Rob Arnott's Editorials in the Financial Analysts Journal

[George$]

New Arnott here.

The cost of noise trading and the structural bias in
our cap-weighted benchmarks may be a major, even
primary, driver of the historical return advantage
associated with equal weighting, with value stocks,
with small-cap stocks, with GDP weighting in international
portfolios, and with rebalancing.

Norbert: Thank you. The Arnott article seems both significant and interesting - but I don't think I fully understand it. Help. How much spoonfeeding are you willing to engage in?

Should the topic have its own thread?

[Norbert Schlenker]

Help. How much spoonfeeding are you willing to engage in?

I think I don't really understand it myself. I'm hoping someone smart will translate it for us all.

[Bylo Selhi]

A nice profile on Arnott as well as a summary of his contention that market-cap weighted indexes are inherently flawed. New Take on Index Fund Concept

Robert Arnott, a Pasadena-based money manager, thinks the modern investment business can learn something from Copernicus' heresy: The industry, he says, should be less afraid to question conventional wisdom — particularly when it's deeply entrenched. Arnott is a well-known name in investment circles who heads a firm called Research Affiliates. Since 2002 he also has served as editor of the Financial Analysts Journal, a forum for Wall Street pros, academics and others with a serious interest in finance. A statistics whiz with a passion for vintage motorcycles, Arnott has used his journal editorials to poke at many classic investment dogmas, question their application in the real world and effectively challenge his peers to wonder whether there are better ways to construct clients' portfolios.

Lately, he has been focused on a portfolio of immense proportions: the estimated $1.7 trillion that U.S. investors have in index stock funds, most of which seek to replicate the performance of the blue-chip Standard & Poor's 500 index. The vast majority of market indexes, and the funds that track them, are capitalization-weighted — meaning, the stocks that dominate the indexes at any moment are the biggest issues in terms of market capitalization (a company's share price multiplied by the number of shares outstanding).

Arnott contends that is fundamentally flawed. Why? Think about the technology companies that were paramount in the market in March 2000, at the peak of the dot-com mania. In a capitalization-weighted index, "You automatically overweight all the overvalued stocks," Arnott says. A better way to construct an index, he maintains, would be to weight stocks based on measures of what firms actually have accomplished, rather than solely on the often fleeting beauty contest of stock capitalization...

[yielder]

A nice profile on Arnott as well as a summary of his contention that market-cap weighted indexes are inherently flawed.

We had our own discussion on the subject.

[rmark]

Originally Wells Fargo had a equal weight index, as opposed to the current cap -weighted. What killed it was transactions costs. With cap weighted, any change in value is relative to the rest of the portfolio, so no transaction is needed. Arnott is essentially suggesting we go back to circa 1972. ( I'll have to find my bell bottom pants)

[DanH]

Originally Wells Fargo had a equal weight index, as opposed to the current cap -weighted. What killed it was transactions costs. With cap weighted, any change in value is relative to the rest of the portfolio, so no transaction is needed. Arnott is essentially suggesting we go back to circa 1972. ( I'll have to find my bell bottom pants)

In a recently published research article, he's suggesting simply looking at better measures of economic significance. In other words, the value of a company's publicly traded shares reflects an outlook for the future (which may or may not materialize). But a company's revenue, for instance, is a concrete and recent measure of how prominent a company is in the context of the economy.

His reasoning makes sense. What he's suggesting is what he consideres to be a better way to construct the index (a market barometer). In his quest for improved measures of economic significance, he not only looked at 'sales' but many other factors that are independent of market prices.

[Bylo Selhi]

Arnott is essentially suggesting we go back to circa 1972.

Arnott is saying that market cap weighting, while the least costly to implement, is probably the worst criterion to use for indexing because "You automatically overweight all the overvalued stocks."

