By Roger G. Ibbotson, Jeffrey J. Diermeier and Laurence B. Siegel The Demand fer Capital Market Returns: A New Equilibrium Theery Investors demand more of an asset, the more desirable the asset's characteristics. The most important characteristic is its price, or expected return. By varying price, any and all assets become desirable enough for the capital market to clear. Asset characteristics other than price include both risk and non-risk characteris- tics. The Capita! Asset Pricing Model and Arbitrage Pricing Theory have described the risk characteristics. The non-risk characteristics are not as well understood. They include taxation, marketability and information costs. For many assets, these non-risk characteristics affect price, or expected return, even more than the risk characteristics. Investors regard asset characteristics as positive or negative costs, and investors evaluate expected returns net of these costs. The New Equilibrium Theory (NET) framework applies to all assets—including stocks and bonds, real estate, venture capital, durables and intangibles such as human capital—and incorporates all asset characteristics. P RICES IN CAPITAL MARKETS are set by ences and aversions. Investors translate each the interaction of demand and supply. characteristic into a cost, and require compensa- This relationship is commonly expressed tion in the form of expected returns for bearing in terms of the "supply of and demand for these costs. Thus, although all investors are capital." But viewing it from the opposite per- assumed to perceive the same before-cost'ex- spective—^that is, in terms of the demand for pected return for any given asset, each has' and supply of capital market rclurti:^—has the individually determined costs he must pay to advantage of focusing our attention on returns hold that asset. On the basis of perceived ex- as the goods being priced in the marketplace. pected returns net of these individually deter- This article provides a framework for analyzing mined costs, investors choose to hold differing the demand for capital market returns, which amounts of each asset. The cost of capital for an we define as the compensation each investor asset is the aggregation of all investors' capital requires for holding assets with various charac- costs on the margin, and represents the market teristics. A companion article will consider the expected return on the asset, supply of returns generated by the productivity Formal demand-side theories such as the of businesses in the real economy, outside the Capital Asset Pricing Model (CAPM) and Arbi- capital market.' The basics of the demand for capital market Roger Ibbotsojt is Senior Lecturer in Finance and returns can be explained in a few sentences. Director of the Center for Research in Security Prices at the Investors regard each asset as a bundle of char- Graduate School of Busine^ of the ilmvcrsny of Chicago . . ,- 1 • 1 I 1 • f /''/fff'V Diermeier is Vice President of the hirst Nntionat actenstics tor which they have various prefer- g,,,,^: ,,^ Qnaixo. Laurence Sieget is an Associate al R.G. 1. FomiHites appiMr ji Olid til .irtiirkv thbotson Associates. Inc., Chicago. FINANCIAL ANALYST5 JOURNAL ' JANUARY-FEBRUARY 1984 D 22
Transcript
By Roger G. Ibbotson, Jeffrey J. Diermeier and Laurence B. Siegel
The Demand f erCapital Market Returns:
A New Equilibrium TheeryInvestors demand more of an asset, the more desirable the asset's characteristics.The most important characteristic is its price, or expected return. By varying price,any and all assets become desirable enough for the capital market to clear.
Asset characteristics other than price include both risk and non-risk characteris-tics. The Capita! Asset Pricing Model and Arbitrage Pricing Theory have describedthe risk characteristics. The non-risk characteristics are not as well understood.They include taxation, marketability and information costs. For many assets, thesenon-risk characteristics affect price, or expected return, even more than the riskcharacteristics.
Investors regard asset characteristics as positive or negative costs, and investorsevaluate expected returns net of these costs. The New Equilibrium Theory (NET)framework applies to all assets—including stocks and bonds, real estate, venturecapital, durables and intangibles such as human capital—and incorporates all assetcharacteristics.
PRICES IN CAPITAL MARKETS are set by ences and aversions. Investors translate each
the interaction of demand and supply. characteristic into a cost, and require compensa-This relationship is commonly expressed tion in the form of expected returns for bearing
in terms of the "supply of and demand for these costs. Thus, although all investors arecapital." But viewing it from the opposite per- assumed to perceive the same before-cost'ex-spective—^that is, in terms of the demand for pected return for any given asset, each has'and supply of capital market rclurti:^—has the individually determined costs he must pay toadvantage of focusing our attention on returns hold that asset. On the basis of perceived ex-as the goods being priced in the marketplace. pected returns net of these individually deter-This article provides a framework for analyzing mined costs, investors choose to hold differingthe demand for capital market returns, which amounts of each asset. The cost of capital for anwe define as the compensation each investor asset is the aggregation of all investors' capitalrequires for holding assets with various charac- costs on the margin, and represents the marketteristics. A companion article will consider the expected return on the asset,supply of returns generated by the productivity Formal demand-side theories such as theof businesses in the real economy, outside the Capital Asset Pricing Model (CAPM) and Arbi-capital market.'
