Hacker News new | past | comments | ask | show | jobs | submit login
Do Stocks Outperform Treasury Bills? (2018) (ssrn.com)
59 points by ZeljkoS 26 days ago | hide | past | web | favorite | 61 comments

This is the second reason why indexing outperforms actively managed funds over the long term (The first reason is the fees).

Price weighed index fund rebalances regularly and never misses out those best-performing four percent. Actively managed fund manager might do a good job valuing companies and might catch some success stories, but it's unlikely that they catch them all consistently.

When companies have now started to go into debt to buyback shares and artificially raise their EPS, who are they targeting? The people who are actively evaluating what they're buying, or those asleep at the wheel buying index funds?

Index funds might have been a good investment in the past, before wallstreet has figured out how to exploit it like they're doing today.

From what I understand this exact same argument is reiterated every decade. "It worked in the past but today is different" - invariably ending up with a suggestion to give up increasingly large amounts of your profits to someone that "manages your investments" for you. I would do it if the fees were fixed, once someone asks for a "percent" I call BS. There is no difference between managing 100K and 1M yet you pay 10 times more.

I'll stick with Vanguard.

It worked in the past but today is different is a vast oversimplification of my argument, and implying I'm either acting in bad faith or hoodwinked by those that are in order to pay off some investment manager is just an attempt to destroy the messenger.

Good luck with Vanguard.

My point is that while Wall Street may have figured out index funds better than it used to - it is nowhere near as "figured" out as that of manipulating individual stocks.

As the paper points out - holding individual stocks is far riskier and inefficient than that of holding an index stock. Hence my point that while you may have different motivations for writing your piece - in the end, you are advocating for a worse potential outcome for most people.

Wall Street would love nothing better than have many people play the stockmarket and try to outsmart the index funds.

Vanguard charges 10x for investing 10x more.

Good point - they do indeed!

What I was really referring to is someone else doing nothing more than buying a Vanguard ETF and then adding their fees (perhaps an order of magnitude higher) on top of those fees - just to "manage" your money.

Huh? Index fund buyers don't read EPS, so the frauds are targeting the dumb active investors.

It's generally agreed that index funds are going to eventually destroy the stock market because they are the free market reinvention of a central planning economy relying on fee riding on the valuation work of a shrinking number of active investors, but that's a different problem.

> Index fund buyers don't read EPS, so the frauds are targeting the dumb active investors.

I'm not saying the buyer's do. The selection criteria of the index fund may include it.

> It's generally agreed that index funds are going to eventually destroy the stock market

Or, like I've been saying, companies will find says to adjust their figures to get selected for index funds. And then eventually implode, due to the cost of doing so, which in most cases is quite a lot, taking down the investors along with them.

Generally agreed?

By whom? Money managers? But of course!

Guess what - if you think index funds are not optimal, deploy your own optimal strategy then profit bigly.

> When companies have now started to go into debt to buyback shares and artificially raise their EPS, who are they targeting?

Their shareholders, which includes management. Buybacks are essentially dividends in another form. Buybacks have the upside of not being expected to go up indefinitely, and there are no dividend cuts to upset investors.

Why does everyone ignore the second part of my point? Corporations are going into record debt to do this. It is one thing to pay back shareholders by a dividend or buyback or whatever, if the company has extra profits, but it is quite another to borrow money to do it.

That works for now, but as soon as those interest rates tick up...

Money is extremely cheap to borrow, and companies have decided the interest rate risk is lower than the short-term share price gains from borrowing money to do buybacks. Perhaps current management assumes they won't be the ones left holding the bag.

I agree with you, when rates go up a large amount of zombie companies being held up by cheap money are going to go bankrupt, and it will have large ripple effects.

Isn't there only a handful of hedge funds that have posted profits every year of operation?

In my dreams, I create 1024 stock-recommending twitter bots.

If half beat the market in a year and half don’t, i’ll Still have one that goes up every year after 10 years, and sell it for a fortune.

> If half beat the market in a year and half don’t

As per this article, that wouldn't happen: more than half of the bots would fail to beat the market.

This article finds that the vast majority of the positive return is explained by relatively few stocks, leading to a positive skewness. The median random choice has a below-mean outcome.

Right. Regression toward the mean is inevitable.

But I think people appreciate a story about picking the right stock more than the right stocks, which helps that problem.

Please read Paulos's classic book Innumeracy, on its 30th anniversary


Yes but most of us aren't allowed to invest in them.