He doesn't say we should go back to equal-weighting. In fact he suggests a combination of several other criteria. From the LA Times article, "A better way to construct an index, he maintains, would be to weight stocks based on measures of what firms actually have accomplished, rather than solely on the often fleeting beauty contest of stock capitalization... The 50-year-old Arnott and his peers at Research Affiliates have spent the last few years developing an indexing system using such measures of business success as five-year average sales and operating earnings, employment and net asset value."

the value of a company's publicly traded shares reflects an outlook for the future (which may or may not materialize)

Based on Graham/Buffett's "Price is what you pay. Value is what you get." I think you meant price, not value

Arnott's contention is that the other criteria are a better reflection of value than price [=> market cap] alone.

[DanH]

the value of a company's publicly traded shares reflects an outlook for the future (which may or may not materialize)

Based on Graham/Buffett's "Price is what you pay. Value is what you get." I think you meant price, not value

Yes, price - or more specifically total market value.

[Norbert Schlenker]

A postscript. Rick Ferri reports on his attendance at a conference where Arnott's thesis met with considerable skepticism.

[DanH]

A postscript. Rick Ferri reports on his attendance at a conference where Arnott's thesis met with considerable skepticism.

Part of Ferri's post...

The astute audience picked up quickly that the past "alpha" claimed by Arnott in his indexes were totally dependent on overweighing value and small cap stocks during the period he back-tested.

IIRC, Arnott acknowledges the small and value tilts present in his fundamental indexes. I think there are good arguments on both sides. For instance, EMHers might argue that current stock prices (and, by association, market caps) aggregates the best available information. But I also see the use of more concrete statistics like dividends or sales as an intuitive measure of 'size' or economic significance.

Side note: For what it's worth, I've reviewed Dalbar's methodology in the past and found it lacking in many ways. While I'm not published, my guess is that it would never make it into a juried journal without major changes.

[martingale]

Firms have a substantial power to manipulate their dividend record. Even in a year when a firm shows a loss, management is frequently able to maintain or even raise the dividend. In general, firms do not cut a dividend until they are absolutely forced to do so. Overall, statistically, a firm is 15 times more likely to maintain or raise its dividend than it is to cut its dividend. If that all sounds wonderful note the downside: By the time management does finally communicate an admission of trouble by cutting the dividend, the trouble is usually so serious that the stock isn't just in the toilet, it's been flushed.

In theory the current price is the best estimate of the present value of all future dividend payments, and so the price will reflect expectations about an upcoming dividend cut long before the cut actually arrives. The dividend history tells you what the price used to be; the price tells you where the dividend is going.

[yielder]

In theory the current price is the best estimate of the present value of all future dividend payments, and so the price will reflect expectations about an upcoming dividend cut long before the cut actually arrives. The dividend history tells you what the price used to be; the price tells you where the dividend is going.

Theory schmeory. The price often is the worst indicator. In the short run, it's a beauty contest measure. Price tells you nothing about where the dividend is going. The dividend tells you where price is going. And the financials & notes give you clues to where the dividend is going. I say clues because they don't say much about the industry or the company management personality.

When I looked at dividend cuts among the S&P Dividend Aristocrats, I found that with one exception all of the freezes/cuts come after significant acquisition/merger activity.

Although never an S&P Dividend Aristocrat, a classic example is Goodyear.



In 1999, their debt increased by 98% in response to an acquisition and a huge increase in CAPEX. All of this debt was funded short term. In 2000 and 2001 they rolled the ST debt into LT debt while debt was increasing again by 36%. Long rates had been steadily declining through 1997 and 1998 until they bottomed below 5% in September 1998. From that point until January 2000 long rates rose 1.5% to 6.5%.

The handwriting was on the wall for the dividend cut but the significant price reaction came only when the dividend was cut. Acquistions and debt aren't necessarily a bad thing but they should be a screaming alarm to watch and listen closely to what is going on.

[DanH]

Firms have a substantial power to manipulate their dividend record. Even in a year when a firm shows a loss, management is frequently able to maintain or even raise the dividend.