The basics of the demand for capital market Roger Ibbotsojt is Senior Lecturer in Finance andreturns can be explained in a few sentences. Director of the Center for Research in Security Prices at theInvestors regard each asset as a bundle of char- Graduate School of Busine^ of the ilmvcrsny of Chicago
. . ,- 1 • 1 I 1 • f /''/fff'V Diermeier is Vice President of the hirst Nntionatactenstics tor which they have various prefer- g,,,,̂ : ,,̂ Qnaixo. Laurence Sieget is an Associate al R.G.
1. FomiHites appiMr j i Olid til .irtiirkv thbotson Associates. Inc., Chicago.
FINANCIAL ANALYST5 JOURNAL ' JANUARY-FEBRUARY 1984 D 22
trage Pricing Theory (APT) have prescribed use-ful mathematical formulations for deriving as-sets' expected returns."^ Both these theories,however, assume perfect capital markets inwhich air-costs are due specifically to risk. TheCAPM specifies the payoff demanded by inves-tors for bearing one cost—beta, or market, risk;APT treats multiple risk factors. Other researchhas addressed non-risk factors, but in isolation,
Our framework, which we term New Equilib-rium Theory (NET), integrates costs arisingfrom all sources—^inciuding various risks, aswell as taxability, marketability and informationcosts—and affecting all assets in an investor'sopportunity set—stocks, bonds, real estate, hu-man capital, venture capital, tangibles and in-tangibles. NET theory does not provide a de-tailed analysis of each particular cost, nor does itspecify a mathematical asset pricing equation.The NET model is useful, however, in explain-ing observed investor behavior.
The New Equilibrium TheoryThe objective of the NET framework is to deter-mine the equilibrium cost of capital, r,, for eachasset i in the market, given the characteristics ofasset j and the utility functions of all the inves-tors in the market. Conceptually, the cost ofcapital is the sum of all capital costs at themargin across ail holders of all claims on asset j ;it is typically expressed as a per year percentageof value. This cost of capital can also be inter-preted as an expected return to investors or as adiscount rate used in valuation.
Unlike most models in finance, which dealspecifically with time and uncertainty, the NETmodel makes use of the simpler, classical instan-taneous supply and demand setting. The instan-taneous setting incorporates perceived uncer-tainties, investment horizon, taxation, entryand exit costs, and other factors that affect aninvestor's perception of an asset's cost, as decre-ments to the present value of the asset. Theinstantaneous setting is suitable as long as allcosts can be expressed as present values.'
The Supply of AssetsInvestors view all assets as bundles of charac-
teristics. The NET framework, in its most gener-al form, makes no assertion that there are com-plete markets for each characteristic. There isnot necessarily a separate market for each char-acteristic, nor any way that an investor couldconstruct his a priori optimum bundle of charac-
teristics, even if there were shadoiv prices forthese characteristics. Thus pricing characteris-tics are translated and aggregated into investorcosts at the individual investor level for eachasset, It is the asset, rather than each of itscharacteristics, that is being priced by the mar-ket.-*
NET assumes the existence of a characteristic-free asset, analogous to the risk-free asset inother models, earning the pure time value ofmoney. Investors can invest in, or lend out,unlimited quantities of this asset at the marketrate, r|. Investors are restricted in their borrow-ing, however, by the inclusion of an investor-specific borrowing cost function, cir, which risesas the amount borrowed increases. Thus inves-tors are divided into lenders and borrowers,with the net quantity of the characteristic-freeasset assumed equal to zero in the economy.'̂Borrowing costs are paid to intermediaries tocover their various costs.
Although the NET framework technically al-lows short selling, the cost of shorting a specificasset is ordinarily too high to make short sellingworthwhile. Short selling is thus restricted tothe characteristic-free asset, and even this bor-rowing is limited by increasing costs. The as-sumption of a characteristic-free return, r̂ , pro-vides a homogeneous benchmark for allinvestors, while the assumption that it is (in-creasingly) costly to short the characteristic-freeasset restricts the market impact of low-wealthinvestors who might otherwise borrow in infi-nite quantities.
