the end of this one, Ben,

0:31 in our conversation with Patrick Adams

0:34 that people love research that supports 100% equity.

0:38 Well, today is not one of those days,

0:42 which is kind of the point of the podcast, right?

0:44 Just to get more information to help people make,

0:46 as you've said 403 times, better, smarter investment decisions.

0:52 And this one was particularly interesting from that standpoint because it

0:56 is a different piece of research which I'll let you describe.

1:00 The second interesting piece I took away because Patrick's

1:02 a PhD candidate at MIT and he talked about the impact

1:07 of your work on the podcast on him and his cohort

1:11 that are looking for subjects for their their PhD work.

1:17 So I thought it was really interesting.

1:18 So with that, why don't you tell us more about Patrick,

1:22 his work, and how you discovered him?

1:24 Yeah.

1:24 So he he did say about the podcast that it

1:26 sounds like a lot of PhD candidates listen uh because it's,

1:31 you know, it's a medium where you can sit down

1:34 and listen to a very accomplished researcher in many cases.

1:38 uh in in this case uh Patrick as he says at the end of the conversation is is

1:42 a relatively new researcher but we're we you still

1:44 get to dig into his research but you get

1:46 to sit down with people who have done research

1:48 um and hear them talk about it in plain

1:51 language uh and really tease out the relevant pieces

1:54 of it for people who are making financial decisions today.

1:56 So it is a I mean it's an interesting medium.

1:58 If I were a PhD candidate I'd probably want to listen to this kind of thing too.

2:03 Uh so Patrick is as you mentioned Cameron a PhD candidate at MIT.

2:07 Uh his research interests include asset pricing,

2:10 household finance, international finance and macroeconomics.

2:13 And he is on the 202526 academic job market.

2:19 Um yeah, I'm sure an interesting time in in his life.

2:22 I I know the closest thing I have to that is when

2:25 I I I mean I guess other than looking for a professional job,

2:28 but I remember being uh a basketball player going around touring

2:32 at universities to trying to decide where to where to commit to.

2:36 I'm I'm assuming it's kind of like that except maybe maybe a little different.

2:42 It's probably it's probably way different.

2:44 Um, prior to MIT, and this is kind of interesting,

2:46 Patrick worked as a senior research analyst in the Federal

2:49 Reserve Bank of New York's macroeconomic and monetary studies function,

2:54 and he graduated from the University of Connecticut.

2:57 Now, his research that we talk about,

2:59 so this is uh this I I discovered his research because uh Jonathan Parker

3:06 uh who's at MIT is on Patrick's

3:09 dissertation committee and he tweeted Patrick's paper.

3:14 I don't even know if I follow him on Twitter,

3:15 Jonathan Parker, but I somehow his tweet shows up in my feed.

3:19 And so I opened the paper and flipped through it and I was like, "Oh man,

3:23 this is so good." Uh so I messaged him right away

3:26 and asked if he uh if he'd be interested in coming

3:28 on the podcast and he told me that he's you

3:30 know listened to many episodes that he's he would love to.

3:33 Uh so we agreed he'd come on.

3:35 Uh but the paper's basically like okay we

3:38 know stocks are not that risky for long-term investors.

3:42 Um, but he asked, "Okay,

3:45 we understand that, but they are kind of risky for long-term

3:50 investors if you need to sell your portfolio in the short term,

3:53 which basically makes you a short-term investor." And so,

3:56 he did something that's like, I mean,

3:58 it's one of those things where it's like, oh, of course, we should look at that.

4:01 Uh, he looked at empirically how high-income households behave

4:06 with respect to the stock market during bad times.

4:11 Uh, and the findings I I won't I won't spoil them.

4:14 Um, but they're they're very interesting and they do

4:18 call into question being 100% equity for your liquid wealth.

4:24 Um, which sounds it sounds like a big finding

4:28 and I I I don't want to diminish the the research.

4:31 It is a very interesting finding.

4:33 Uh, but I think it boils down to and we talked about this during the episode.

4:37 You need to have an emergency fund.

4:40 You can invest in stocks, but your over overall asset allocation,

4:44 including your emergency fund, uh is going to be very conservative.

4:48 In your liquid accounts, like your non-retirement accounts you could sell from,

4:52 uh when you account for, you know, a 12-month of expenses emergency fund,

4:55 if you think about your overall wealth,

4:57 when you have relatively low liquid wealth,

5:00 uh your equity allocation in your liquid accounts, of course,

5:02 is going to be super low because it's mostly in it

5:05 should mostly be in an emergency fund in many cases.

5:07 And then as as your wealth increases,

5:09 your overall allocation to equities can increase.

5:13 But I think this this research does kind of flip flip

5:16 on the head the idea that well young investors should be 100% in stocks.

5:19 Um and I think he does that in a very a very interesting and and useful way.

5:24 Yeah, it's really interesting intersection of you know stock returns,

5:28 the risk on your own personal income,

5:30 your labor income and as he said the consumption commitments you've made.

5:36 So you intersect all these and that's what

5:38 helps you get to an asset allocation information piece.

5:42 It really makes you think about the added risk of consumption commitments.

5:50 So that's like a mortgage.

5:52 You're not going to miss a mortgage payment.

5:54 Yeah.

5:54 And so if the stock market drops and your income drops at the same time,

5:58 which is empirically what he's showing tends to happen,

6:01 particularly for very high income earners,

6:04 like you're not going to for you're not going to let your house

6:07 get foreclosed on or whatever happens if you start missing mortgage payments.

6:11 um you're going to sell some of your stocks

6:12 and if you sell them when they're down a lot,

6:15 all of a sudden the whole stocks for long-term investors thing,

6:18 it kind of kind of dies and and you're you're really uh you're you're

6:23 taking on a huge implicit cost by selling

6:27 your stocks at the worst possible time.

6:30 Anyway, agree.

6:31 It's the kind of thing where it's like, oh well, you wouldn't do that.

6:32 You wouldn't sell your stocks when they're down.

6:34 But Patrick's research is showing, well,

6:35 a lot of high- income earners, in fact, do that.

6:38 Do Yeah.

6:40 So, we can say, you know, stay stay disciplined,

6:43 that you have good behavior, look at the data, all that kind of stuff.

6:46 Um, but Patrick's showing in a lot of cases people just aren't able to do

6:51 that because they don't want to stop

6:52 paying for daycare or miss their mortgage payment.

6:55 Anyway, super interesting research.

6:56 So, we we talked about that paper with Patrick.

6:58 We went through kind of all the thinking behind

7:00 it and how it affected his thinking about asset allocation.

7:05 Excellent.

7:06 Okay, let's go at episode 403 with Patrick Adams.

7:09 Let's go.

7:13 Patrick Adams, welcome to the Rational Reminder podcast.

7:17 Dan Cameron, thank you very much for having me.

7:19 So, you know, as I've told you before,

7:20 the podcast played an important role actually

7:22 in the research I'm going to talk about today.

7:25 So, it's great to be here.

7:26 Uh I do just need to first state

7:28 one important disclaimer since the paper uses data

7:31 from the US Census Bureau and that's just that they

7:33 haven't reviewed the paper for any accuracy or reliability.

7:36 Uh they don't endorse its contents and all

7:38 the conclusions I'm going to express today are

7:41 purely my own and don't represent the views

7:43 of the Census Bureau or any other government agency.

7:46 However, the census has reviewed the results that I'm

7:48 going to talk about to ensure appropriate access,

7:51 use, and disclosure avoidance protection of the confidential

7:54 source data used in the paper.

7:57 Interesting.

7:58 Interesting.

7:58 Uh disclaimer.

8:00 Uh and also very cool to hear that this podcast played a role

8:04 in your kind of thinking and your your approach to the research.

8:09 Definitely.

8:08 Yeah.

8:09 Uh okay, so let's kick this off.

8:10 And and by the way, I read your paper and I as soon as I read it,

8:13 even when I got to the abstract before I had even read the paper,

8:16 I was like, "Oh man,

8:16 we need to have Patrick on because it fits so well with a lot

8:19 of the other uh research we talked about in the last couple years." Uh okay.

8:24 So to to to get us started here, can you explain why stocks are often

8:28 characterized as relatively safe for long-term investors?

8:32 Sure, definitely.

8:33 So, in a lot of conventional investing wisdom, particularly for US investors,

8:38 stocks are considered safe for a lot of long-term

8:41 investors because in the past in the US data,

8:44 they've historically had relatively safe returns

8:46 over long holding periods such as 5, 10, 20, or 30 years.

8:52 Even when we adjust those returns for inflation,

8:55 going back all the way to the start of the 19th century,

8:57 there's not even a single 20-year period

8:59 where stocks had a negative cumulative real return.