As long as this isn't done extensively, it's not a big deal. So what if a firm uses excess cash reserves to sustain a dividend one year? It would be a sign of confidence of better times ahead - provided management isn't brainless

Besides, when talking of groups of stocks - like a benchmark - the exceptions you highlight get drowned out by the more common examples of solid dividend payers.

Allow me to edit and mark up your next clip...

In theory the current price is the best estimate of the present value of all future dividend payments, and so the price will reflect expectations about an upcoming dividend cut long before the cut actually arrives. The dividend history tells you what the price used to be; the price tells you where the dividend is going.

I'd agree with the first part - that is the theory. The last part sounds much more certain that the market is not only more correct than most individuals but quote precise as indicated by your wording. As yielder so eloquently pointed out: theory schmeory;)

Just about every dividend cut I can recall was followed by swift, downward price action. Remember TRP and Rogers Sugar Income Fund in 1999 (or 2000)? Both were trucking along, but then got pummelled after announced dividend/distribution cuts.

[yielder]

As long as this isn't done extensively, it's not a big deal. So what if a firm uses excess cash reserves to sustain a dividend one year? It would be a sign of confidence of better times ahead - provided management isn't brainless

Back in the early 90s when IBM was bleeding red ink, it maintained its dividend and was yielding over 8%. The stock price was hammered as the company was punished for maintaining a huge dividend that they were funding from debt offerings. Then the div was cut by 55% in Q1/93 and again by 54% 6 months later.

[martingale]

Theory schmeory. The price often is the worst indicator. In the short run, it's a beauty contest measure. Price tells you nothing about where the dividend is going. The dividend tells you where price is going.

Theory schmeory indeed. Fortunately, these days, the scientific method pervades the study of finance and so theory tends to be based on extensive analysis of real world data.

In June 2004 Malcom Baker and Jeffrey Wurgler waded through dividend returns from 1962 to 2000 and determined that aggregate dividend payouts are best explained by investor demand: Managers cater to investors by increasing dividends when dividends are in demand, and omitting dividends when investors are less interested in them. Dividend payouts did not signal increased future financial health, but rather, investor demand for dividends.

In April 2002 Rodney Boehme and Sorin Sorescu looked at all returns from 1927 to 1998 and determined that dividend policy changes had no observable impact on future prices for US large cap stocks (representing 88% of the market). They did find some evidence of post policy change price drift for smaller stocks but only in the period from 1965 to 1998 and only when the portfolio was equal weighted. They concluded that this effect could be explained by a change in the riskiness in the firms concurrent with the dividend change. In other words, the result agreed with the EMH.

In Dec. 1992 Harry and Linda DeAngelo, and Douglas Skinner picked through NYSE returns from 1980-85, lookng at the 167 firms that recorded an annual loss and comparing them to the 440 that did not. They determined that an annual loss tended to be a necessary condition for a dividend cut in firms with an established dividend record. The cut came after the loss.

Overall, the "theory schmeory" about dividend policy changes not having any impact on future prices (edit: other than changes explained by other concurrent news) is remarkably well supported by the dividend record of the last 78 years. The three studies quoted above are from the Journal of Finance. If you dig around in other journals, you could no doubt find many more examples.

[jiHymas]

Price may not be a particularly good predictor of future dividends, but I don't know of a better one.

Just about every dividend cut I can recall was followed by swift, downward price action. Remember TRP and Rogers Sugar Income Fund in 1999 (or 2000)?

To the extent that the decision to cut the dividend represents previously undisclosed material information, such price declines would be expected.

[yielder]

Theory schmeory indeed. Fortunately, these days, the scientific method pervades the study of finance and so theory tends to be based on extensive analysis of real world data.

I'm not rejecting theory, just the EMH part that says "price price will reflect expectations about an upcoming dividend cut long before the cut actually arrives."

Managers cater to investors by increasing dividends when dividends are in demand, and omitting dividends when investors are less interested in them.

