Do Dividends Really Matter?

Selected Paper No. 57

Do Dividends ReallyMatter?

Merton H. Miller

Graduate Schoolof Business

The Universityof Chicago

Merton H. Miller, Leon Carroll MarshallDistinguished Service Professor of Financein the University of Chicago GraduateSchool of Business, has been active in theapplication of economic analysis to a widevariety of management problems in finance.Before joining the Graduate School of Busi-ness faculty, Professor Miller had been aneconomist with the U.S. Treasury Depart-ment and with the Board of Governors ofthe Federal Reserve System, a lecturer at theLondon School of Economics, and a facultymember of Carnegie Institute of Technol-ogy’s Graduate School of Industrial Ad-ministration. He received the A.B. degreemagna cum laude from Harvard and thePh.D. from Johns Hopkins. He is a formerpresident of the American Finance Associa-tion and is a fellow of the Econometric Soci-ety. He serves as coeditor of the Journal ofBusiness. Professor Miller has presentedthese views on dividends recently as part of aseminar on current management issues inStockholm, Sweden.

Do Dividends Really Matter?

I. The Question

There are few aspects of corporate financialpolicy where the gap between the academicsand the practitioners is larger than that ofdividend policy. The academic consensus isthat dividends really don’t matter very much.The market does not, and should not be ex-pected to, pay premium prices for firmsadopting what are sometimes called “gener-ous” dividend policies. If anything, generousdividends may actually cause the shares tosell at a discount because of the tax penaltieson dividends as opposed to capital gains.

Most practitioners on the other hand, bothamong corporate officials and investmentbankers, still continue to insist that a firm’sdividend policy matters a great deal. Andthey’ll cite example after example of com-panies whose price collapsed after passing aregular dividend; or whose price jumpedafter announcing a resumption of regulardividends.

My role will be that of a peacemaker try-ing to reconcile these conflicting views. Inparticular, I want to try to explain to thepractitioners, why, in the face of all this evi-dence of price gyrations in response to divi-dend announcements, otherwise sensibleacademics believe that a firm’s dividend pol-icy really doesn’t make much difference.

I’11 argue that the seeming evidence thatdividends do matter-evidence I hope atleast some of you believe, so that I’m not justpreaching to the choir-is not to be trusted.It’s an optical illusion.

An example I often use with my studentsis a stick in the water. If you use your eyesand look at the stick, it appears bent. But ifyou feel it with your fingers or if you pull itout of the water-that is, if you think aboutyour observation more deeply-you will re-alize that the stick is not really bent. It just

2 Selected Paper No. 57

looks bent. And similarly with dividends.They don’t really affect the value of theshares, but they may seem to if you do notthink about your observations more deeply.

II. The Explanation

My explanation of the illusion-why divi-dends look l ike they affect value eventhough, in a deeper sense, they do not-willhave two strands, which can be traced backultimately to two academic journal articlesthat appeared almost exactly twenty yearsa g o .

One article is devoted explicitly to thesubject of dividends and tries to explain why,as a matter of economic theory, no relationbetween dividends and value should be ex-pected. That is, it explains why the stick doesnot really bend.

The other deals with what seems to be anentirely unrelated problem, but turns out tohold the essential clue to the other side ofthe puzzle: why the stick looks bent, that is,why dividends look like they matter eventhough they do not.

The M-M Article

The first article entitled “Dividends,Growth and the Valuation of Shares,” is oneI wrote jointly with Professor Franco Mo-digliani (now of the Massachusetts Instituteof Technology) for the Journal of Business ofthe University of Chicago. We argued in ourarticle that part of the feeling that thereought to be a dividend effect was caused byan imprecise use of words.

To speak only of a firm’s dividend policy isincomplete. It’s like the sound of one handclapping, or a one-sided coin. The money topay the dividend must come from some-where. Uses of funds must equal sources offunds. Debits must equal credits. If a firmpays out a dividend (which is a use of funds)

Merton H. Miller 3

something else has to change in the sourcesand uses statement.