The Investor Demand for AssetsTo focus on the composite market's cost of
capital for asset j , we assume that Investorshave homogeneous expectations concerning Vythe asset's expected retum before investors'costs, as well as r,v the rate of return on thecharacteristic-free asset/' Our second key as-sumption is that investors have heterogeneous,or individually determined, costs associatedwith the holding of asset j . These differing costsare a natural consequence of the fact that inves-tors differ in regard to wealth, risk aversion,access to information, tax bracket and numer-ous other traits. The individual investor mayevaluate an asset's characteristics according tohis own classification scheme, and he may mea-sure an asset's characteristics according to hisown judgment. Thus each investor will have hisown particular utility function, according to
FINANCIAL. ANALYSTS I O 1 . ' K \ : A L / [ A NU A RY-FEBRUARY 1984 Ql 2 3
Figure A INDIVIDUALSupply (S|) and Demand ^d^^) for an Asset (j) as seen by the Individual (i)Given the Homogeneous Before-Cost Expected Return (r.)
returns net ofcosts (investor
q (quantity)
which he translates all asset characteristics, in-cluding all risks, into costs.^
The costs associated with holding each incre-mental unit of asset j may fall at first because ofeconomies of scale in information, transactionor other costs. At some point, however, thecosts of holding additional units of asset j willbegin to rise. The major cause of this rise is theincreasing lack of diversification in the inves-tor's portfolio.
A Graphical Description of NETEquilibriumEigure A diagrams the individual's demand
and supply equilibrium. The expected return onasset j , viewed homogeneously by all investors,is denoted by the horizontal line rj. The expect-ed return on the characteristic-free asset is thehorizontal line rf. The latter also represents anopportunity cost curve, Cf, in this context. The
cost to investor i of holding asset j will be thesum of this opportunity cost plus all the othercosts he associates with holding the asset, Cy.̂This sum, Cf -i- Cjj, equals the individual's de-mand curve, dij, which is upward sloping be-cause it is presented in re turn-quantity space.^The investor stops buying the asset when hismarginal costs rise to equal the return on asset j ,rj. Thus the before-cost expected retum, rj,serves as the individual supply curve, Sj.
Alternatively, we can view the investor asdemanding a return net of his costs, rj - Cjj.This return net of costs may also be labeledinvestor surplus, analogous to the consumersurplus in the classical economic framework.The investor purchases asset j until his returnnet of costs equals the return on the characteris-tic-free asset, rf; that is, he invests in asset j untilrj minus Cjj equals rf.
The individual investor demand curve from
FINANCIAL ANALYSTS JOURNAL / JANUARY-FEBRUARY 1984 D 24
Figure B MARKETAggregation of Individual Demand Curves (d,|) to form Market Demand Curve (D.)and Determine Asset Price and Return (r.)
q (quantity)
Figure A, dy, is also displayed in Figure B. Here,individual demand curves, dy, are summedhorizontally to form the aggregate demandcurve for security j , Dj. The aggregate supply ofshares of security j is perfectly inelastic, and isrepresented by the vertical line, Sj (= Q,). Theintersection of D, and Sj forms the market re-turn, rj. This retum is the same as the individualsupply curve, rj, shown in Figure A.
Figure C restates Figure B in conventionalprice-quantity (rather than return-quantity)space. In Figure C, both the individual (djj) andaggregate (Dj) demand curves for asset j aredownward sloping. As in Figure B, the supplycurve (Sj) for asset j is vertical, The market price(Pj) and individual and aggregate quantities ofasset j demanded (qjj and Qj, respectively) aredetermined by the intersection of the supplyand demand curves.
Figure D shows what happens when theindividual investor is offered the opportunityeither to invest in (lend) or short (borrow) thecharacteristic-free asset. If he lends, he gets thehomogeneous return, r̂ , offered to all investors.If he borrows, he pays heterogeneous borrow-ing costs, Cjf, and consequently Borrows at raterf + Cjf. The individual investor's wealth con-straint, and the continuously rising cost func-tions depicted in Figures A and D, require thatthe investor either borrow or lend, which avoidsany corner solutions. The investor borrows hisoptimum quantity of the characteristic-free as-set; the cost of borrowing is included in his assetcosts in Figure A, so that c.j includes a constantCif for each asset j . ' "
Figure E depicts the aggregate demand curvefor the characteristic-free asset, Dt. This aggre-gate demand curve is the horizontal sum of
RNANCIAL ANALYSTS JOURNAL / JANUARY-FEBRUARY 1984 D 25
Figure C MARKETAggregation of Individual Demand Curves, Restated in Price-Quantity Space
q (quantity)
individual demand curves, djf. The net supplycurve for the characters tic-free asset is zero,and is depicted as Sf at a quantity (Qe) of zero.The characteristic-free asset earns rate r| in equi-librium, where the demand curve, D,, intersectswith the supply curve, S|.