Yep. In general, that's true. That's why I'm usually only interested in companies with a dividend track record. They cater to the investor who always wants dividends.

They concluded that this effect could be explained by a change in the riskiness in the firms concurrent with the dividend change.

Yep, dividend history is useful in identifying possible investments but sustainability is more important in the buy decision.

They determined that an annual loss tended to be a necessary condition for a dividend cut in firms with an established dividend record. The cut came after the loss.

Too bad they didn't dig further to see what led to losses. That would be useful and interesting.

Overall, the "theory schmeory" about dividend policy changes not having any impact on future prices is remarkably well supported by the dividend record of the last 78 years.

I agree but that's not what you said: "price will reflect expectations about an upcoming dividend cut long before the cut actually arrives."

[George$]

It's possible that this has been posted before, but it's very good and very short (2-page pdf)

Dividends and the Three Dwarfs (Arnott - 2003)

Some text from it

The importance of dividends for providing wealth
to investors is self-evident. Dividends not only
dwarf inflation, growth, and changing valuation
levels individually, but they also dwarf the combined
importance of inflation, growth, and changing
valuation levels. This result is wildly at odds
with conventional wisdom, which suggests that,
while the return from bonds is wholly dependent
on income, stocks provide growth first and income
second. It is startling to realize that dividend
growth has averaged less than 1 percent above
inflation during the past 200-year period. And it is
shocking that real per-share dividend and earnings
growth on the S&P 500 Index since 1965 has been
zero.

[Norbert Schlenker]

Disentangling Size and Value

[gyrfalcon]

RE: "Disentangling Size and Value"

Bob Tattersall is an Active manager who has invested according to these concepts for his entire career. Consider the following ratios, which have been **featured** in the regular fund reports for a number of years:

......................................12/01 12/02 12/03 12/04 9/05
Saxon Small Cap MF P/BV.... 0.93 0.92 1.30 1.38 1.37
TSE 300 > TSX Composite... 1.94 1.80 2.01 2.20 2.27

.....................................12/01 12/02 12/03 12/04 9/05
Saxon Small Cap MF P/CF.... 4.89 6.71 9.20 7.40 8.46
TSE 300 > TSX Composite.. 9.27 9.47 10.10 9.91 10.52

......................................12/01 12/02 12/03 12/04 9/05
Saxon Small Cap MF P/Rev.... 0.35 0.40 0.50 0.52 0.58
TSE 300 > TSX Composite.... 1.14 1.22 1.36 1.54 1.56

The 7-yr compounded annual return on SSC MF to 9/05 was 15.87%. I've enjoyed all but 4 months of that. [The 15-yr figure is 14.85%]. [I realize the "Table" is not perfect, but Preview tells me it's readable]. gyr.

[Taggart]

New ETF Targets Stocks With Best Stats

Published: December 19, 2005 12:25 PM Dow Jones

By John Spence

BOSTON (Dow Jones) -- PowerShares Capital Management is at it again, unveiling another exchange-traded fund that challenges traditional notions on index investing.

On Monday, the PowerShares FTSE RAFI U.S. 1000 Portfolio (PRF) began trading on the New York Stock Exchange. The ETF is based on a controversial new "fundamental" indexing strategy pioneered by Robert Arnott, chairman of Research Affiliates and editor of the Financial Analysts Journal.


LINK

[Bylo Selhi]

New ETF Targets Stocks With Best Stats

OTOH, from FTSE/RAFI PowerShares ETF launches on Big Board

A number of index industry experts have claimed that they have been unable to duplicate Arnott’s results, and attack the data as back-tested and possibly even inaccurate. If this is so, we can expect intensive reviews of the data as more products are launched. In any event, the proof will be in the pudding. And the pudding, in this case is “future returns”.