9:02 And the same is not true for inflationadjusted returns

9:05 on other safe investments like cash or long-term bonds.

9:09 Now of course over shorter holding periods stocks can look quite

9:12 risky and even in the 20th first century so far we've already

9:16 seen two separate 24-month periods where those real cumulative stock returns

9:21 fell as low as for uh negative 40% for the overall market.

9:26 Now, one of the key empirical facts in finance, though,

9:29 is that stocks have historically recovered from many of these large

9:32 price crashes within the span of often just a few years.

9:36 This is one particular example of Robert

9:38 Schiller's Nobel Prize-winning insight that of course

9:40 stocks tend to be excessively volatile and returns mean revert over time.

9:45 Now, all of these observations are

9:47 what ultimately motivate the portfolio prescriptions,

9:50 for example, in Jeremy Seagull's famous book, Stocks for the Long Run,

9:53 that stocks are a relatively safe investment for long-term buy

9:57 and hold investors who can avoid selling during these short-lived price crashes.

10:04 When does this logic break down?

10:06 Definitely.

10:07 So, I think one of the most important ways that this logic can break down is

10:12 if you have to sell or rotate out

10:14 of stocks during one of these large price crashes.

10:17 And essentially, by selling the dip,

10:19 you're going to miss out on the recovery that tends to follow.

10:22 I think a few examples here help to illustrate this.

10:25 So, suppose you're an investor who starts the year 2008

10:28 with $100,000 invested in a broad US stock market index.

10:33 Now, by the end of the year,

10:34 you're going to be down almost 40% from where you started.

10:38 Now, if you stay fully invested in the stock market in this case,

10:40 you would recover most of that initial investment

10:43 in real terms by roughly the end of 2012.

10:46 However, if you shift out of stocks at the end

10:48 of 2008 into some safer store of value,

10:52 then shift back into stocks only later at the end of 2010,

10:56 it may ultimately take you as late as 2016

10:58 to recover that initial investment from before the crisis.

11:02 We could even do the same exercise at the start of the year 2000.

11:05 A buy and hold investor there would ultimately lose more

11:09 than 40% of their initial wealth by the end of 2002,

11:12 but would be roughly back to the level they started at around the end of 2007.

11:17 Now, if that investor instead shifted out of stocks at the end of 2002,

11:22 then back in at the end of 2004, they would potentially have to wait until well

11:26 after the financial crisis to recover that initial investment.

11:30 But in either case, if that investor instead had to withdraw

11:33 $50,000 from their brokerage account at the bottom of the market,

11:37 they would wipe out nearly all of their financial wealth

11:40 and would barely benefit from that recovery in stock prices.

11:44 So, as portfolio guidance for long-term investors,

11:47 stocks for the long run really relies crucially on your ability

11:51 to avoid selling your holdings during these large price crashes.

11:54 That's right, man.

11:57 And so you you you have done this research that asks what is kind of such

12:03 an obvious question now that you've asked it

12:05 um but no nobody had really done before.

12:07 So empirically what do typical high-income households flows

12:11 into and out of the stock market look like?

12:14 Right.

12:14 So in the paper I'm going

12:16 to ultimately use information from individual income tax

12:18 returns to study these savings behavior

12:21 and investment decisions of highincome working age households.

12:25 And just to set the stage,

12:27 these are going to be households that rank in the top 20%

12:30 of the wage and private business income distribution within their age group.

12:34 So those households collectively own about 1/3 of total US household

12:39 stock market wealth including retirees and about half of that wealth

12:43 is held in taxable brokerage type accounts while the other

12:46 half is in tax advantage retirement accounts like IAS and 401ks.

12:51 So the main measure that I construct in the paper is

12:54 their flows into and out of the stock market within their taxable accounts.

12:58 And I'd love to get into the details of exactly how

13:00 that's measured from the information we see in the tax data.

13:03 But I think what's really striking about these measured

13:05 flows is that they are very strongly procyclical.

13:09 So during these major stock market crashes in the early and late 2000s,

13:13 we see large average outflows from stocks among these households.

13:17 In contrast with pretty steady inflows

13:19 during the expansion periods before and after.

13:22 And moreover, those large average outflows are pretty concentrated among a small

13:26 subset of investors making really large draw downs from their existing holdings.

13:31 That's to say the proportion of individuals who are liquidating a large

13:34 share of their existing holdings varies a lot over the business cycle.

13:39 So in the data, many highincome working age

13:42 households do not seem to behave like that idealized

13:45 long-term investor who can weather the storm and ride

13:48 out these temporary crashes in the stock market.

13:50 And what I wanted to understand on the basis

13:52 of that is what drives their savings

13:54 and investment decisions and what implications does this have

13:58 for how they should allocate their financial wealth.

14:02 Now you touched on this at the beginning,

14:04 but what data set are your results based on?

14:07 Right.

14:08 So the main data set that I'm going to be

14:09 using in the paper consists of some select information we see

14:13 from individual income tax returns for a very large random sample

14:17 of US tax filers over the period from 1998 to 2023.

14:23 So right now tax season is approaching and you or many of your listeners

14:27 may be filling out your own individual income tax return form 1040.

14:31 Now, that form has many different lines on it that report the total

14:34 income that you and potentially your spouse

14:37 have received from many different sources,

14:39 like a job that pays wage and salary income,

14:42 any dividends or interest that you receive

14:44 from your asset holdings outside of tax advantage retirement accounts,

14:48 and even any income that you receive from a closely

14:51 held private business like an S corporation or a partnership.

14:54 Now, the information that I work with ultimately consists

14:56 of the dollar values from those lines on form 1040.

15:00 And I want to take one step back here.

15:02 This is really an example of what we

15:04 call administrative data and finance and economics research.

15:07 Now, over the past 20 years, there's been a huge surge in research

15:11 using big data provided by governments and large

15:14 financial institutions to uncover new facts about

15:17 individuals risk exposures and their economic decisions.

15:21 Now, the advantage of these data sets is their large size and broad scope.

15:25 They allow us to follow, in my case,

15:26 millions of individuals over long time periods spanning multiple boom

15:31 bust cycles in the stock market and the real economy.

15:34 But the challenge of working with these data sets is that they were often

15:36 collected for a different purpose than what we want to use in our research.

15:40 For example, there's not much information in these tax

15:42 forms about how many stocks you purchase this year

15:45 because the IRS really just cares about the capital

15:47 gains taxes that you'll eventually pay when you sell them.

15:51 So for that reason we often have to be a bit creative to get

15:53 the information we want from the information

15:55 that we have in these government data sets.

15:59 I I do want to ask more about how you estimated the flows

16:02 from the data data that you have but uh I want to ask first.

16:05 So you you mentioned that the flows into the stock market are proy proyical.

16:11 How do the flows into bonds differ from the flows into stocks?

16:15 Right.

16:15 It's a great question.

16:16 And so if we look at the aggregate level for this group of households,

16:19 their flows into safe fixed income assets look

16:22 much less procyclical than their flows into stocks.

16:25 So I construct a similar measure for their net

16:27 flows into say safe bank deposits uh government bonds,

16:32 municipal bonds and other safe fixed income assets.

16:35 During these large stock market crashes,

16:37 many of these households are actually actively moving into fixed income

16:41 even though interest rates tend to be quite low at these times.

16:44 So again here, many of the highincome working age investors I

16:48 study in the data do not appear to behave as our idealized

16:51 long-term investor who leaves their portfolio untouched during these large stock

16:56 market crashes or even actively buys the dips in the stock market.

17:01 Okay.

17:01 So you're working with income tax data tax return data which

17:06 doesn't tell you it doesn't have flows like there's no flows data.

17:09 How much stock did people buy?

17:10 So how how did you estimate flows for the research?

17:13 Yeah, exactly.

17:14 So again, the challenge of working with the income tax

17:17 data is that form 1040 doesn't have a line that reports,

17:21 as you say, this is the amount of stock that I bought or sold this year.

17:26 It does have a line that reports this is

17:27 the amount of dividend income I've received this year,

17:30 which provides some indirect information about my taxable stock holdings.

17:34 And I ultimately use that information to estimate households

17:37 net stock sales and purchases using something called capitalization methods.

17:42 So essentially, if I see your dividend income grow much faster

17:45 than the overall stock market's dividends over a given time period,

17:48 I'm going to infer that you bought a lot of stock then.

17:51 Whereas, if it falls a lot more than the market,

17:53 I'm going to infer that you sold stock over that period of time.

17:56 The capitalization part of the name refers

17:58 to how we ultimately estimate the dollar

18:00 value of those stock sales or purchases from that change in dividend income,

18:05 just like an investor would value a stock based on its current dividends.