We argued in our paper that if you holdconstant the use represented by the firm’scapital budget, that is, its investment spend-ing, then paying out more dividends justmeans you will have to raise more fundsfrom bank loans or outside flotations ofbonds or stocks. A firm’s choice of dividendpolicy, given its investment p o l i c y is thusreally a choice of financing strategy. Doesthe firm choose to finance its growth by re-lying more heavily on external sources offunds (and paying back some of those fundsin higher dividends) or by cutting its divi-dends and relying more heavily on internalf u n d s ?

Put this way, it is by no means obvious thatthe generous dividend/heavy outs idefinancing strategy is always the best one, orvice versa. In fact, when we academics saythat dividend policy doesn’t matter much, weare really saying only that, given the firm’sinvestment policy (which is what really drivesits engine), the choice of dividend/financingpolicy will have little or no effect on itsvalue. Any value that the stockholders de-rive from the higher dividends is more orless offset, because they must give outsidersa bigger share of the pie.

If you find yourself resisting this notion, itmay be because you skipped over the crucialqualifying phrase, “given the firm’s invest-ment policy.” You may think of cases inwhich a firm doing poorly replaced its man-agement, and the new managers promptlyannounced an increase in the dividend (to dosomething for the long-suffering stockhold-ers) with the market presumably showingits appreciation by a big jump in the price.But was the price increase caused by the in-creased dividend? Or was it caused by thepresumed change in the investment policies


of the former management-the policies thatwere acting as a drag on the firm’s price?What would have happened to the price ifthe old management had announced the in-crease in the dividends but also the con-tinuation of its former investment policies,along with a proposal to f inance thosepolicies no longer with retained earnings-those earnings now going to the sharehold-ers in bigger dividends-but with the pro-ceeds of new issues of common stock or newbank borrowing?

You may think even in this case share-h o l d e r s w o u l d s t i l l p r e f e r t h e h i g hdividend/high financing strategy. At leastthey would have cash in their pockets. Butremember that if cash is what your share-holders want, they can get it even when youfollow the low dividend/low outside financ-ing strategy. They simply sell off part of theirholdings. It works in the opposite direction,too. If you adopt the high payout strategy,any investors who want to build up theirequity in your firm can do so by reinvestingthe unwanted cash in the additional sharesyou will have to issue. In fact, that presum-ably is what dividend reinvestment plans areall about.

Personal portfolio adjustments of thiskind are not costless, of course, particularlyonce we allow for taxes. That is why it maypay your firm to adopt and announce a divi-dend policy, even though in a deeper sense,dividends do not matter. If your firm an-nounces a high payout/high outside financingpolicy, you may attract a clientele of peopleor financial institutions preferring that pol-icy, and vice versa if you opt for the lowdividend route. To say that dividends do notmatter under these conditions is simply tosay that one clientele is as good as the other.

In sum, much of the belief that dividendsare terribly important is basically a confusion

Merton H. Miller 5

of the firm’s dividend policy with its invest-ment policy. Given the firm’s investmentpolicy-and again, I cannot overemphasizehow important it is to make that qualificationbefore beginning to appraise the role ofdividend policy-the dividend decision canbe seen as essentially a decision aboutfinancing strategy. You can always pay yourstockholders higher dividends while main-taining capital spending, if you are preparedto sell off more of their firm to outsiders.

A neat, almost literal, illustration is pro-vided by a story that appeared in the WallStreet Journal (March 4, 1982, p. 20) afterRCA cut its dividend for the first time sinceit began payouts in 1937. The Journal re-porter wrote:

According to John Reidy, analyst atDrexel Burham Lambert, the cut was asound management decision. He notedthat the alternative would have been toconsider liquidating some of RCA’s as-se ts .

RCA, the Journal goes on, “recently put itsHertz Rental car unit up for sale. Analystssaid a cut could have been prevented if abuyer had been found.”

1 sometimes wonder why RCA didn’t justpay out the Hertz shares as a dividend.Perhaps that would have made it all too ob-vious that for stockholders, a dividend pay-ment is merely putting money in one of yourpockets by taking it out of another.

If you still resist the notion that thefinancing drag or divestiture drag will cancelout the dividend boost, it may be becausethe phrase “dividend boost” brings to mindanother scenario that seems to have nothingto do with the firm’s investment or financingpolicy. In that scenario, a firm suddenly, byluck or skill, generates big increases inearnings and cash flow-so big in fact that it


can afford a substantially more generousdividend payment to the shareholders with-out any change in its investment budget orany resort to outside financing.

Under those conditions most real worldobservers would expect the firm’s share priceto rise, and I would agree. But what is re-sponsible for the higher value? Is it the use ofthe funds for more generous dividend? Or isit the source of the funds, to wit, the assumedbig increase in the current cash flow? Do youreally believe the price would always belower if instead of paying out its windfall asdividends, the firm announced its intentionto use the cash for retiring outstanding bankloans’or for buying shares in other com-panies (or, for that matter, for buying backits own shares)?

But you may say that you know caseswhere there was no significant increase incurrent cash flow (or at least none that themarket had been told about), and yet theprice still jumped up when a dividend boostwas announced.

I will agree with you. Prices often changewhen dividends are announced even whennot accompanied by an earnings announce-ment or forecast. There are even cases wherethe price followed the dividends despite anearnings move in the opposite direction. Thefirm announced a decrease in its accountingearnings, but an increase in its dividend, andits share price still rose.

These are the cases, par excellence, ofwhat I called earlier the “dividend opticalil lusion”-dividends can look as if theymatter even when they do not. To under-stand the source of the illusion we must turnto the second of the two fundamental papersI referred to earlier.

The article in question was entitled “RationalExpectations and the Theory of Price

Merton H. Miller 7

Movements” and was written by John F.Muth, now at Indiana University, and pub-lished in Econometrica.

Muth’s “Rational Expectations” paper hascome to be recognized as one of the mostimportant and influential papers written ineconomics in the twentieth century. In thelast ten years, the central notion of thepaper, as developed by my colleague RobertLucas at Chicago, among others, has virtuallydestroyed the intellectual underpinnings oftraditional macroeconomic theory and pol-icy, both of the Keynesian and of the morenaive monetarist varieties.

Like many other important ideas in eco-nomics, the central notion is basically a sim-ple one, though its development and elab-oration can become complicated and subtle.Reduced to its essentials-and I hopeeconomists will pardon me for taking suchliberties with the doctrine-the rational ex-pectations approach says that what matters ineconomics, and especially in policy makingin economics, is often not so much what ac-tually happens as the difference between whatactually happens and what was expected tohappen.

Politicians in America relearn the crueltruth of this point again and again during ourprolonged presidential election campaigns.Our presidential primary system involves asequence of trial elections within each politi-cal party, leading ultimately to the selectionof the party’s candidate during its conven-tion.

Had you been a foreigner visiting theUnited States in March 1980, during theweek of the Massachusetts primary, youmight have read in the morning paper thatGeorge Bush received 60 percent of thevotes and Ronald Reagan only 40 percent.As a visitor you could be pardoned for be-lieving that Bush had won and Reagan hadlost. But as we know Bush actually lost be-


cause he was expected to get 75 percent of thevote. The fact that he won a plurality of 60percent is of no consequence; the point isthat he did worse than expected, and so helos t .

It is bad enough when a candidate is run-ning for office, but the nightmare gets worseafter the election and the hard policy de-cisions must be made. Then, says the doc-trine of rational expectations, only the sur-prises in the policies will have the effects thepolicy maker is trying to achieve.

3

Suppose, for example, that an administra-tion takes office at a time of substantial un-employment in the economy. The adminis-tration’s economic advisers will tell the pol-icy makers about the so-called Phillipscurve-the trade off between inflation andunemployment. The monetarists among theadvisers will warn that the trade off can workonly in the short run. The Keynesians amongthem will argue that the trade off will applylonger than the monetarists suspect; but, inany event, in the long run the administrationwill be dead or out of office if no immediateaction is taken.