But before you get too excited, take a look at this chart (from JoI [no link provided...Bylo]) stretching back into the 1960s. The US RAFI 1000 does not really begin to outperform until the very tail end of the Internet bubble. That covers a unique period when the market underwent a huge correction – primarily to large cap, growth-oriented technology stocks...

[yielder]

New ETF Targets Stocks With Best Stats

FWF'sdiscussion on market cap based indexes.

[George$]

And another link courtesy Bylo's link above ---
from the NY Times Dec 18
Start With a Stock Index. Now Try to Turbocharge It. By JACK EGAN -- in part

A NEW kind of designer index fund has arrived on the scene. Instead of trying merely to match the performance of the stock market, as measured by some broad index like the Standard & Poor's 500, these new funds are aiming to outperform it.

The S.& P. 500, the most widely used benchmark, compounded by 9.5 percent a year, on average, over the 10 years ended Sept. 30. Statistics show that about two-thirds of actively managed equity mutual funds failed to outperform the market over 3-, 5- and 10-year periods, let alone longer spans. That has been the impetus for index funds that aspire merely to match the market's return.

Backers of the new variety of index funds say that the products can beat the market and other index funds, either because they have found ways to exploit stock market inefficiencies or because the indexes, especially those based on market capitalization, like the S.& P. 500, are themselves flawed.

None of the handful of funds that fall into this category have been around long enough to register persuasive long-term results. Nevertheless, the best performers have managed to more than just outpace benchmarks. In fact, they have clobbered them, behaving a bit like index funds on steroids.

PowerShares Capital Management, a small but fast-growing issuer of exchange-traded funds, the increasingly popular alternative to traditional mutual funds, has been the main innovator in this arena. Based in Wheaton, Ill., it has started a series of index funds that it calls X.T.F.'s and whose goal is to beat benchmark indexes.

Like other E.T.F.'s, they trade on exchanges the way stocks do. But they are different in their composition. They mirror custom-built indexes based on so-called Intellidexes, created by a quantitative unit of the American Stock Exchange. In these formulations, stocks in an existing index are sifted and ranked for potential capital appreciation. Each stock is scored by using 25 variables such as cash flow, earnings growth, share-price momentum and timeliness. The 100 top-ranked stocks make up the custom index, with the added requirement that sector weightings parallel the broader index. If, say, 20 percent of an index is composed of financials, so is the Intellidex for that index.

PowerShares offered the first two of these funds at the beginning of May 2003. From their start, both have run well ahead of their index bogeys. Its flagship core fund, Dynamic Market Portfolio, is up 70 percent since it started, compared with a 39 percent gain for the S.& P. 500 over the same period. Its Dynamic OTC fund, has climbed 75 percent, while the Nasdaq 100 index is up 53 percent.

In a variation on the theme, Pimco, the money management giant based in Newport Beach, Calif., started two hybrid equity mutual funds at the end of May. The funds, called Fundamental Index Plus and Fundamental Index Plus Total Return, are based on a proprietary index of large-capitalization companies that also seeks to outgun the S.& P. 500. The funds are run by Pimco's star portfolio manager, William H. Gross, who is better known for his bond-fund expertise. Both contain a small bond component but are mainly built around the Research Affiliates Fundamental Index, developed by Robert D. Arnott, who serves as a subadviser.

[Taggart]

Index Mutual Funds Get an A in School, C- in Life: Chet Currier

Nov. 18 (Bloomberg)

So index funds don't always get above-average results, even with their natural advantage of lower costs (less trading, no company research to do). Lately, an intriguing explanation has emerged to account for this. In indexes that give the greatest weight to stocks with the biggest market value, as many do, overpriced stocks naturally tend to dominate underpriced stocks.

``There has been a lot of focus in the investment literature recently on the idea that market capitalization weighted indexes contain systematic pricing error,'' said Hugh Whelan, a fund manager and analyst at ING Investment Management in Hartford, Connecticut. ``If there is pricing error, it is axiomatic that an index overweights every `overvalued' stock and underweights every `undervalued' stock.''

LINK