Fach investor holds a different portfolio,based on his own particular heterogeneouscosts. Nevertheless, each investor is relativelydiversified, because he has calculated diversifia-ble risk as one of his costs. By holding a posi-tive, zero or negative position, each investor'smarginal cost equals his return for every asset inthe market. This means that each investorwould change his position in an asset in re-sponse to any change in the asset price—i.e.,before-cost expected return on the asset—or inresponse to any change in the aggregate costshe incurs by holding that asset. For a given
asset, all investors have equal costs on themargin, constituting the cost of capital for eachInvestor and for the market.'' Prices are thus setin aggregate by all investors. In sum, bothinvestors and issuers treat the asset characteris-tics as marginal investor costs that, in the aggre-gate, sum to the cost of capital.
The Role of Financial IntermediariesThe NET framework can readily be expanded
to include repackaging opportunities on thepart of issuing firms or financial intermediaries.The role of the financial intermediary is torepackage the pricing characteristics so as toreduce investor costs. One way intermediariesaccomplish their task is by making the marketsfor pricing characteristics more complete. Byunbundling asset characteristics, for example,they increase the likelihood that those investors
FINANCIAL ANALYSTS JOURNAL •JANUARY-FEBRUARY 19M LI 26
Figure D INDIVIDUALIndividual (i) Return On Lending or Borrowing the Characteristic-Free Asset
Shorting or Borrowing
r (return)
Lending or Investing
with lower costs for a particular characteristicwill hold that characteristic in their portfolios.Another way intermediaries reduce investorcosts is by optimal bundling of asset characteris-tics to take advantage of economies of scale.
Investors perceive financial intermediaries asadditional asset offerings, whereas issuers per-ceive them as additional investors. Assumingperfect competition, intermediaries act to maxi-mize aggregate investor surplus by minimizingthe sum of all investor costs (not just marginalcosts) across all assets for all the pricing charac-teristics.
The Pricing Characteristics of AssetsWe have developed a framework in which in-vestors view assets as bundles of characteristicsthat investors then translate into their ownheterogeneous costs. Thus far, these pricingcharacteristics—those attributes of an asset thataffect an investor's f.Yrt»/t'costs—have remainedin the background.'- As the opportunity set ofinvestor assets is worldwide in scope and in-cludes stocks, bonds, real estate, human capital,venture capital, tangible and intangible goods.
we may expect a wide range of pricing charac-teristics to exist. We delineate below some of themore important pricing characteristics, and sug-gest informally how they are bundled into fa-miliar assets, how they might be priced, whotheir investor clienteles are likely to be, and howfinancial intermediaries might reduce theircosts. We consider both risk and non-risk char-acteristics, the latter including taxation, market-ability and miscellaneous pricing factors.
Risk CharacteristicsThe CAPM states that only one risk pricing
characteristic exists—namely, market risk. APTprovides for multiple risk pricing characteris-tics, and treats each risk as orthogonal to all ofthe others, so that the market payoffs are addi-tive. The NET framework does not directly takesides in this controversy, but does allow formultiple pricing characteristics. '̂ We focus hereon four of the most intuitive types of risk—beta(market), inflation, real interest rate and residu-al risk.
Market, or beta, risk is the risk that the return ofan asset will fluctuate with the market portfo-
FINANCIAL ANALYSTS JOURNAL / JANUARY-FEBRUARY 1984 D 2 7
Figure E MARKETAggregate Demand (D,) and Supply (S,) Curves for a Characteristic-Free AssetWith Individual Demand Curves (d̂ ,) for a Borrower and a Lender
lio's return. According to CAPM, beta risk is theonly risk that affects expected return, it is as-sumed that the rational investor will diversifyaway {at no cost) all other risks. In the NETframework, as noted, each investor translatesrisks into costs by assigning a price at which heis indifferent between buying and not buyingmore of the risk.
Inflation risk is the risk that an asset's realvalue will fluctuate because of unanticipatedchanges in the inflation rate. This risk is bestexemplified by a long-term government bond,which is relatively free of mcist other pricingcharacteristics. The bond is a nominal contract,and its yield to maturity consists of three com-ponents—the expected inflation rate, the ex-pected real interest rate and the risk premium (ifany) associated with inflation and real interestrates. Although the market anticipates all threecomponents over the bond's life, unanticipated
changes in current and expected inflation ratescause variations in the bond's real return.
Inflation risk arises when one side explicitlyor implicitly contracts in nominal, instead ofreal, terms. For this pricing characteristic to benonzero, at least one side must have negativeinflation risk costs and be willing to pay theother side to create these risks.'"* The inflationrisk premium may be positive for investors inthe stock market and for holders of short-term,and possibly long-term, bonds.'"^ Other assetslikely to contain a nonzero amount of inflationrisk include real estate, gold and any otherassets whose real returns are correlated (posi-tively or negatively) with unanticipated changesin the inflation rate.