In the face of this advice an administrationmay well conclude: We really have no choicebut to open the monetary and fiscal spigots abit and trade some inflation for the benefitsof getting the economy going again. But thepublic knows the kind of advice the ad-ministration will be getting, so they antic-ipate that the spigots will be opened and thatprices will rise. That means that when theplanned price rise does come, it is no longera surprise. And if the inflation is not a sur-prise, the doctrine of rational expectationsimplies there will be no improvement in em-ployment. In fact, if the administration is notcareful, or not lucky, opening the monetaryspigots may produce less inflation than thepublic is expecting. Then the policy pro-duces the worst of both worlds: inflation and

Merton H. Miller 9

rising unemployment at the same time.Sound familiar?

Similar rational expectations nightmarescenarios can arise in the area that concernsus here-the effects of management’s divi-dend decisions on the value of the firm. Asthe date for announcing the regular quarterlydividend approaches, the market decideswhat dividend to expect, based on itsestimates of the firm’s earnings, investmentopportunities, and financing plans, which arein turn based on information the market hasabout the state of the economy, the industry,the firm’s past dividend decisions, the recentdecisions of other similar firms, changes inthe tax trade offs, and so on.

If the actual announced dividend is justwhat the market expected, there may be noprice movement at all, even if the announceddividend is larger than the previous one. Itwas expected to be larger and was fully dis-counted long ago. But if the announced divi-dend is higher than the market expected, themarket will start rethinking its appraisal.

They tell a story about the great diplomat,Prince Metternich, during the prolongednegotiations at the Congress of Viennawhich redrew the map of Europe after theNapoleonic Wars. After one particularly dif-ficult session Metternich was notified thatthe Russian ambassador had just died. AndMetternich is reported to have replied, “Ah,what could have been his purpose?”

Now the real world financial markets maynot be quite as suspicious as Prince Metter-nich, but when they see an unexpected in-crease in the dividend they, too, begin towonder: What does management mean bythat? They answer their question by saying:We can’t be sure yet what the firm’s earningsreally are, but given what we know of thefirm’s investment opportunities, manage-ment is certainly behaving the way we wouldexpect them to behave if they had turned in


better earnings than we had been thinkingthey would. And, since it is the earnings assources of funds and not the dividend as oneparticular use of funds that concern the mar-ket, we should expect upward price revisionseven though no earnings figure was actuallyannounced-just a dividend payment thatwas higher than expected.

And similarly in the other direction. Weall know cases where a firm cut or passed adividend unexpectedly, thereby setting off acrash in its price. Protestations by manage-ment that all was really well for the long pulland that this was just a way of redeployingcash to more profitable uses were of no avail.The market reasoned thus: They may say thepicture is rosy, but talk is cheap; and they areacting as if they had just experienced a dropin their earning power that was far worsethan had been expected or that they are let-ting on.

The following comments in the Wall StreetJournal (February 18, 1982, p. 36) on thedecision by AT&T to hold its payout to$1.35 per share neatly illustrate this interac-tion between the market’s expectations ofearnings and the firm’s dividend decisionsthat give rise to the dividend illusion.

Many analysts had expected an increaseof 10 cents or more in the quarterlyrate. . . .Apparently investors were disappointedthat the dividend wasn’t raised. AT&Twas the most active stock . . . yesterday,down $1.125.Analyst Steven Chrust of Sanford C.Bernstein and Company said continuingthe current dividend rate lends cre-dence to the recent cuts in earningsprojections for AT&T.

In sum, unexpected dividend actions in aworld of rational expectations provide the


better earnings than we had been thinkingthey would. And, since it is the earnings assources of funds and not the dividend as oneparticular use of funds that concern the mar-ket, we should expect upward price revisionseven though no earnings figure was actuallyannounced-just a dividend payment thatwas higher than expected.

