The Unbearable Snarkiness
of Dividends

by

Ross M. Miller
Miller Risk Advisors
www.millerrisk.com
September 11, 2006

[This commentary  appears in the September/October 2006 issue of Financial Engineering News. This is the raw, unedited version of that commentary.]

The simple idea of issuing periodic cash dividends to stockholders was an early triumph of financial engineering. Not so long ago, common stock was considered a more or less disreputable investment. To attract equity, companies were forced to pay dividends that were large enough that their stock yield would exceed the interest rate for their bonds.

The invention of dividends was not just an engineering coup; it was a practical application of behavioral finance long before that field formally came into existence. After all, companies that paid dividends were merely giving their investors' money back to them in a way that framed stocks in a positive light relative to bonds.

Companies no longer need to pay high dividends to sell their equity. The big turnabout came during the go-go years of the 1960s. Not coincidentally, in 1961 the Journal of Business published an article written by Franco Modigliani and Merton Miller (M&M, for short) that, together with an earlier 1959 article on corporate capital structure, permanently altered the financial landscape. M&M demonstrated that a company's dividends payments were "irrelevant" when investors could re-engineer its stock's cash flows through their own purchases and sales without affecting the company's value.

Since dividends do not matter in this theoretical sense and the taxes on dividends (which are outside the M&M model) matter in a negative way, why do many companies continue to incur the expense of paying them? A great deal of financial theory has been hatched in an effort to answer this question.

My personal objection to dividends is that they are inelegant or, as my British friends would say, they are snarky. Consider option valuation theory. Everything goes along beautifully, with closed-form solutions galore, until dividends make the scene.

One aspect of dividends is so unbearably snarky that most financial engineering textbooks have chosen to ignore it completely. Dividends, it so happens, do not materialize at a moment in time; instead, they are the end product of a time-consuming process. This process starts with an official corporate announcement. Included in the announcement are the amount of the dividend, the date on which the stock must be held in order to receive dividends (known as the record date), and the date the dividends are finally paid. Because stock purchases take time to settle, the critical date for dividends is the ex-dividend date. This is the date on which the stock must trade in the marketplace in order for the buyer to take ownership of the shares on the record date. At present, there is about a one-month lag between the ex-dividend date and the payment date for most stocks; however, with some companies have stretching this out to two months.

At the time when M&M were writing about dividends, traders with low transaction costs could profit nicely by engaging in dividend capture arbitrage—purchasing a stock on the day before the ex-dividend date and holding it overnight to get an entire quarter's worth of dividends in a single day. Such freebies were too good to last, especially after financial deregulation led to lower transaction costs. Nowadays, the bid and offer prices of U.S. stocks are automatically marked down by the dividend payment as soon as they go ex-dividend.

The M&M article defined a stock's total return so that only dividends paid out during the holding period were counted. It is now standard practice for researchers, following the lead of the University of Chicago's Center for Research in Securities Prices (CRSP), to compute stock returns assumed that the dividends are received on the ex-dividend date regardless of when the payment date is. Consider a stock that is trading at $100 per share and makes quarterly dividend payments of $1 per share. If one were to hold the stock overnight in order to capture the dividend and if a fortunate turn of events caused the price of the stock to climb to $101 per share (ex-dividend) during that time, according to CRSP one would receive a total return (net of commissions) of 2%: 1% in stock appreciation and 1% in dividends. Unfortunately, the $1 dividend cash flow is not immediately available and effectively becomes a short-term unsecured obligation of the issuer. Assuming a 6% discount rate and two months until the dividends are paid, the delay knocks a full basis point (b.p. = 0.01%) off the value of the dividend. (CRSP, it should be noted, provides information that can be used for making such adjustments to their database's stock returns.)

One basis point is generally insignificant, but when exchange-traded funds (ETFs) are involved, basis points can add up. In a 2002 Journal of Business article, "Spiders: Where are the Bugs?" Edwin J. Elton and three coauthors showed that in the case of SPDRs, a popular ETF that tracks the S&P 500 Index, the treatment of dividends significantly affects their rate of return. SPDRs accumulate the dividends paid by the S&P 500 companies and distribute them quarterly. While awaiting distribution, the accrued dividends stay in a non-interest bearing account.

Using a sample period during which the S&P 500 Index increased by 22% a year, Elton found that SPDRs lagged that benchmark by nearly 10 basis points a year in addition to their management fees, which were 18.45 b.p. at the time. Given that many index products now available in the retail market only lag their benchmark index by as little as 10 b.p. a year, an additional 10 b.p. represents a major shortfall in a highly competitive market.

Most of this shortfall comes about because parking the dividends in cash precludes them from being reinvested in the index, which is particularly damaging when the market is rising. (Large investors who are allowed to convert their SPDRs into its S&P 500 components can get their hands the cash.)

In essence, one should not view money placed in SPDRs as a pure investment in the S&P 500 index, even though they are marketed that way. SPDRs are like a very dry index martini—lots of index with the slightest splash of cash.

Because dividend payments for stocks in all the major indexes are a matter of public record, one could dynamically undo the cash holdings of the SPDRs in a way that keeps one fully invested in the index at all times. If one were lazy and willing to endure some tracking error, one could simply estimate the average beta of SPDRs the course of a year (Elton gets 0.998 to the nearest hundredth) and leverage accordingly. For the typical investor, who is not inclined toward do-it-yourself financial engineering, the dividend treatment of SPDRs is a very real strike against them.

Although companies are certain to persist in paying dividends, there is something they can do to diminish their snarkiness. The lag between the record date and the payment date is just an artifact from the days of paper recordkeeping. Given the electronic nature of both shares and money, why not move the record date up to the payment date, or the business day before at the earliest? It might take amending a rule or regulation here and there, but such a change should be possible. Anything that brings the world of practice closer in line with the world of theory is a good thing.