The real interest rate is the difference betweenthe instantaneous nominal interest rate (on acharacteristic-free bond) and the instantaneousinflation rate. Since real interest rate changes
FINANCL^L ANALYSTS JOURNAL / JANUARY-FEBRUARY 1984 O 2 8
are unanticipated, the investor who rolls over aseries of short-term bonds receives an uncertainreturn in real terms. The investor in long-termbonds can lock in the real rate over the bond'slife, but incurs inflation risk in the process. It is,of course, possible to construct a long-termcontract in real terms and avoid both inflationand real interest rate risk for any given timehorizon. Since these contracts are not common-ly seen, we have to presume either that issuersand investors have differing time horizons, orthat issuers and/or investors believe that thecosts of writing contracts in nominal terms,including the costs of inflation and real interestrate risk, are less than the costs of writingcontracts in real terms. '̂
Residual risk is the risk resulting from lack ofdiversification in a portfolio. Assuming that therisks already described account for an asset'sundiversifiable risk, residual risk is the oneremaining risk factor. We propose that residualrisk, like the other risk factors, may be an ex antepricing factor.
in CAPM, the rational investor perfectly di-versifies so as to eliminate entirely all residualrisk. NET assumes that it is costly to diversify.The factors that make perfect diversificationeither impossible or suboptimal are related tonon-risk pricing characteristics. For example,many investors wish to own their residencesOutright. The large unit size of other real estateinvestments, along with the high cost of creat-ing divisibility mechanisms such as condomini-ums and limited partnerships, imposes highcosts on investors seeking diversification. Thusmost investors do not hold a diversified realestate portfolio—that is, one that is spread overvarious geographical locations and types of landand structures.
Human capital is subject to even more ex-treme constraints on diversification. Once ac-quired, human capital cannot readily be sold,and is usually rented out for wages in the labormarket. It follows that one cannot easily buy aportion of another person's human capital inorder to diversify within the asset class.
TaxabilityTaxability often has a substantial impact on an
asset's cost of capital, The taxability characteris-tic is inherently complex because of the intrica-cies of the U.S. (and other countries') taxationsystems. This complexity consists of (1) thestepwise ("tax bracket") and multiplicative attri-
butes of the tax function; (2) the fact that taxeson a given asset are contingent on the perform-ance (effect on income) of other assets in one'sportfolio; (3) the differential treatment of ordi-nary income and capital gains; (4) special taxlaws, such as those allowing depreciation muchfaster than the useful life of certain assets; and(5) multiple taxing authorities. These attributescause the tax costs for the same asset to differacross individuals. The general principle is thathighly taxable assets are lower priced—i.e.,have a higher before-tax expected return—thanless highly taxed assets.
For example, municipal bonds, whose cou-pons are free of U.S. federal income taxes, yield20 to 50 per cent less than fully taxable corporatebonds of comparable risk. A similar relationshiphas been suggested for high dividend versuslow dividend stocks.'^ Constantinides providesa personal tax equilibrium that includes thetiming option for the realization of capital lossesand the deferral of capital gains.'" Most of theseand other tax-related theoretical results can beintroduced into the general NET framework,because NET does not specify actual investorcosts.
Real estate, venture capital, hedging portfo-lios and leasing arrangements provide specialopportunities for financial intermediaries toseparate out tax characteristics and repackagethem for the appropriate clienteles. After re-packaging, many investments may be tax shel-ters having negative tax rates.'"^
In summary, an asset may generate taxes(positive or negative) on income, expenses orcapital appreciation. The investor includes thesetax costs in his pricing process. The complexityof the taxation system and the interaction oftaxes with other pricing characteristics make itdifficult to specify this pricing characteristic.Nevertheless, the magnitude of taxes is suffi-ciently large that it must be included in anyexposition of the NET framework.
Marketability CostsWe group all the entry and exit costs associat-
ed with buying or selling an asset into thecategory of marketability costs. The NET frame-work is instantaneous, so that it provides nodescription of how an investor came to hold hisparticular portfolio or when or how he mayrebalance his portfolio. For the NET equilibriumto be descriptive, each investor must reduce thevalue of his assets by a present value amount to
FINANCIAL ANALYSTS JOURNAL /JANUARY-FEBRUARY 1984 D 2 9
cover these costs."" These marketability costsinclude information, search and transaction,and divisibility costs.