I don't pretend to know the numbers, but my guess is that Survivorship Bias might account for that to a substantial amount. I just can't imagine that there are people with glass bowls that good to make consistent alpha over the long term.

ALmost all have posted profits, only a very few have posted profits higher than their risk adjusted peers.

" indexing outperforms actively managed funds"

Most index funds are actively managed.

Seeking to replicate the performance of an index by trading a basket of stocks/futures/options is not active management.

For instance, when an institutional investor gives SPDR say, $300,000,000 for a million shares of SPY, the fund goes out at purchases the underlying stocks in proportions that match the composition of the S&P 500, and then issues the SPY shares and holds the underlying and dividend payments in trust. SPDR does not have the option to choose the shares, they purchase a basket that replicates the S&P 500.

This is passive management, there is no one actively picking stocks, and when to buy/sell them.

So who decides what the DJIA and S&P 500 stocks actually are? A random number generator?

I will assume your interest is genuine, here are links to the documents explaining the methodology behind the various S&P[0] and DJ indices[1]. I'm sure you can find documents for /RUT and IWM (Russell 2000), as well as /NQ and QQQ (Nasdaq 100) if you would like to see those as well.



The point, with which I agree, is that it's still active, just much less so. It's a bit pedantic, but assuming ETFs are passive by definition could get people crushed when these themed tickers get outflows.

I'm not claiming all ETFs are passive by definition, or without risk.

ETFs do have additional liquidity risk that mutual funds lack. If liquidity is lacking, it could lead to the bid falling below NAV, potentially quite significantly, and investors who chose to sell at that point in time would be losing additional money.

Mutual fund sales are processed at NAV at the end of the market day when the sale takes place, which ensures you will receive the NAV when you sell.

ETFs allow more flexibility, with the drawback of possibly trading below NAV or above NAV, which can be good or bad depending on if you are buying or selling.

If the price of the ETF falls too far below NAV, big traders will scoop up the ETF shares to redeem for the discounted underlying stocks as an arbitrage play, bringing the ETF's price closer to the NAV.

This isnt really surprising to me. What's the average lifetime of a company, even one that has reached maturity? I would guess much less than 100 years.

I suppose the waters are muddied somewhat by consolidation, in 1927 there were 10s(?) of thousands of car companies, today we're basically down to 20(?). No doubt many went bust, but many also got subsumed.

The fact that 3/7 companies have outperformed Treasuries since 1927 is the surprising thing for me, not that 4/7 havent.

Make it even more extreme. I bet that the Treasuries would also outperform if you take 10'000 year timespan. Simply because: Which company would really survive 10'000 years, huh? So the total return would be 0. And as long as Treasuries>0, that research paper is little more than a fairly obvious statement, at best.

What government survives 10K years?

This is a very valid point. You should include now-defunct companies, as current investors can’t tell whether a company goes bankrupt next year.

The study does include now-defunct companies.

What does the average lifetime of a company have to do with it? I'm not really understanding why you think that is relevant to what the paper is discussing?

The paper isn't comparing "Company A that was in existence from 1926-1944" vs. "Treasury Bills from 1926-2018". It is comparing "Company A from 1926-1944" vs. "Treasury Bills from 1926-1944".

So the average lifetime of a company is irrelevant.

I also don't follow why you think the waters are muddied by consolidation? The paper explicitly discusses how it handles "merger, exchange, or liquidation (CRSP DLSTCD with 2, 3, or 4 as first digit)". If Company B is founded in 1931 and merges with another in 1943 then its lifetime is 1931-1943 and it is compared against Treasury bills over that period, so there's nothing muddied.

Company A started in 1926. It is either still in business, consolidated with other companies, or gone bust.

If the company has gone bust (which I would suggest is most of them) it has probably underperformed treasuries, it is of course possible that dividends could make up for the total loss of capital though.

A company that exists from 1926 - 1944 has seen its value go from 0 to X and back to 0, whereas the Treasuries haven't.

Consolidation muddies the waters because you have potentially 10s of companies into 1. You've highlighted 1 way of accounting for that, there are other legitimate ways of doing it though. I was though thinking more that you can't look at a current list of stocks and a 1926 list of stocks, you aren't necessarily comparing apples to apples.

Followup paper "Do Global Stocks Outperform US Treasury Bills?": https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3415739

Note how this is about GLOBAL stocks, not US Stocks, which would be the better comparison

A little bit OT: did anybody come up with a way to graph two stocks such that you see individual long term performance but also relative performance?