18:09 Now to take a step back,

18:11 this measure can be pretty noisy

18:12 for a single investor over a single time period.

18:15 Of course, some people hold stocks like

18:17 Amazon that don't pay dividends and other people hold stocks whose dividends may

18:21 fluctuate more than the overall stock market,

18:23 even if they don't actively buy or sell.

18:26 So, a lot of work goes

18:27 into validating these estimates using information from other

18:30 data sources and ultimately determining what we can and can't say with them.

18:35 The bottom line there, I think,

18:36 is that when we average over thousands or even millions of people

18:40 in a given year that I see in the tax data,

18:42 we can actually get a pretty good estimate

18:44 of their average flows into or out of the stock market.

18:49 This is really interesting.

18:51 So, how would you describe the financial situation of the household you studied?

18:56 Right.

18:56 So, I think to understand their financial situation,

18:59 it's ultimately useful to focus on their assets and income.

19:03 you drawing on some data outside of what

19:05 we see directly in the tax returns and some

19:07 surveys from say the Federal Reserve that provide

19:10 a really detailed snapshot of their balance sheet.

19:13 So first in terms of their assets these are generally households

19:17 with pretty high wealth and net worth in absolute dollar terms.

19:21 However, a large share of that net worth tends to be

19:24 tied up in the form of assets that are not very liquid.

19:28 These are things like a large house or a private

19:30 business that may be difficult to sell on short notice

19:34 or retirement accounts like a 401k or IRA that they would

19:38 have to pay a significant tax penalty to actually withdraw from.

19:42 If we focus on just their liquid financial assets,

19:45 that is their bank account balances, stocks and bonds that they hold in taxable

19:49 accounts and mutual funds or other assets that they can draw down on without

19:54 paying these high transactions costs or tax penalties.

19:57 That's a smaller share of their net worths.

19:59 And just to put some numbers on it here, if we focus on 40-year-olds,

20:03 say in the top 1% of the earnings distribution at their age group,

20:08 the median household in that group has only about 2/3

20:11 of their pre-tax earnings saved up in liquid assets or about 8 months.

20:15 If we look further down in the earnings distribution,

20:18 that number is even smaller.

20:19 So if they had to tap into those liquid assets to cover

20:22 their typical expenditures over a period of even just a few months,

20:26 they would potentially exhaust a large share of them.

20:29 Now the most important asset they own

20:32 that doesn't directly appear on any balance sheet

20:34 and this been the discussion of a lot

20:36 of work on this podcast recently is their human capital.

20:40 So even though these households own a lot of financial assets,

20:43 the vast majority of their spending is financed not by returns on those assets,

20:48 but by the income they earn from their job or a closely held private business.

20:52 Now, I think one of the most important empirical facts

20:55 that we've learned from work with administrative data over the past

20:58 two decades is that this income is actually quite risky

21:01 and quite correlated with the stock market for many top earners.

21:05 in particular Fatiguan at the University of Toronto

21:08 and his collaborators have shown in a series of very important

21:11 papers that many of these top earners experience large

21:15 declines in their earnings around these major stock market crashes.

21:19 So, for those same 40-year-old households that start out before

21:22 the crash in the top 1% of the earnings distribution,

21:26 about 17% of them are going to lose half or more of their annual

21:30 earnings during the stock market crashes in the early and late 2000s.

21:35 Now, these people don't necessarily go through prolonged periods

21:37 of unemployment like those at the bottom of the income distribution,

21:41 but they may have bonus pay or stockbased compensation that's

21:44 very sensitive to the stock market and the overall economy.

21:48 uh they may have to switch from an initial high-paying job to a different one

21:51 that doesn't pay quite as well or they

21:53 may own a private business with very volatile profits.

21:56 But to sum up, these households that I study ultimately

21:59 have a lot of illquid assets like housing and retirement accounts,

22:03 a smaller share of liquid assets

22:05 and bank accounts or taxable brokerage accounts,

22:08 and volatile labor or private business income that may force them to draw

22:13 down on those liquid assets exactly when the stock market is doing poorly.

22:18 So if we look at just those liquid assets,

22:21 what proportion does the typical household in your sample invest in stocks?

22:26 They do.

22:26 And the typical average share of liquid wealth that's invested in stocks

22:30 for the typical household in my sample is about a quarter or 25%.

22:35 Although this varies a lot across households.

22:37 Some of them have no taxable stock holdings at all and may hold stocks only

22:41 in their retirement accounts while others have half

22:44 or more of their entire liquid wealth invested in stocks.

22:48 And I think it's important also here just to remember that this measure of total

22:51 liquid wealth in the denominator there includes assets

22:54 held outside of brokerage accounts like bank deposits.

22:57 That's the single largest liquid assets on most households balance sheet.

23:02 Um if we look instead at their holdings

23:03 within their retirement accounts that average

23:05 stock share is quite a bit higher at around 60% for these households.

23:09 Uh but the stock share of liquid wealth

23:11 also varies a lot across age groups ranging

23:13 from less than 20% around age 30 to more

23:17 than 30% around 60 as households approach retirement.

23:20 And that increasing stock share over the life cycle is essentially the opposite

23:24 of what would be predicted by many

23:25 portfolio choice models that have safe bond-like human

23:29 capital or with strong horizon effects where

23:32 young households with long expected holding periods

23:35 have a better ability to ride out

23:36 these short-lived price crashes in the stock market.

23:41 So in addition to age, how does this vary across the income distribution?

23:45 Right?

23:46 So higher higher income households tend to actually hold

23:48 a larger share of their liquid wealth in stocks.

23:52 And you know I think this is in part attributable to the fact

23:55 that more of their financial wealth is held in those taxable

23:58 liquid forms rather than tax advantage

24:01 retirement accounts due to the contribution

24:03 limits on those accounts that are particularly binding for high-income people.

24:08 However, their stock shares of their total financial

24:10 assets including their retirement accounts are also higher.

24:13 And I think this is a bit of a puzzling

24:15 fact because these higher inome households actually face

24:18 more income risk in their jobs and their private

24:21 businesses compared to households further down in the earnings distribution.

24:25 So my own view here is that these patterns probably reflect

24:29 some differences across households and their risk aversion with more risk

24:33 tolerant people selecting into highinccome

24:36 high-risisk professions or businesses but also

24:39 at the same time taking on more risk in their financial portfolios.

24:43 H okay.

24:44 So what what is the relationship

24:45 between household income and stock market flows?

24:48 Like like kind of like you're just talking about it.

24:50 Are the highest income households different in that sense?

24:53 Exactly.

24:54 They are different because their flows into the stock market are much

24:57 more volatile and proyclical than households

25:00 further down in the income distribution.

25:02 I think there's two main facts that really drive this.

25:05 The first is they have more liquid wealth overall and as I mentioned just now,

25:10 they tend to invest more of that in stocks.

25:12 So they have a lot more to potentially sell.

25:15 The second is that their non-financial income,

25:17 their wage and private business income is again much more volatile

25:22 and correlated with the stock market compared to lower inome households.

25:25 And this is really again a fact that we only

25:27 learned somewhat recently when researchers gained access to these large

25:31 highquality data sets where we can follow these uh top

25:35 earners over many years and multiple uh real economic business cycles.

25:40 So these higher inome households have both

25:42 a larger stock share of their liquid assets

25:44 and a greater potential need to draw down

25:47 on those assets during a stock market crash,

25:49 which is exactly what we see them doing in the data.

25:53 So what tends to be happening to a household's non-financial

25:56 income when they're actually taking money out of the stock market?

26:01 Right?

26:01 So it tends to be falling.

26:02 And this is one of the nice advantages of the income

26:05 tax data that we can really zoom in on individual

26:08 households and get a pretty complete picture of both

26:10 their financial and non-financial income for every investor in my sample.

26:15 So the investors who are taking large sums of money out

26:18 of the stock market or their savings accounts are much more likely

26:22 to have experienced these large 25 or 50% declines in their wage

26:27 or private business income over the same time period that they're withdrawing.

26:32 And we can even flip the same exercise around and ask, you know,

26:35 when an investor gets hit with a bad shock to their non-financial income,

26:39 for example, maybe there's a local economic downturn in their zip code,

26:43 how do they adjust to that shock?

26:45 And what we see in the data is that they

26:46 mostly adjust by drawing down on their liquid financial assets.

26:51 Their net savings ultimately falls by between 50 to 85 cents

26:55 for each lost dollar of wage and business income that we see.