Information costs are the costs that an investormust pay to learn the value of an asset. Sincethe NET model assumes homogeneous expecta-tions, we have already in some sense assumedthese costs away. Nevertheless, we can infor-mally apply the NET model by suggesting thatinvestors must pay some costs to learn what thehomogeneous expectations are."' In such aworld, investors with comparatively lower in-formation costs for a particular asset would tendto own that asset. For example, U.S. investorsown stocks and bonds of U.S. corporations indisproportionately large quantities because ofthe cost of acquiring information across nationalboundaries. Moreover, assets that are difficultto learn about, such as stocks of small or newcompanies, should have higher before-cost ex-pected returns than assets that are easier tolearn about, such as large company stocks.Finally, information costs tend to favor the largeinvestor, since there are economies of scale ininformation use.
Search and Iran^action cost'^ include the costs oflooking for the other side of the transaction, aswell as the costs of actually closing the transac-tion. The costs may include the bid-ask spread,the waiting time beyond the investor's desiredhorizon, the possibility of having to take a priceconcession, the paperwork and legal costs ac-companying a transaction, the cost of advertis-ing or other efforts to locate the other party tothe transaction, and the cost of any brokers oragents used to effect the transaction. Thesecosts are treated in search and bargaining the-ory literature. In the NET framework, thesecosts are merely estimated by the investor astheir present value equivalent costs.
Divisibility costs arise from the large and dis-crete scale of some investments, such as realestate, venture capital, large-denomination cer-tificates of indebtedness, and certain discretehuman capital decisions. Divisibility interactswith many of the other pricing characteristics.Indivisibility's chief burden to investors may bethat it forces them to take substantial residualrisk. It also causes some investors to hold asuboptimal quantity of a particular in vestment.-"
Human capital, once acquired, is often con-sidered nonmarketable as well as indivisible. Itcan be rented and, to some extent, it can be putup as collateral for loans. When invested in a
business, portions of it can sometimes be sold.In the NET framework, we can regard these ashigh, but not insurmountable, divisibility costs.In some models, an equilibrium is arrived at inwhich human capital is literally treated as non-marketable."'
One of the principal roles of financial interme-diaries is to repackage securities in such a wayas to reduce divisibility costs. A saver (smalllender) would have great difficulty in finding aborrower with whom to transact and still main-tain the liquidity of his savings. By pooling thesavings of many persons, a bank can do exactlythat. Money market funds reduce the minimuminvestment amount for cash instruments from510,000 to very little. Real estate investmenttrusts and limited partnerships lower the sizebarrier for investing in large properties from therange of millions of dollars to the range ofthousands or less. Each of these mechanisms forreducing divisibility costs is itself costly. Formany investors, however, paying the costs ofinvesting through a financial intermediary in-creases their investor surplus.
Other miscellaneous factors may affect the priceof a capital market asset. These include nonpe-cuniary costs or benefits, all of which we wouldtreat as positive or negative costs. In addition,certain expenses such as management, mainte-nance and storage costs are best treated as costsof capital, rather than as decrements to cashflow. This is because they differ across inves-tors. Because investors seek to maximize re-turns net of all costs and benefits, these factorsshould be included in the set of NET pricingfactors.
Application of the NET FrameworkIn NET, we go back to economic basics. Weapply the classical supply and demand model tothe pricing of assets. Individual investors havehomogeneous expected returns but heteroge-neous costs associated with multiple risks, vari-ous forms of market imperfections and otherpricing characteristics. These characteristics arepriced at the individual level and treated aspresent value equivalents in the instantaneoustime setting. This simple equilibrium frame-work cannot provide a pricing equation. It can,however, describe investor portfolios and assetexpected returns in broad realistic terms.
Each individual investor maximizes his owninvestor surplus. He does so by investing ineach asset until his own unique costs equal the
FINANCIAL ANALYSTS JOURNAL / JANUARY-FFRRL'ARY 1984 P 3 0
asset's>expected return on the margin. Thus noone holds the market portfolio. Rather, eachinvestor selects his asset holdings according tohis comparative cost advantage. Clienteles arisebecause groups of investors have similar costs.
Assets are treated as bundles of pricing char-acteristics. The cost of capital for a given asset isthe sum of the costs of all its characteristics andis equal to the asset's expected return. Givencontinuous cost curves, all investors face thesame cost on the margin for a given asset.However, the cost for each characteristic of theasset is not the same for each investor.
Financial intermediaries act to maximize the8̂8̂ *̂ 83*6 SLini of investor surplus across all
investors by repackaging assets to take advan-tage of pricing characteristics that are less costlyto some investors. They bundle and unbundlecharacteristics to move toward complete mar-kets and produce investor economies of scale.
The NET framework can readily be adapted toinclude heterogeneous expectations. In fact, do-ing so may shed further light on the mecha-nisms by which Investors price assets. We havekept the homogeneous expectation assumptionso that we could relate expected returns to thecost of capital.
