functional spin off

Functional Spin-Off: A Key toAnticipating Change inDistribution StructureBRUCE MALLEN

Building on the concepts of marketingfunctions and division of labor, this article

develops an approach to aid channeldesigners in anticipating changes in distrib-ution structure before such changes develop

into obvious trends. The author suggeststhat an understanding of how and why

producers spin off marketing functions willin turn lead to an understanding of change

in all dimensions of channel structure.

• ABOUT THE AUTHOR.

Bruce Mallen is chairman of Graduate Studies in theFaculty of Commerce and Administration, Sir GeorgeWilliams University, Montreal, Canada.

Journal of Marketing, Vol. 37 (July 1973), pp. 18-25.

T HE purpose of this article is to suggest toanyone interested in understanding distribu-

tion channel change an approach whereby thechannel designer may anticipate distributionchange in his industry before such change hasdeveloped into an obvious trend. The channel de-signer would then be in a position to incorporatethis information in planning his distributionstrategy and in adapting to his distribution en-vironment.

To successfully complete his task, the channeldesigner must closely analyze five factors: ^

1. The selected target markets2. The rest of his marketing mix: price, prod-

uct, promotion, physical distribution, etc.3. His company's resources4. Competition and other external forces5. Current and anticipated distribution struc-

tures in his industry

Perhaps the most difficult of these factors toanalyze is the future changes in distributionstructure—those that have not yet developed intoobvious trends. Typically, the channel designermust limit himself to reading futuristic type ar-ticles on distribution trends^ and/or to surveyinga cross-section of opinions in his industry. Theproblem with the first information source is thatsuch articles are usually too general to be ofdirect benefit to the reader and are extrapola-tions of current obvious trends rather than an-ticipations of changes which have not yet de-veloped into trends. The problem with an opinionstudy is that the consensus may be completelywrong—nothing more than the reflection of acommon pool of ignorance.

This article does not attempt to provide a com-prehensive explanation of distribution structure

1. For a detailed description of the selection process,see Bruce Mallen. "Selecting Channels of Distribution forConsumer Products," in Handbook of Modern Marketing,Victor P. Bueli. ed. (New York: McGraw-Hill, 1970), pp.4-15 to 4-30.

2. See, for example, William R. Davidson, "Changes inDistribution Institutions," JOURNAL OF MARKETING, Vol, 34(January 1970), pp. 7-10; and Philip B. Schary, "ChangingAspects of Channel Structure in America." British Jour-nal of Marketing, Vol. 4 (Autumn 1970), pp. 133-147.

18

Functional Spin-Off: A Key to Anticipating Change in Distribution Structure 19

based on empirical research. Rather it presentsa sequence of relationships which can be usedto aid in anticipating change.

The Conceptual Approach

For approximately 60 years, economic and mar-keting scholars have recognized the concept ofmarketing functions and have related them in amore and more exact fashion to the determina-tion of channel structure. Early contributionswere made by Butler. Shaw, Weld, Cherington,Clark, Breyer, and Converse. More recently Stig-ler, Vaile, Grether, Cox, Alderson, and Bucklinhave been major contributors to functionalismas it relates to channel structure.^

The basic message of all channel functionalistsis as follows:

1. Marketing functions are the various typesof job tasks which channel members under-take.

2. These functions can be allocated in differentmixes to different channel members.

3. The functional mixes will be patterned in away which provides the greatest profit eitherto the consumer (in the form of lower pricesand/or more convenience) or the channelmembers with the most power {which de-pends on market structure).

4. Should one or more channel members (orpotential members) see an opportunity tochange the functional mix of the channel inorder to increase his profits, he will attemptto do so.

5. Should the attempt be successful, and ifthe functional mix change is big enough, itwill (by definition) change the institutional

3. Ralph S. Butler, Selling and Buying, Part II. Ad-vertising, Selling and Credits of Modern Business. Vol. IX(New York: Alexander Hamilton Institute, 1911), pp. 276-277; Ralph S. Butler, H. F. Debower, and J. G. Jones, Mod-ern Business, Vol. III. Marketing Methods and Sales-manship (New York: Alexander Hamilton Institute. 1914),pp. 8-9; Arch W. Shaw, Some Problems In MarketDistribution (Cambridge. Mass.: Har\'ard UniversityPress, 1915). pp. 4-28; L. D. Weld. "Marketing Functionsand Mercantile Organization." American Economic Re-view, Vol. 7 (June 1917). pp. 306-318; Paul T. Cherington.Elements of Marketing (New York: MacMillan, 1920), pp.44, 56-59; Fred E. Clark, Principles of Marketing (NewYork: Macmillan, 1922); Ralph F. Breyer. The MarketingInstitution (New York: McGraw-Hill, 1934); P. D. Con-verse, Essentials of Distribution (New York: Prentice-Hall, 1936); George J. Stigler. "The Division of Labor isLimited by the Extent of the Market," Journal of PoliticalEconomy. Vol. 54 (June 1951). pp. 185-193; R. S. Vaile. E. T.Grether, and R. Cox. Marketing in the American Economy(New York: Ronald Press, 1952), pp. 121-133; Wroe Aider-son, Marketing Behavior and Executive Action (Home-wood. 111.: Richard D. Irwin. 1957); Louis P. Bucklin. ATheory of Distribution Channel Structure (Berkeley,Calif.: Institute of Business and Economic Research, Uni-versity of California, 1966).

arrangement in the channel, i.e., the channelstructure.