And similarly in the other direction. Weall know cases where a firm cut or passed adividend unexpectedly, thereby setting off acrash in its price. Protestations by manage-ment that all was really well for the long pulland that this was just a way of redeployingcash to more profitable uses were of no avail.The market reasoned thus: They may say thepicture is rosy, but talk is cheap; and they areacting as if they had just experienced a dropin their earning power that was far worsethan had been expected or that they are let-ting on.

The following comments in the Wall StreetJournal (February 18, 1982, p. 36) on thedecision by AT&T to hold its payout to$1.35 per share neatly illustrate this interac-tion between the market’s expectations ofearnings and the firm’s dividend decisionsthat give rise to the dividend illusion.

Many analysts had expected an increaseof 10 cents or more in the quarterlyrate. . . .Apparently investors were disappointedthat the dividend wasn’t raised. AT&Twas the most active stock . . . yesterday,down $1.125.Analyst Steven Chrust of Sanford C.Bernstein and Company said continuingthe current dividend rate lends cre-dence to the recent cuts in earningsprojections for AT&T.

In sum, unexpected dividend actions in aworld of rational expectations provide the

Merton H. Miller 11

market with clues about unexpected changesin earnings. These in turn trigger the pricemovements that look like-but only looklike-responses to the dividends themselves.

I wish I could safely conclude on this noteof harmony in which the academic view thatdividends do not matter, and the practition-ers’ view that they matter very much, arenicely reconciled. Both views are correct intheir own ways. The academic is thinking ofthe expected dividend; the practitioner, of theunexpected.

But I am afraid I have one more piece ofbad news to deliver to the practitioners: theunexpected can become expected. Some ofyou may be wondering: If the market isgoing to read an estimate of my earnings intomy dividend announcements, why not justpay out more than is expected, even if itmeans cutting back on profitable in-vestments? That will run the price up in theshort run and benefit those shareholderswho are selling out in this period. It will alsoget me off the hook with all those pensionfund managers who keep threatening todump the shares if the price is low at the endof the quarter when their performance isevaluated. True, passing up investment op-portunities to prop dividends and prices willhurt those not selling out, but the immediatebenefits of the price rise or prevented fallmay well be larger than the loss to the stayersover the long run.

Unfortunately (or perhaps fortunately),this tactic cannot be counted on to work in aworld of rational expectations. The marketwill say: We know how managements think.They’ll figure that we think they are tying thedividend to earnings in the same old way;and they’ll raise the dividend above what weexpect to create the impression of an unex-pectedly large increase in earnings. But weknow they’ll be tempted to do that, and forthat reason we’ll expect an even larger divi-


dend than before at any given level of earn-ings .

And now you’re really in trouble, just likethe policy maker in the inflation case. Youdare not raise the dividend even further inhopes of generating a surprise; the cost interms of foregone future earnings would betoo great. And you dare not cut back to amore sensible dividend because the marketis already assuming that you’ll pay a biggerdividend than you can really afford in thehopes of getting the price up, and so willinterpret the disappointingly low dividend asa sign of bad news.

So you’re stuck. You’ve lost your ability toconduct an independent dividend policy.You have to deliver the dividend that themarket expects you to deliver. If you don’tdeliver, you’ll have to pay for it either inprice falls (if you deliver too little as didAT&T) or lost opportuni t ies and un-necessary financing expense (if you pumpout too much).

III. Conclusion

I hate to end on such a negative note, par-ticularly these days when there’s so littleanywhere to be cheerful about. So I will re-call the positive side for practitioners in theacademic view of the dividend problem.

Dividend policy may not be an effectivemanagement tool and may not even be com-pletely under your control in a world of ra-tional expectations, but there are things thatdo matter and over which you do have morecontrol. I refer, of course, to the firm’s in-vestment decisions and to the engineering,production, personnel, marketing, and re-search decisions that underlie them. Thesedecisions are in what economists call the“real” side of the business, and they generatethe firm’s current and future cash flows.

And that, you’ll find, is what really matters.

Documents

Do Dividends Really Matter?