As a practical matter, it may be useful to makeadditional assumptions in order to estimate themarket cost of capital for an asset. If we assume
Table I The Pricing Characteristics of Assets
that there exists a "representative" investorwhose cost functions mimic those of all inves-tors in the aggregate, and who perceives allcharacteristics except tax costs as orthogonal,we can sum the characteristic costs for therepresentative investor to obtain the after-taxmarket cost of capital. We may assume that thetax cost, unlike other costs, is multiplicative;thus we divide the after-tax cost of capital byone minus the representative investor's tax rateto arrive at the before-tax cost of capital.
Table 1 presents a heuristic summary of select-ed assets matched with pricing characteristics.The investor must make his own judgmentsabout the quantity of each characteristic embed-ded in each asset and about the costs he asso-ciates with the characteristic. Although NETcannot make these judgments for the investor,NET provides a framework within which aninvestor can analyze the wide range of assetsand characteristics available.
NET as it presently stands is a characteriza-tion of the way investment practitioners viewthe world, to some extent explicitly and in largepart implicitly or intuitively. NET theory servesto draw all pricing factors into a unified frame-work. Investors need to know what affectspricing, with or without an algebraic pricingequation. Consideration of all pricing character-istics may be more important to them than a
Note: Low, medium, high, etc. retcr to positive coefficients unless indicated to be negativt',*Finandal intermediaries are likely to be important in reducing these costs.
FINANCIAL ANALYSTS JOURNAL ' JANUARY-FEBRUARY 1484 G 31
rigorous partial equilibrium treatment of one ora few characteristics. Our hope is that the inte-grated view of asset characteristics and investorcosts presented in this article will be given moreexplicit consideration by both practitioners andresearchers. •
Footnotes1. Jeffrey |. Diermeier, Roger G. Ibbotson and Lau-
rence B. Siegel, "The Supply of Capital MarketReturns," Financial Analysts loumal, forthcoming,1984.
2. William F. Sharpe, "Capital Asset Prices: A The-ory of Market Equilibrium Under Conditions ofRisk," foumal of Finance, September 1974, pp.425-552; John Lintner, "The Valuation of RiskAssets and The Selection of Risky Investments inStock Portfolios and Capital Budgets," Revieio ofEconomics and Statistics, February 1965, pp. 13-37;and Stephen A. Ross, "The Arbitrage PricingTheory of Capital Asset Pricing," journal of Eco-nomic Theory, December 1976, pp. 341-360.
3. Since there is only one instant of time in themodel (the present), the investor is assumed toselect his optimal portfolio in a tatonnementprocess, given the prices (before-cost expectedreturns) of all the assets in the economy. ThetStonnement process is described in Alfred Mar-shall, Principles of Economics, 8th edition (London:Macmiilanand Co., 1920).
Although the Law of One Price holds for eachasset in the economy, other market imperfectionsare expressed as some of the pricing characteris-tics.
4. This avoids many aggregation issues presentedin John Lintner, "The Aggregation of Investors'Diverse Judgments of Preferences in Purely Com-petitive Markets," journal of Financial ami Quanti-tative Analysis, November 1969; Mark Rubinstein,"An Aggregation Theorem for Securities Mar-kets," journal of Financial Economics, September1974; and Sherwin Rosen, "Hedonic Prices andImplicit Markets: Product Differentiation in PureCompetition," journal of Political Economy, Jan-uary/February 1974.
5. The net quantity of the characteristic-free assetmay be assumed to be other than zero withoutaffecting the analysis. For example, an exogenousgovernment may issue this asset in some quanti-ty, similar to the money supply.
6. As we shall see later, the NET model is robustwith respect to forming an equilibrium (i.e., asingle market price for an asset) given heteroge-neous expectations. The concept of cost of capi-tal, however, has very little meaning when relat-ed to the diverse expectations of investors.Heterogeneous expectations are better treated in
the specific context of information costs and indebating market efficiency, rather than herewhere we seek to determine an equilibrium costof capital.
7. To take the most familiar example, that of map-ping beta risk into a cost in the CAPM context,each investor's cost function rises as he takes onincremental beta risk. The investor then assigns aprice at which he is indifferent between buyingand not buying more of the risk. Given completemarkets and homogeneous expectations, themarket provides one clearing price for beta risk.Each investor buys beta risk until his increasingrisk costs equal the market risk payoff. The mildlyrisk-averse investor has a lower cost for a givenamount of beta risk than the highly risk-averseinvestor, and therefore holds a higher beta port-folio.
8. These costs, Cj,, include the borrowing costs, Cjf, ifthe investor is a borrower. Borrowing costs aredescribed in Figures D and E.
9. In Figure A, return is on the vertical axis andquantity on the horizontal axis. Since return isrelated to the reciprocal of price, supply anddemand curves are "upside down" relative to themore familiar price-quantity diagrams. This re-turn-quantity space was made familiar by MertonH. Miller in "Debt and Taxes," journal of Finance,May 1977. '
10. In the instantaneous framework presented here,the investor finds his optimum borrowing quanti-ty of the characteristic-free asset and makes hischoice of other assets (based on their returns netof perceived costs) simultaneously, using thetitonnement process referred to in footnote 3.
11. In practice, the utility function that translatescharacteristics to costs may generate a discontin-uous cost curve because of indivisibility of assetsor restrictions on short selling. Thus, for manyassets, investors will hold only zero or wholenumber positions. For these assets, there mayexist heterogeneous costs on the margin.
12. Although we have noted that NET does notpostulate complete markets for these characteris-tics, which are priced at the individual level, wemight have alternatively assumed that completemarkets existed for each pricing characteristic, sothat the characteristics could be priced at themarket level. The assets would then also bepriced at the market level and represented ascombinations of characteristics. We deliberatelyavoid making this assumption, because many ofthe characteristics we are about to discuss areinseparable, complex, and contradict the com-plete markets assumption by their very nature.
13. Nai-fu Chen, Richard Roll and Stephen A. Ross,in "Stock Markets, Factors, and the Macroecon-omy" (working paper, 1982) have isolated fiverisk factors—(1) beta risk, (2) the rehjm differen-
tial between low and high grade bonds (related tothe return differential between small and largecapitalization stocks), (3) the return differential 18between short and long-term bonds, (4) anticipat-ed inflation, and (5) unanticipated inflation.
14. Jeffrey F. Jaffe, in "Corporate Taxes, inflation, the 19Rate of Interest, and the Return to Equity,"Journal of Financial and Quantitative Analysis,March 1978, shows that inflation risk is notreadily separable from taxation costs. Thus it maybe impossible to contract totally in real terms.
15. Charles R. Nelson, in "Inflation and Rates ofReturn on Common Stocks," journal of Finance, 20May 1976, and Eugene F. Fama and G. WilliamSchwert, in "Asset Returns and Inflation," jour-nal of Financial Economics, November 1977, pres-ent results suggesting a positive inflation-riskpremium for stocks. J. Huston McColloch, in "AnEstimate of the Liquidity Premium," journal ofPolitical Economy. January/February 1975, shows 21,that the inflation-risk premium is positive tobondholders.
16. in high-inflation countries such as Brazil andIsrael, bonds with price-levei-indexed principalhave existed for years. The United Kingdom andFrance now issue them also. Bertrand Jacquillatand Richard Roil, in "French Index-Linked Bondsfor U.S. Investors?" journal of Portfolio Manage- 22.ment, Spring 1979, have found French price-level-indexed bonds to have high historical re-turns to a U.S. investor.
17. See Robert H, Litzenberger and Krishna Ramas-wamy, "The Effect of Personal Taxes on Divi- 23.dends on Capital Asset Prices," journal of Finan-cial Economics. June 1979. Both their theory andtheir empirical results are disputed by MertonMiller and Myron S. Scholes in "Dividends and
Taxes," journal of Financial Economics, E)ecember1978.George Constantinides, "Capital Market Equilib-rium With Personal Tax," Econometrica, May1983.Merton H. Miller ("Debt and Taxes," op cit.)shows that the taxation issue cannot be fullyanalyzed on the investor demand side alone.Corporations (like intermediaries) may react toany pricing characteristic by attempting to offerthe type of issues that have the lowest costs in themarket.A problem with applying the NET framework tomarketability costs is determining the investor'sstarting portfolio. For example, if the investorholds all cash, then he must pay entry costs oneach asset to obtain his optimum portfolio. Onthe other hand, the investor may already hold hisoptimum portfolio, so that he has no entry costs.The NET framework cannot directly include re-sults from the variety of information modelsfound in the financial literature. To apply NEThere, we have to assert that the investor knows inadvance both that he will use the information tobuy the appropriate quantity of the asset, andthat the purchase of the information is worth-while.The NET framework as presented in Figure Auses continuous co.st curves, c,,. When divisibilityis a problem, cost curves will be discontinuous,and it will no longer be true that the marginal costof capital of an asset is the same for all investors.See for example, David Mayers, "Non-Market-able Assets and Capital Market Equilibrium Un-der Uncertainty," in Michael C. Jensen, ed..Studies tn the Theory of Capital Markets (New York;Praeger Publishing, 1971).
FINANCIAL ANALYSTS JOURNAL / JANUARY-FEBRUARY 19W D 3 3