26:59 And just to take a step back here, I think there's some sense among academics

27:02 and maybe sophisticated investors uh that these retail investors

27:06 outflows during stock market crashes in particular are

27:09 driven by purely behavioral factors like panic and fright.

27:13 But I think what this new evidence suggests is that these losses

27:17 in their non-financial income also play

27:19 an important role in driving these flows.

27:21 And in some sense, this is a deeper

27:23 problem for these investors and their portfolio

27:25 allocation problem than just trying to avoid

27:28 panic selling when the market crashes.

27:31 Yeah, that's a much bigger deal because you can't just say,

27:33 "Well, no, don't sell.

27:34 Don't don't panic.

27:35 It's going to be okay.

27:36 Look at the data." But if people are

27:37 selling because they have to whatever fund expenses,

27:40 it's a totally different conversation.

27:43 Um so why why are households spending

27:47 their financial assets rather than you know holding

27:50 their stocks and being disciplined um and adjusting

27:53 their consumption when they have income losses?

27:56 Exactly.

27:56 It's a great question.

27:57 It's key to the paper and to try to understand

27:59 that I draw on a different survey data set that provides

28:03 a lot of detailed information about what these highincome working

28:07 age households actually spend their money on in a typical year.

28:10 And what we learned from that is

28:11 that a large share of their typical budget goes towards

28:14 forms of spending that would probably be difficult

28:16 to reduce or defer over a relatively short time period.

28:21 So take housing for example.

28:22 For the typical 40-year-old highincome household,

28:25 housing related expenditures are going to account

28:27 for about a third of their annual budget.

28:30 Many of these people live in large expensive

28:32 houses and have large mortgages associated with that.

28:35 and the cost of servicing that debt plus utilities,

28:38 property taxes and maintenance.

28:40 Uh they all add up pretty quickly.

28:42 Apart from their home,

28:43 they also spend a lot of money on other things that may be difficult to cut

28:46 back on within the span of say

28:48 a year without making big changes in their lifestyle.

28:51 Things like health care and insurance premiums,

28:54 child care, school or college tuition for their children.

28:58 All of these types of services that involve some form

29:01 of commitment over some fixed time period account for a large

29:05 share of what they spend their money on rather than

29:08 going out and buying boats and dinners at fancy restaurants.

29:12 So taking a step back again,

29:14 economists like Raj Cheddy and Adam Sidle have thought in the past about

29:19 these sorts of consumption commitments and how

29:22 they might influence an investor's portfolio choice.

29:25 But the point I want to emphasize here is that these consumption commitments

29:28 are really very relevant for households at the top of the income distribution,

29:33 not just lower inome households, uh,

29:35 compared to what people may have thought previously.

29:38 I think this ultimately helps to explain why

29:40 these households have to dip into their financial

29:42 assets so much when their income from their job or private business dries up.

29:47 Man, so this really relates like a household's fixed expenses

29:52 back to their ability to take risk in the stock market.

29:57 Absolutely.

29:57 Yeah.

29:58 The the it's almost like a form of leverage you can think of.

30:01 Uh a lot of these commitments don't show up,

30:04 you know, explicitly as debt on the balance sheet.

30:07 Uh but if you have to pay say insurance premiums or child care

30:11 over the course of a year and it's maybe difficult to cut back

30:14 on those in a financial emergency uh then that's almost another form of leverage

30:19 that these households are taking on even

30:20 if it isn't an explicit debt commitment.

30:24 What types of households seem to be

30:26 particularly vulnerable to vulnerable or resilient

30:30 to the selling your portfolio in bad times effect that you documented?

30:35 Right.

30:35 So I think there's ultimately three main things

30:37 that broadly determine how vulnerable investors are to this risk.

30:41 Uh the income risk that they face, the level of their liquid wealth,

30:44 and their ability to cut back on that spending quickly if needed.

30:48 So first, for your income risk,

30:50 if you're in a job or profession where you're more likely to experience a large

30:54 drop in your income when the economy or the stock market are doing poorly,

30:59 you're more likely to end up having to liquidate your portfolio in bad times.

31:03 Now, this risk is particularly high

31:05 for private business owners and for W2 employees.

31:09 It's going to be very high for workers

31:10 that are in cyclical industries like finance, tech,

31:13 or consulting and within a given industry or firm

31:16 for individuals higher up in the earnings distribution

31:20 that may have more volatile bonus pay or stock-based

31:23 compensation that tends to fall at these times.

31:26 But at the same time, there are many high-income jobs or professions

31:29 that are not very exposed to the business cycle.

31:32 I think perhaps the best example

31:34 of this is tenured professors at business schools.

31:36 And many of the ones that I've talked with over the past

31:38 few months tend to be quite aggressive in their stock market investments.

31:42 So the second factor is going to be the amount of liquid

31:45 assets that you have relative to your typical income and spending.

31:48 And that's going to determine the share

31:50 of those assets that you'd have to exhaust if

31:52 you were to draw down on them to cover

31:54 your typical expenditures for some period of time.

31:57 If you don't have a lot of liquid assets,

31:59 then you're much more likely to deplete almost all of them in that scenario,

32:03 and you don't benefit from the fact that stock returns are high going forward

32:07 if you don't have much left in the stock market in your personal account.

32:11 So, in contrast, the households in my data that have a lot of liquid

32:14 assets do tend to significantly draw down on them after a bad income shock,

32:19 but still have plenty left over after doing so.

32:22 And just finally, again,

32:23 the third factor is your ability to cut that spending quickly if needed.

32:27 How much of your annual budget goes to these sticky expenditures like rent,

32:32 mortgage, utilities, child care, health care,

32:34 and other things that would be difficult to cut back on without

32:37 having to make large and potentially disruptive adjustments to your lifestyle.

32:42 So, here I think there's really two types

32:43 of households that are particularly at risk in these cases.

32:46 uh homeowners that have a large mortgage

32:49 and working parents that have multiple children.

32:52 All of which tends to come with a bunch of expenditures

32:54 that may be difficult to cut back on in a time of need.

32:57 And in contrast, young households with more financial flexibility,

33:01 uh say the, you know, fire, financial independence,

33:04 and early retirement investors,

33:05 uh may be in a much better position to adjust to these shocks if they occur.

33:10 Man, so it sounds like a big a big thing is

33:13 not letting your fixed expenses absorb variable portions of your income.

33:19 Like if you get a lot of your income

33:20 from stockbased compensation or bonuses or whatever,

33:23 letting your regular fixed lifestyle expenses creep

33:26 up to that level of income is kind

33:28 of sounds like a big one of the risks that you're that you're finding here.

33:32 Absolutely.

33:32 I mean it ties in with I think a broader societal discussion about lifestyle

33:36 creep keeping up with the Joneses and the people who step a bit too far

33:41 in this direction who are working in some

33:43 of these risky industries and maybe uh you

33:45 know financing a nice car lease out of their uh bonus pay in a given year.

33:50 They're going to be more at risk of having to make some painful financial

33:53 adjustments when things don't look quite as good

33:56 in their job or their private business.

33:59 Yeah.

33:59 which could make that luxury purchase way

34:02 more expensive if they have to sell sell

34:03 their stock portfolio after it's dropped to keep

34:06 making the payments or whatever it is.

34:08 Exactly.

34:08 And of course that's not going to be on the sticker

34:10 price of the car that they sell to you.

34:12 Yeah.

34:12 How do your empirical findings compare

34:14 to the predictions of common savings and consumption models?

34:18 Right.

34:18 So the financial adjustments that I see

34:20 households making in the data after a shock

34:22 to their income differs quite a bit from what

34:24 typical models of consumption and savings decisions would predict.

34:28 And this ultimately makes a big difference

34:30 in the optimal stock portfolio shares implied by the models.

34:33 But to take a step back here,

34:35 what do these typical models actually predict about households would adjust

34:38 to this kind of shock to their wage or private business income?

34:41 Well, if that shock is expected to be pretty persistent rather than temporary,

34:46 as is the case for a lot of the large

34:48 income shocks that these high-income investors face in the data,

34:52 and if they can easily cut back on their spending at those times,

34:56 then those benchmark models that we work with predict

34:58 that this would actually be the main margin they would adjust on.

35:01 That's to say, if a person loses their job

35:04 and switches into a different one that pays $100,000 less,

35:08 these models essentially predict that they would cut their spending by about 80

35:12 to $100,000 immediately in that year and then in every year going forward.

35:17 And this is essentially what Milton

35:19 Friedman's permanent income hypothesis model predicts

35:22 for how households should adjust their consumption

35:24 following a permanent shock to their income.

35:27 Now, just thinking intuitively for a second,

35:30 that's a pretty big adjustment to make in your annual

35:32 spending within a window of one or two years.

35:35 I think a lot of working age households would struggle to cut that much

35:38 spending over such a short horizon without

35:41 making some major adjustments to their lifestyle,

35:44 such as moving to a different house

35:45 or apartment or finding some alternative child care arrangements.

35:49 And when we ultimately look in surveys to see how

35:52 these households say they would adjust in difficult financial situations like

35:55 this, they don't plan to cut back on their spending or borrow

35:59 money through a credit card or home equity line of credit.

36:02 They plan to tap into their liquid assets at that point.

36:05 And this is consistent with how we

36:07 see households actually adjusting in the data.

36:09 When they face these large and persistent

36:11 shocks to their wage or business income,

36:14 they're primarily adjusting by dipping into their liquid savings to smooth

36:18 out that shock rather than cutting back sharply on their spending.

36:22 Now, how does this ultimately matter

36:24 for the models of portfolio choice we work with?

36:26 Well, most of the models in that literature have

36:29 this traditional model of consumption and savings behavior at their core.

36:33 And what this means is that they

36:35 ultimately understate the liquidity needs of these households.

36:38 That is if the stock market crashes today and someone loses their job,

36:42 these models essentially predict that they should avoid

36:44 dipping into their liquid sav their uh liquid stocks,

36:47 their bonds, their savings deposits and instead they should

36:50 cut their spending aggressively leaving their financial assets mostly untouched.

36:55 In practice, we think this is going to be difficult for many of them to do.

36:58 And so the challenge is to then rethink their consumption, savings,

37:02 and portfolio choice decisions in a model

37:05 that more realistically captures their liquidity needs.

37:09 Okay, we do have questions about the model,

37:10 but I just want to make sure it's clear for listeners real quick.

37:12 So you you observe in the data that when people have a an income shock

37:16 when their income falls that they're pulling

37:18 out of their financial assets and then you

37:20 kind of cross reference that empirical observation

37:22 with survey data where people are telling you

37:24 or telling the survey that that is exactly what they would do in those cases.

37:29 Exactly.

37:29 And again, that information is coming from these this great survey,

37:33 the survey of consumer finances run by the Federal Reserve where again they go

37:37 in and ask households if you were

37:39 to face one of these hypothetical financial emergencies,

37:42 how would you adjust through your spending,

37:44 through your uh drawing down in your savings,

37:47 through borrowing and this is I think bringing in the rich set

37:51 of data that we have to understand what we see in other data sets

37:55 that don't have as much granularity and then ultimately try and map that back

37:59 to the models we work with and bring them closer into line with reality.

38:03 Yeah.

38:03 Super interesting.

38:04 So, it sounds like we need a better model.

38:06 We do.

38:06 Yeah.

38:07 And that's a big task of the second part of the paper.

38:10 How did you set up your life cycle portfolio choice model

38:14 to study optimal stock allocations in light of your empirical findings?

38:20 Right.

38:20 So I work in the essentially second part of the paper

38:23 with one of these life cycle portfolio choice models that follows

38:27 households from age 25 when they enter the labor market through

38:31 age 60 when they retire and then through their retirement period.

38:35 And at each point in the life cycle they're

38:37 going to decide how much to spend and consume,

38:39 how much to save, and how much of those savings

38:42 to allocate to risky stocks versus safe risk-free assets.

38:46 There are really three key features of the model

38:48 each of which has been studied in several previous papers

38:51 in the literature but not really combined in the unified

38:54 way that matters a lot in the model I work with.

38:57 So the first is that expected stock

38:59 returns and investment opportunities vary over time.

39:03 Just like in Robert Mertton,

39:04 a guest of the podcast's uh intertemporal capital asset pricing model.

39:09 That is the model can capture the salient pattern

39:12 in the data that stocks occasionally suffer from these large price crashes

39:16 that are followed by swift recoveries and high returns for investors

39:20 who stay in the market or even buy the dip.

39:23 And this is what makes stocks look safe

39:24 for long-term investors despite their volatile short-term returns.

39:28 A point that's, you know, of course emphasized in Jeremy Seagull's book,

39:31 but also in more for uh in some of these portfolio

39:35 choice models in the literature uh studied by people like John Campbell,

39:38 Luis Visera, Jessica Wter, Nicholas Barbaris and many others.

39:43 So that's the first feature of the model.

39:45 The second is labor income risk and its relationship with stock returns.

39:50 So shocks to households earnings are going

39:53 to be drawn from a fat tailed distribution.

39:56 So that most households in a given year experience

39:58 relatively little change in their long-term earnings prospects while

40:02 a small subset of these unlucky investors experience really

40:05 large negative shocks capturing events like job loss or transition.

40:10 Now in the model, a large crash in the stock

40:12 market is going to directly shift the left tail

40:16 of that earnings growth distribution which is going to lead

40:19 to only a modest decline in average or median earnings.

40:22 Uh but a really substantial change

40:24 in the probability of these disastrous labor market outcomes.

40:28 And all the parameters there are going

40:29 to be disciplined by measures of labor income

40:31 risk computed from this administrative income data

40:34 and its correlation with stock returns over time.

40:36 And modeling labor income in this particular way is ultimately going to make

40:40 human capital look much more like a risky stock than a safe bond.

40:45 And just finally, the third key

40:46 feature is frictions in adjusting household consumption.

40:49 I'm going to assume they have to pay an extra cost

40:52 to reduce their spending below its level from the previous year.

40:55 And this cost is much higher for very large cuts in their spending.

40:59 This is ultimately going to force them to actually

41:01 draw down on their savings following these large

41:04 shocks to their income instead of simply cutting

41:06 their spending aggressively and leaving their financial assets untouched.

41:10 And the magnitude of those costs is

41:12 ultimately going to be disciplined by that size

41:15 of that savings response to these income shocks that I estimate in the tax data.

41:19 And these costs are going to be crucial for realistically

41:22 capturing those households liquidity needs during a stock market crash.

41:27 Okay, so everyone is on the edge of their seat

41:30 for the to waiting for the answer to this question.

41:33 What what does the model say about

41:34 the optimal equity share for a working age household?

41:38 So the model says that the optimal share

41:40 of liquid wealth invested in stocks is relatively low

41:43 for the average working age household ranging from 10

41:46 to 40% depending on factors like their age, risk aversion and the uh magnitude

41:51 of these frictions in adjusting their spending.

41:54 This optimal stock share also increases strongly with the household's age

41:58 and with the amount of liquid wealth that it owns relative to its labor income.

42:02 Now, these stock shares of liquid wealth

42:04 and their patterns over the life cycle fall

42:06 in the ballpark range of values that we

42:08 see most of these households investing in the data.

42:11 However, it's a lot lower than the optimal shares implied

42:14 by many life cycle portfolio choice models in the literature,

42:17 particularly for those young households.

42:19 So what is it then that makes stocks so risky for these young investors

42:23 in the model despite the fact that returns

42:25 are relatively safe over their long potential investment horizon?

42:29 It's ultimately these liquidity needs that the they

42:32 face during these stock market crashes.

42:34 So in the model and in the data, many of them experience these large

42:38 declines in their income during these crashes.

42:42 And if it's difficult for them to cut their spending in response,

42:45 they're going to have to draw down other liquid assets at that time.

42:48 And if you have those liquid assets fully invested in stocks,

42:51 then they're going to have to liquidate their holdings exactly when

42:54 prices are low but expected to bounce back in the near future.

42:57 They won't actually benefit from that rebound if they have

43:00 little financial wealth left after smoothing out from this shock.

43:04 So as a result, the optimal stock share

43:06 of liquid wealth is quite low in the model,

43:08 particularly for these young households without much liquid financial wealth.

43:12 It's that combination of their risky income

43:14 and inflexible spending that forces them to draw

43:17 down on their assets in bad times

43:19 and makes stocks ultimately look less attractive.

43:23 So, let let me give you a very basic interpretation of your findings

43:26 and and you can tell me if it's if it's right.

43:28 Um, it sounds like you've basically found in this research

43:31 that people should have an emergency fund, right?

43:34 Because that's that's going to be liquid wealth and it's

43:36 going to be decreasing as their overall liquid wealth increases.

43:41 Sorry.

43:41 like a fixed uh period of expenses in an emergency fund.

43:45 Like you should have whatever 12 months is 12 months of expenses in an emergency

43:49 fund which would line up with your finding that as liquid wealth increases

43:53 your share of equities can increase but as you've got when you have low

43:55 liquid wealth the share of equities is really low the optimal share of equities.

43:59 So it's really like you can invest in stocks

44:01 with your liquid wealth but you should probably have an emergency fund.

44:05 Absolutely.

44:05 And personally, when I think about how this evidence changes my own investing,

44:09 that's exactly the way I frame it.

44:11 Or when talking with some of my friends working

44:13 in some of these risky industries like finance or portfolio management,

44:17 uh how many months of your difficult to adjust expenditures do

44:21 you have saved up in these safe liquid assets uh rather

44:24 than some of these riskier assets that may decline a lot

44:27 in value exactly when you need to draw down on them.

44:30 So you can exactly think of this as a fixed emergency

44:33 fund covering say 12 months of your difficult to adjust expenditures.

44:38 Uh that's going to represent a smaller share

44:40 of your overall liquid wealth as you start to accumulate more

44:43 of it and can then think about aggressing uh investing

44:46 that more aggressively in some of these riskier assets like stocks.

44:52 How do high consumption adjustment costs affect optimal savings behavior?

44:57 Right?

44:57 Right.

44:57 So in the model I think you know as we've sort of talked

44:59 about we think is going on the uh in the data and in reality.

45:03 So those high consumption adjustment costs are going to give households

45:06 a very strong incentive to save

45:08 more particularly in these safe risk-free assets.

45:12 And the reason is that they're going

45:13 to anticipate having to draw down on those savings

45:15 if they face a large income loss

45:17 and they can't cut their spending easily in response.

45:20 And actually in the current calibration of the model that I show in the paper,

45:24 this actually leads households to accumulate a bit more

45:26 liquid wealth than what we see in the data.

45:29 And you know, this is a paper that I'm always working to improve.

45:32 It's partly, I think, because of strong assumptions I make about

45:36 the illquidity of their housing and their retirement accounts.

45:39 I'm currently working with some, you know,

45:40 high-powered Nvidia GPUs to enrich the model along these dimensions.

45:45 But despite that high level of liquid wealth that they accumulate,

45:49 the optimal share of it that they invest in stocks is still quite low.

45:54 So if you have high fixed expenses, you need a bigger emergency fund basically.

45:59 Precisely.

45:59 And on the other hand,

46:01 if you can easily cut back on that spending and bad times,

46:04 you can feel free to invest a lot more

46:06 of those liquid assets in some of these riskier forms,

46:09 provided that you have the market discipline

46:11 to avoid selling them if there's a big crash.

46:14 Man, so interesting.

46:14 I I I originally thought about this paper as well, it still is.

46:19 Maybe this won't make any sense, but I'll say it.

46:21 Um I thought about it as like an asset allocation paper, which it still is.

46:25 Um but it's really it's really about emergency funds.

46:30 Exactly.

46:30 What what form of asset allocation, right?

46:34 Yeah.

46:34 You can think of this almost as really

46:35 taking like a risk management or cash flow management

46:38 sort of view of the household balance sheet

46:40 and the different sources of risk that they face.

46:42 And you know, one way I try to frame it is that, you know,

46:45 they've got sources of funds uh that are pretty

46:47 volatile and uncertain uh especially for these highincome wealthy households.

46:53 And they've got uses of funds that are pretty sticky uh even

46:56 beyond their actual fixed debt commitments and payments that they have to make.

47:00 And I think that's a very useful way

47:02 to think about again household risk management uh

47:05 which is ultimately an asset allocation problem between

47:08 these safe and risky or liquid and illquid assets.

47:12 Yeah.

47:12 Yeah.

47:12 Super interesting.

47:14 Uh, what does the model say about,

47:15 you've mentioned this a couple times, but I'm still gonna ask it.

47:18 What does the model say about the difference in optimal

47:20 equity shares between a working age household and a retired household?

47:24 So the model is ultimately going to say that optimal stock shares

47:28 that they should invest in their liquid wealth should actually be lower

47:31 for most working age households compared to most retired households which I

47:36 think would be a bit of a puzzling result from a lot

47:38 of the portfolio choice models we've seen uh in the prior

47:41 literature and for recent retirees that optimal stock share in the model

47:45 is generally going to be above 50% with the precise value depending

47:49 on some factors like their risk aversion in their remaining investment horizon.

47:53 Now that value I think should be interpreted a little

47:55 bit carefully in my current set of results because

47:57 of exactly how I model the income flow and asset

48:00 allocation in their retirement accounts or don't model it actually.

48:04 Uh but the broad point is that within the model stocks ultimately

48:08 look much less risky for retirees than for these young working age investors.

48:16 Wow.

48:18 So what explains this counterintuitive difference?

48:22 Right.

48:22 Right.

48:22 So, I think this result can really seem counterintuitive

48:24 from the work that we've seen previously given that we're used

48:28 to thinking of stocks as being particularly attractive for young investors

48:31 who have long potential investment horizons and safe bond-like human capital.

48:37 And so, I think there's really two major

48:39 places where this intuition breaks down in my model.

48:42 First, when we look at the administrative data,

48:44 investors labor income actually looks quite risky.

48:47 And the model captures that observed relationship between stock

48:50 returns and these negative tail outcomes in the labor

48:53 market and ultimately makes households human capital look much

48:57 more like a risky stock than a safe bond.

49:00 If you think of retirees on the other hand,

49:02 they have much more stable income sources

49:04 in the form of social security and pensions.

49:07 Uh second, I think the other important factor is that while households have

49:10 long potential investment horizons spanning all

49:14 the way until they retire and beyond

49:16 that, their effective investment horizon may end up being quite a bit shorter if

49:20 they experience a shock that leads them

49:22 to draw down on their assets before then.

49:25 Now, early on in their career,

49:27 most of their usual spending is financed exactly by this risky labor income.

49:31 As we discussed previously,

49:33 if it's difficult for them to cut back on that spending when that income falls,

49:37 they're going to have to draw down on their assets,

49:39 particularly at times when the stock market is doing poorly.

49:42 And as a result, their effective investment horizon shrinks and becomes

49:46 much shorter at exactly the most inopportune times as an investor.

49:51 And if you're a retiree instead who consumes only

49:53 a small fraction of your financial wealth every year,

49:56 those same-sized withdrawals wouldn't really jeopardize

49:59 your financial situation and put you

50:01 in a scenario where you would exhaust most of that liquid wealth.

50:04 This is what's ultimately going to allow these older

50:06 investors to ride out these stock market crashes more effectively.

50:10 So ultimately, I don't believe we should think of most young investors,

50:15 particularly at the top of the income distribution,

50:17 as having the safe bond-like human

50:19 capital and long effective investment horizons.

50:22 You should think of them as having

50:24 risky stock-like human capital and potential liquidity

50:27 needs that may lead them to draw down on their assets well before they retire.

50:33 You kind of did this early on, but c can you walk us through a hypothetical

50:38 hypothetical scenario that illustrates why the optimal equity

50:41 share is so low for the working age households?

50:44 Yeah, definitely.

50:44 So, I think it's helpful here to step into the shoes

50:47 say of a hypothetical household around the global financial crisis in 2008.

50:51 And based on some of the statistics I talked about previously,

50:55 I'll assume they start the year in 2008 with around $300,000

50:59 in annual income from their jobs and $200,000 in liquid financial assets.

51:05 Now, suppose October 2008 comes and at the height of the financial crisis,

51:10 one of the two people in the household loses

51:11 their job and their total income falls by $150,000.

51:16 Now, based on my estimates from the tax data,

51:18 I'm going to assume that they draw down 2/3

51:20 of that amount from their liquid assets over the following months.

51:24 So, 2/3 of $150,000, that's $100,000.

51:28 Now, if we suppose that their liquid assets start

51:31 out fully invested in stocks at the start of 2008,

51:35 then by the time they actually start

51:36 drawing down on their stock holdings in October,

51:39 they will have declined by almost a third in value.

51:42 And at that point, that $100,000 outflow represents a much larger share

51:46 of their remaining liquid wealth than it did just a few months prior.

51:50 And if we follow them through to the end of 2010,

51:53 when a buy and hold investor would have fully recovered their initial wealth,

51:56 our hypothetical household has only about $50,000 left in that case.

52:01 In addition to the $100,000 that they

52:04 would actually withdraw from their account,

52:05 there's effectively an extra $50,000 cost they pay from the high returns they

52:10 forego after having to liquidate their stock

52:12 holdings exactly when prices are low.

52:14 Now, if they were to invest only half

52:17 of those funds in stocks over the same time period,

52:20 that extra loss would be only half as large,

52:22 amounting to only about $25,000 in that case.

52:26 And if they keep them fully invested and say safe bank deposits,

52:29 then they're only out that $100,000 outflow and they

52:32 could potentially even buy stocks when prices are low.

52:35 I think the key idea here is that even

52:37 if stock prices recover from some of these big crashes,

52:41 you don't personally benefit from that as an investor if you end

52:44 up having to exhaust most of your liquid financial wealth during the crash.

52:49 And that's what's ultimately makes it risky to invest

52:51 a large share of that wealth in stocks in my model.

52:56 So let's put a finer point on this.

52:58 Which parameters in your model have the biggest impact on the headline

53:02 result of the lower optimal equity share for the working age households?

53:08 Definitely.

53:08 So the most important parameters that matter for those working

53:11 age households portfolio decisions are going to be ultimately

53:15 the relationship between stock returns and their labor income risk

53:19 and how costly it is for investors to cut their spending.

53:22 So first on the labor income risk side

53:25 in my baseline calibration I assume that a large crash

53:28 in the stock market significantly increases the probability that a working

53:33 age investor experiences a large persistent negative income shock.

53:38 If we were to weaken that channel

53:39 by making labor income shocks either independent from stock

53:42 returns or even normally distributed uh without

53:45 the fat tails that we see in the data, then the optimal equity share is much

53:49 higher in those cases because human capital

53:52 ultimately looks more like a safe bond

53:54 under those assumptions than a risky stock.

53:57 And for a subset of investors who are employed in professions

54:01 that are relatively safe for insulated from the business cycle,

54:05 this may be the appropriate case for them to consider.

54:08 But for the typical highincome working age investor,

54:10 they do face significant labor income risk and that should

54:14 play a big role in determining their portfolio allocation decisions.

54:19 Yes.

54:19 Now super sorry.

54:21 No, just re really interesting.

54:22 It's it's cool to hear you go through the the parameters.

54:24 I think that's super helpful for for listeners.

54:27 Exactly.

54:27 And you know, of course,

54:28 the other one that's going to matter here is those consumption adjustment costs.

54:31 It just determines how much you actually have to draw down

54:33 on those liquid assets when you face one of these big income shocks.

54:37 And if it's easy to cut your spending, you can leave those assets untouched.

54:40 You can weather the storm, ride out these price crashes,

54:43 and act like that ideal Jeremy Seagull long-term investor who

54:47 doesn't have to sell their stock holdings at those times.

54:50 H.

54:51 So, we've had a couple guests,

54:54 James Choy most recently and Scott Cedarberg a while ago now,

54:57 whose life cycle portfolio choice research generally

55:00 suggests 100% equity portfolios for working age households.

55:04 And they both have I mean their papers are really comprehensive and they

55:08 thought about modeling labor income in different

55:10 ways and all that kind of stuff.

55:12 What would be the key differences driving

55:13 the differing advice between your model and their models?

55:18 Right.

55:18 So both of them are obviously really fantastic researchers and those are

55:23 great papers that I personally learned a lot from reading.

55:26 But I think the primary difference at the end of the day

55:28 between those models and mine is how we capture the relationship

55:32 between stock returns and labor income risk and whether this makes

55:36 human capital look more like a risky stock or a safe bond.

55:39 So in my model again a large crash in the stock

55:42 market has a very direct effect on the probability

55:45 that a given household experiences a really large decline

55:48 in their earnings consistent with what we observe in administrative income data.

55:52 And here I should note I'm building off of similar models studied by one

55:56 of my adviserss Lawrence Schmidt at MIT

55:59 and Silvan Katherine at Wharton who showed

56:01 that modeling labor income in this way makes human capital again look much more

56:05 like a risky stock and can even help to explain the equity premium puzzle.

56:10 That is why stocks are so risky and earn

56:12 such high average returns in the first place.

56:14 Now, importantly, labor income still looks very risky in these models,

56:18 even though the average household's earnings don't

56:20 fall dramatically when the stock market crashes.

56:23 Most households experience a small change in their income,

56:26 but an unlucky subset of them lose a large share of their initial earnings.

56:30 And I should also note international stocks don't really look like

56:33 a great hedge against this form of tail risk in the labor

56:36 market because they also tend to crash in years when

56:39 US stocks do as we've seen over the last 25 years.

56:43 And finally, I just want to note I think another important difference between

56:46 my model and many previous ones studied

56:48 the literature are these consumption adjustment costs because

56:51 these models uh have been uh used in past work often assume that households

56:57 can pretty easily make these large immediate

56:59 cuts to their spending if they have to.

57:01 And it turns out to be the main way

57:03 that they adjust following these persistent shocks to their income.

57:06 It's at odds with what I find in the tax data

57:09 and the consumption adjustment costs I put in the model help

57:12 to generate a much more realistic savings response and liquidity needs

57:16 that make stocks look very risky for these working age investors.

57:22 So I'm sure this next question is shared by many listeners.

57:25 Is there a set of parameter specifications where your model does

57:29 find an optimal 100% equity share for a working age household?

57:35 Yes, definitely.

57:35 And we can kind of push on I would say five

57:38 different factors in the model that are going to nudge households

57:41 towards an optimal 100% equity share even when they are still

57:45 saving for retirement and face this income risk in the labor market.

57:50 Uh so those factors are essentially if they have low risk aversion,

57:54 low labor income risk, low consumption adjustment costs,

57:58 high liquid wealth relative to their income and also you know because

58:02 the equity premium is varying a lot over time in my model.

58:05 If it's a time period at that uh you

58:08 know in that given year where the equity premium

58:10 is currently very high say you know 10 or even

58:13 15% instead of the 5% baseline uh normal times.

58:18 So all of these factors are either going to decrease the risk that the household

58:22 will have to liquidate most of its assets during a stock market crash

58:27 or in the case of you know risk aversion or a high equity premium

58:31 makes them more willing to bear that risk

58:33 in pursuit of those high average returns.

58:36 But, you know, I think for the typical highincome household that does

58:39 have risky labor income and moderate

58:41 financial wealth and more conventional risk preferences,

58:45 the optimal stock share is going to be much lower

58:47 than 100% consistent with what we observe in the data.

58:50 I think I would encourage listeners to maybe

58:52 figure out which of those five buckets they

58:54 fall in if they're really tempted to select

58:57 a 100% equity share in their liquid portfolio.

59:01 Yeah, people love the 100% equity advice when

59:04 when a paper comes out that supports it.

59:06 Well, I don't know.

59:07 I don't know if it's because it's just so straightforward or I don't know.

59:10 People just like stocks.

59:11 I guess at least they do.

59:13 Maybe that's because the returns have been so good for for a while.

59:16 Maybe if we have a prolonged bare market, people will change their minds.

59:19 Yeah, exactly.

59:20 Well, let's hope not.

59:21 At least for my sake.

59:22 But uh what kind of comments are you getting when you present this paper?

59:25 Like you're kind of going against the grain

59:27 of a lot of the the past research in the space.

59:29 What are you hearing from other researchers when you present?

59:33 Right.

59:33 So, I should say, you know,

59:34 everyone I've talked to, even the people kind of working

59:36 on similar topics who have conclusions that differ a bit from mine,

59:40 they've all been very receptive, very uh, you know,

59:43 I think very thoughtful and scholarly

59:46 in terms of their engagement with the work.

59:48 And so, I'm really delighted by, you know,

59:50 how great the people I've talked with have been,

59:52 uh, even when I say things that may go against uh,

59:55 some of the things they claim.

59:57 uh but a lot of the people I present this paper to are academics

1:00:00 in particular early career ones with a mortgage

1:00:02 and young children and so this they find

1:00:05 this idea of consumption commitments very personally relatable

1:00:08 because they ultimately spend a lot on housing

1:00:10 and child care and many of these high cost of living areas around the US.

1:00:15 Now at the same time many of those people are also

1:00:17 tenured professors with stable predictable income

1:00:20 who have that bond-like human capital.

1:00:23 So they don't personally find that part of the paper very relatable.

1:00:26 But many of them do have friends, family,

1:00:29 or even former PhD students who work in risky industries like finance or tech.

1:00:34 And some of those people did lose or switch their jobs

1:00:36 during these large stock market crashes over the last 25 years.

1:00:40 And some of those people did end up having

1:00:42 to sell their stocks at the bottom of the market.

1:00:44 Uh so these anecdotes from other people's past experiences do line up with what

1:00:48 I document in the paper and I think helps to convince them of my results.

1:00:54 So how's your work on this affected

1:00:56 your own decisions around your own asset allocation?

1:01:00 Right.

1:01:01 So I think there's one very practical way that's been related to my own job

1:01:04 search over the last 12 months which is that we know that from prior research

1:01:09 using this administrative data that people searching

1:01:12 for a new job are particularly vulnerable

1:01:15 to the state of the business cycle because

1:01:17 these job openings tend to be very volatile

1:01:20 um compared to say layoffs for existing hires.

1:01:23 And so because of that, I personally held off on moving some of my own savings

1:01:28 into stocks over the past year while I

1:01:30 was waiting for that particular uncertainty to resolve.

1:01:33 But you know, looking forward,

1:01:35 I now think a lot more, as I mentioned previously,

1:01:37 about how much of a safe liquid asset buffer I

1:01:41 should have given the risks I face in my income

1:01:43 and some of those spending commitments uh that are starting

1:01:46 to add up to a larger share of my own annual budget.

1:01:49 And I should say I'm definitely more hesitant now

1:01:51 to sign up for some of these large expenditure

1:01:53 commitments like a big apartment lease or a mortgage

1:01:56 than I was before I started working on this paper.

1:02:00 You mentioned listeners wanting to consider which of the five

1:02:02 buckets that you talked about a minute ago, which one they fall in.

1:02:06 What would you say to the working age

1:02:07 listeners who do currently have 100% stock portfolios?

1:02:12 Right.

1:02:12 So I think the usual caveat first applies here which is

1:02:15 that I am not a financial adviser and this is not investment advice.

1:02:19 But that being said I would you know first

1:02:21 remind them that these stock portfolio shares that we're talking

1:02:24 about are measured as a fraction of their total

1:02:27 liquid financial assets not just what's in their brokerage account.

1:02:30 So almost all of them surely have

1:02:32 some bank deposits or other safe liquid assets.

1:02:35 And I think the key question is whether those assets

1:02:38 are enough to get them through some potentially difficult financial situations.

1:02:42 I think it's useful here to take a riskmanagement perspective.

1:02:45 Again, similar to a corporation thinking about its own cash flow management.

1:02:49 So think about your sources of funds that you have.

1:02:52 In particular, what risks do you face in your job,

1:02:54 your labor income, maybe in a private business you own?

1:02:57 And are these risks likely to be correlated with the overall stock

1:03:01 market in the real economy or are they mostly independent of that?

1:03:05 On the other hand, think about your uses of funds that you have.

1:03:09 What expenditures do you have that would be difficult to cut

1:03:12 back on within the span of a year or two?

1:03:14 And how many months of those expenditures

1:03:16 could you fund with just your existing safe

1:03:18 liquid asset holdings before you would have

1:03:21 to sell stocks or withdraw from your retirement accounts.

1:03:24 And you know finally just one thing I want to say I don't focus as much on asset

1:03:27 allocation within retirement accounts because of some limitations

1:03:31 on what we can see in the tax data.

1:03:33 However these investors liquidity needs actually

1:03:36 may flip some of the conventional way that we think about tax

1:03:39 optimization for these highincome working age investors.

1:03:43 Many of them are told to think

1:03:45 about holding their fixed income assets like bonds

1:03:47 in their retirement accounts given the high ordinary income

1:03:50 tax rates that they face on their interest income.

1:03:53 But if you have to draw down on your financial assets in a time of need,

1:03:57 it may make sense to hold more of those safe fixed income assets outside

1:04:01 of retirement accounts as a first or second buffer against a large income shock.

1:04:06 And any stocks that you hope to avoid selling at those times,

1:04:09 you could instead hold in retirement accounts,

1:04:12 even if that's going to give up some of the preferential capital gains

1:04:15 tax treatment that these assets enjoy

1:04:18 when they're in your taxable brokerage accounts.

1:04:20 I think this is obviously a really important set of issues that are

1:04:23 going to be a main focus of some of my work going forward.

1:04:27 Oh man, I was expecting your asset allocation commentary.

1:04:30 I was not expecting you to also flip asset location on its head.

1:04:34 That was that was awesome.

1:04:36 Exactly.

1:04:36 Yeah.

1:04:37 Think carefully about that given the liquidity

1:04:39 needs uh that these investors face.

1:04:43 Final question, Patrick.

1:04:44 How do you define success in your life?

1:04:47 So, this is a tough one because I'm just starting my academic career.

1:04:51 So, I think I'm probably the least accomplished

1:04:53 person to appear on your podcast and its history.

1:04:56 Uh, so this is all mostly based

1:04:57 on my aspirations and not my actual achievements so far.

1:05:00 But, you know, I hope that this research

1:05:02 and other projects that I do can provide individuals, firms,

1:05:06 and policy makers with some useful information that they

1:05:09 can take to go and make better financial decisions.

1:05:12 I think a lot of these high-income households in particular work in, you know,

1:05:16 important high growth industries or maybe

1:05:18 entrepreneurs creating jobs through their private business.

1:05:21 And so I hope that this evidence can potentially help them to better

1:05:24 prepare and respond to the risks that they face in that work.

1:05:28 Uh on the other hand, in my future teaching,

1:05:29 I hope to similarly provide the students I work

1:05:32 with uh with the tools to make sound financial

1:05:34 decisions in their personal lives or their work

1:05:36 in the private sector or maybe even in public service.

1:05:40 But, you know, as many people on the podcast say, you know,

1:05:43 I think the richest parts of my life are definitely outside of work.

1:05:46 Uh, there I'm very fortunate to have

1:05:47 a wonderful family and a wife who's been here

1:05:50 with me on this academic journey for over

1:05:52 a decade since we first met in college.

1:05:54 And that's obviously the single biggest part of how I define success in my life.

1:06:01 Awesome.

1:06:01 Great answer.

1:06:02 Yep.

1:06:02 Great answer.

1:06:03 Thanks a lot, Patrick.

1:06:04 This is a great conversation.

1:06:05 We really appreciate you coming on the podcast.

1:06:07 Thank you, Ben.

1:06:08 Thank you, Cameron.

1:06:09 been great to finally talk with both of you and keep up the great work

1:06:12 inspiring lots of PhD students to get

1:06:15 the ideas for their dissertations with this excellent podcast.

1:06:18 Thank you.

1:06:19 We'll do.

1:06:24 Hey everyone, it's producer Matt.

1:06:26 Thank you so much for tuning in to this week's episode.

1:06:28 Before we sign off, here's the disclaimer you've been waiting for.

1:06:32 Portfolio management and brokerage services in Canada are

1:06:35 offered exclusively by PWL Capital which is regulated

1:06:38 by the Canadian Investment Regulatory Organization and is

1:06:42 a member of the Canadian Investor Protection Fund.

1:06:45 Investment advisory services in the United States of America

1:06:48 are offered exclusively by One Digital Investment Advisors LLC.

1:06:52 One Digital and PWL Capital are affiliated entities

1:06:55 and they mostly get on really well with each other.

1:06:58 However, each company has financial responsibility

1:07:01 for only its own products and services.

1:07:04 Nothing herein constitutes an offer or solicitation to buy or sell any security.

1:07:10 Occasionally, we tell you not to buy crappy investments in the first place.

1:07:13 But that's not the same thing as telling you to sell them.

1:07:16 This communication is distributed for informationational purposes only.

1:07:20 The information contained herein has been

1:07:22 derived from sources believed to be truthy, but not necessarily accurate.

1:07:28 We really do try, but we can't make any guarantees.

1:07:31 Even if nothing we say is fundamentally wrong, it might not be the whole story.

1:07:36 Furthermore, nothing herein should be construed as investment,

1:07:40 tax, or legal advice.

1:07:42 Even though we call the podcast your weekly

1:07:44 reality check on sensible investing and financial decision-making,

1:07:48 you shouldn't rely on us when making actual decisions, only hypothetical ones.

1:07:53 Different types of investments and investment strategies have varying

1:07:56 degrees of risk and are not suitable for all investors.

1:07:59 You should consult with a professional adviser to see how

1:08:01 the information contained herein may apply to your individual circumstances.

1:08:06 It might not apply to all.

1:08:07 Honestly, you can probably ignore most of it.

1:08:10 All market indices discussed are unmanaged,

1:08:13 do not incur management fees, and cannot be invested indirectly,

1:08:17 which is a shame because it would be awesome if you could.

1:08:21 All investing involves risk of loss, including loss of money,

1:08:25 loss of sleep, loss of hair, and loss of reputation.

1:08:29 Nothing herein should be construed as a guarantee

1:08:31 of any specific outcome or profit.

1:08:34 Past performance is not indicative of or a guarantee of future results.

1:08:38 If it were, it would be much easier to be a Leafs fan.

1:08:42 All statements and opinions presented herein are those of the individual hosts

1:08:47 and/or guests and are current only

1:08:50 as of this communication's original publication date.

1:08:52 No one should be surprised if they have all since recanted.

1:08:56 Neither one digital nor PWL Capital has any obligation to provide

1:09:00 revised statements and/or opinions in the event of change circumstances.

1:09:05 See you next time.