I.e. you can plot them on different scales, so that they both start at the same position. Then you easily see which stock outperformed. But what if you want to see when it was a good time to switch as well?

You could potentially plot spread like this: http://archive.nytimes.com/www.nytimes.com/interactive/2011/...

It sounds like you are talking about a "price relative" or "relative strength" or "Bogle tell tale" chart.

Use a log scale, it makes easier to compare the performance in different periods and the spread between the two lines gives you the relative performance (when both stocks have the same performance the lines move in parallel, when the spread widens the upper line is performing better and vice versa).

This ask avoids the problem of having the entire comparison determined by the arbitrary or deceptive choice of start date for baseline price.

Divide each price x_t by x_0 and multiply by 100, so at t=0 the price equals 100 and at each t>0 x_t-100 is the percent change relative to t=0. Since your profit/loss is proportional to the stock's relative change since you bought it, this is what you want to know.

You plot them as ROI or other normalized value, eg by dividing the price of the stock by its initial value.

Then you can compare two stocks by looking at their return at T1 relative to their price at T0. This solves the problem that you’re better off investing in a $5 stock going to $10 than a $100 stock going to $115, since it normalizes it to 1 going to 2 and 1 going to 1.15 respectively.

For switching you may need something more complicated, like looking at the price at Tn divide by the price at Tn-1.

Please note that this is GLOBAL stocks.

I'm not going to buy any bonds in the current environment: Swedish 3-month bills yield -0.5%, 10-year bonds yield -0.2%. Whatever the historical data says, I can't see anyone buying at those prices unless they have to.

Well, if you’re a US based investor then it might not be a bad idea to pick up some of those -0.5% yield bonds when you consider the effect of currency hedging.

Conversely, for EUR based fund managers you’ll likely be better of sticking to negative yielding EUR denominated bonds vs treasuries.




There are a lot of entities with enormous sums of money who must buy quality debt at whatever yield is available, and they dwarf retail's appetite for bonds by several orders of magnitude

yeah but long term treasury and investment-grade medium duration corporate bonds are still yielding 2-4% . Ppl thought bonds were expensive in late 2018, and now in 2019 LQD, a bond etf, is up 15%.

Why not compare Philippine Treasuries and US Stocks next? #1 rule in finance: use the correct benchmarks ;D

Does this have any implication for indexing criteria?

And who is the author? A graduate student?

Do we really need a paper for something that takes 5 seconds to backtest?


Your backtest doesn't address anything that the paper covers and seems completely unrelated. It feel like you read the headline on HN and assumed you knew what the paper was about.

Here's the beginning of the abstract, which only takes four or five seconds to read:

"The majority of common stocks that have appeared in the Center for Research in Security Prices (CRSP) database since 1926 have lifetime buy-and-hold returns less than one-month Treasuries. When stated in terms of lifetime dollar wealth creation, the best-performing 4% of listed companies explain the net gain for the entire US stock market since 1926, as other stocks collectively matched Treasury bills."

What part of that does your 5 second backtest show?

Touché. I did think it's about the asset class as a whole and didn't bother reading the article because it looked like something as useful as a "are we in a recession" clickbait.

But that isn't surprising either, on a long enough timeline everyone's survival rate drops to zero (hence most individual stocks die, eventually). Also, stock returns are Pareto distributed, which means that a small fraction represents the majority of the gains (hence there will always be massive outperformers that dominate the space). Nothing surprising here.

It's a cool tool, thanks for sharing. I've been looking for something like this.

Quick question: why are long term treasuries returning 7.8% say for 2010 to 2019? I thought those treasuries were in the 3-5% range?

Because yields compress and realize gains ahead of time. Imagine someone issues a risk free 10% bond for 1 year, you buy it for 90 and it will pay 100 in a year. Next, imagine that the rate compresses to 5% within a week. The bond trades at ~95. You can sell it for a ~5% gain realized within a week.

Yields have been collapsing since 1981 (Volker inflation arrest):


BTW: a similar thing happened to stocks, but with far more volatility:


GLOBAL stocks. You linked to the Vanguard 500 index which is US only companies.

I tried to use ishares MSCI ACWI as the benchmark, but apparently that only goes back to 2009.

(Not that the paper is trying to do something like that. It is more about the performance problems of individual stocks. It is not claiming that treasuries outperform a stock market index.)

I leave it as an exercise to you to select "global stocks" as one of the tested portfolios ...

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact