Buy and hold, please

For Nasdaq,
the difference between buy-and-hold and dollar-weighted returns is even
larger. An investor who bought the Nasdaq index in 1973 and sold in
2002 would have earned an average yearly return of 9.6 percent. But the
typical investor in Nasdaq earned only 4.3 percent over this period.
This is true not just in the United States – the same thing occurred in
18 of 19 international markets that Mr. Dichev examined.

That’s Hal Varian, who argues that actively trading investors are excessively swayed by what is in the news, thus earning less than random returns.  If you can’t restrain your trading, at least chuck your newspapers and magazines.


This was the thesis in Fooled by Randomness, which deserves a mention in such a conenction..

So if the stocks managed 9.6%, but the "typical" investor only saw 4.3, where did the rest of the $ go? Shouldn't the sum of investors come back to 9.6?

Speaking of something not being 'news'....

You guys missed Varian's point. The "missing" return is based on investors trading in and out of the Nasdaq. the 9.6% is the time-weighted return, the 4.3% is the dollar-weighted return. Investors buying at the top and selling at the bottom is what leads to the huge mismatch b/w time and dollar weighted returns.

There is a ton of data on mutual fund performance and the huge delta b/w time and dollar weighted returns. David Swensen of Yale has calculated that over the period 1996-2005 investor peformance in mutual funds lagged the time-weighted return by 13% per year (not percentage points, so if the return was 10% in the fund, investors got 8.7%). Another example is that many investors in the tech bubble got killed. A "hot" new tech fund would be up say 100% in 1999 on $100M of assets. Investors would notice the performance and pour assets into the fund, ballooning assets to say $1B by the end of Q1 2000. The fund then loses 75% over the next two years. The funds three year performance is then down 50% on a time weighted basis, but on a dollar weighted basis it will be much closer to 75%, since the vast majority of assets were invested AFTER the fund had doubled in 1999. To add insult to injury, investors also got hit with huge capital gains bill in 2000, sometimes equal to as much as 20% of assets. (these numbers are made up, but there are plenty of real examples, I'm just too lazy to find them. The Jacobs Internet Fund and Munder NetNet fund are two that come to mind).

I think Varian has misread the paper. The paper really says nothing about styles of trading like buy and hold investors or active traders. Indeed, mathematically it is impossible for the aggregate return to be affected by who buys and sells a particular stock at which time. If person #1 sells a stock to person #2 that might change person #1's return but it cannot change the aggregate return.

However, firms CAN affect the aggregate return by issuing or buying back stock. This is the point of the article. What the author documents is that companies issue stock when the market is high (duh, think late 1990s tech IPOs) and buy back stock when they think their companies' stock is undervalued. Apparently they do a better job of timing the market than individual investors. This is the source of the discrepancy between time weighted and dollar weighted returns. It's got nothing to do with being a buy and hold investor versus an active trader.

I don't see an obvious solution for individual investors--all mutual funds or index funds must almost by construction expand and contract with aggregate capital flows into and out of the market. Maybe I should sell when I see lots of IPOs and buy when I see lots of buybacks...

Adam what you say would only make sense if there was a fixed number of investors over the period of time who were each invested in every stock in the market. It is possible that I can sell Home Depot in 1999 at $60 to some other rube willing to pay 50x earnings for a retailer and then go to cash, or buy an undervalued small cap company or invest in timber, etc... The NASDAQ example shows that people's market timing - trading in and out of the NASDAQ - is very poor and they suffer as a result. If the only choice was to invest in the NASDAQ, then sure the aggregate returns would have to equal the index return, less transaction costs. But reality is quite different.

Adam, after closer examination, your comment appears correct. Varian did misread the paper. It has nothing to do with trader behavior or getting in and out at the wrong time. My previous comment was meant to bring up the point that for every buy transaction there is a comparable sell transaction, so that the "typical" investor is held neutral to any amount of trading activity. The only exceptions to this (pointed out by Adam) are secondary stock issues, buybacks, and dividends where the counterparty is the company itself.

A couple of thoughts: 1. can we look at corporate stock buybacks as a massive insider trade (being done on average in times just before better-than-average returns) and 2. does this show that companies who issue dividends tend to perform better in the future (or, to be causality-neutral, companies in a position to do better in the future are more likely to issue dividends now)

Steve, regarding your points:

1. Please note that I acknowledged your exceptions in my post. My comment is meant to refute the arguments of traders all being of the variety where they chase rising stocks only to get crushed later. Other than the exceptions we agree on, there is a trader on the other side who benefitted to an ofsetting amount.

2. If you are right, then I bet the results from the whole paper can be explained by this. All of those dividends probably affect the calculations much more than secondary offerings, so if a small div payer outperforms a non-payer by 1% and a large div payer outperforms the small by another 1%, then...

I'm sure there are a bunch of discussions and papers on why div payers could do better in the long run (signaling, advantage in capital access, etc...)-- so does that mean this particular paper is just regurgitating old news using different calculations?

Outsider trading? I meant insider. Oops.

Comments for this post are closed