Thus, the channel functionalist attempts to an-swer two basic questions: What is the most effi-cient functional mix in a given situation, andhow will this functional mix affect the channelstructure?

There are four dimensions of distribution struc-ture in which change can be anticipated:

1. The number of channel levels2. The number of channels or whether one,

two (dual), or more (multi) channel typeswill be used

3. The types of middlemen that will evolve4. The number of middlemen that will develop

at each level

Although the goal to attain market power andto manipulate demand is an important considera-tion in understanding structural change, the drivefor efficiency is also of primary importance. Thefundamental premise of this paper is that givena specific level of demand, firms will try to maxi-mize profits by designing or selecting a channelwhich will generate the lowest total average costsfor their organizations.

This drive for efficiency and its anticipatedeffects on the four dimensions of channel struc-ture can be evaluated through the concept of"functional spin-off."

The Basic Concept of Functional Spin-off

The basis for the functional spin-off conceptis a 1951 article by Stigler.^ In this article, Stigierprovides a most important conceptual frameworkfor measuring and anticipating channel struc-tural arrangements. His approach to isolating thereasons why firms will "subcontract" some func-tions is to analyze or break down the averagetotal cost curve of the firm by function ratherthan by the normal category of expense calcu-lations such as salaries and interest. Includedwould be costs associated with functions such asownership, promotion, information gathering,risk taking, negotiation, and so on. Each functionwill then have its own cost curve, and the sumof the cost curves for each function will be thetotal average cost curve of the firm.

These functional cost cur\^es will have variousshapes, and each may differ from the other tosome degree. Average cost curves for some func-tions will increase with increasing volume where-as others will decrease with increasing volume.The average cost curves for some functions willassume a U-shaped design: they decrease withincreasing volume and at some point start to in-crease with increasing volume. (Stigler assumes

4. Stigler, same reference as footnote 3.

20 Journal of Marketing, July, 1973

a U-shaped design for the total average costcurve.)

Functional spin-off and the number of channellevels. It is economically beneficial to spin offto marketing specialists those distributive func-tions which have a decreasing curve as volumeincreases when the firm has a relatively smallvolume. When a firm enters or creates a newmarket, it typically produces a small volume inthat market. Assuming the middleman specialistfaces the same cost curve as the producer, theindividual producer at this low volume will havea higher average cost for performing a functionwith a decreasing cost curve than the specialistwho can combine the volumes of a number ofproducers and thus benefit from the economiesthat the performance of this particular functiongenerates at higher volumes. If the middlemanspecialist passes on all or some of the lowercosts, the producer's total average cost will de-cline as a result of this spin-off of the distribu-tive function. In effect, the middleman has gen-erated the basic raison d'itre for his own exist-ence by providing external economies to pro-ducer firms.

Although the falling functional cost curve isof most interest because it is probably the mostcommon situation, the reverse curve also hasimplications for channel structure. With a risingfunctional cost curve, it would make sense eco-nomically to spin off certain functions to smallspecialists when the firm has achieved a highvolume. These small specialists can perform therising cost function at lower costs if they staysmall and do not combine the volumes of toomany producers. If they are competitive ( a morelikely event than in the falling cost functionalcurve situation) such savings will be passed onto the producer.

Some qualifications. At the beginning of thedevelopment of a new market, there may not beenough volume for a middleman to enter themarket since there may not be enough producersfrom which the middleman can draw suppliesto create the large volume required for a profit-able operation. In this case, the producer willnot have any middleman to whom he may spinoff a high cost function. The situation may alsooccur during the declining stage of the productlife cycle when industry volume has decreasedto a point where insufficient total sales exist tojustify a functional specialist. In the second case,vertical reintegration becomes necessary as mid-dlemen leave the market.

As the market develops, more producers enter,industry volume increases, and it becomes viablefor middlemen to operate. It is possible thatwith even greater volume, a given spin-off func-tion will in turn be broken down into severalsubfunctions, some of which may be spun off

by existing middlemen to even narrower spe-cialists; for example, import distributors mightspin off certain types of selling to domestic whole-salers. Thus several levels and other types ofmiddlemen may be added to the structural ar-rangement.

It should be noted that even if the middlemanis a monopolist, he cannot exploit his situationcompletely. A producer will distribute directlyif a middleman attempts to charge more forhis services than the producer would have topay with direct distribution. In other words, themiddleman faces an elastic demand curve for hisservices. At most, he can take all of the efficien-cies that he provides in the form of his ownprofits, but no more. The middleman monopolistis more likely to be present at the beginning ofa new market situation, where volume does notwarrant the entry of competitive middlemen.Eventually, however, with increasing volume andno artificial barriers to entry, competitive middle-men will enter the market.

It should also be noted that functions are notindependent but are interrelated. Therefore, thespin-off of one function could have repercussions,up or down, for the cost of one or more otherfunctions. For example, coordination costs mayfall with the spin-off of a function. That is, ifa given function is not being performed in acompany, there is no longer the need for internalcompany communication between the people thatwould have been performing the function and therest of the company.

Extensions of the Functional Spin-off Concept asRelated to Channel Levels

U- or L-Shaped average cost function. Thereare a number of important implications for in-dustrial structure and, more specifically, channelstructure on which Stigler did not elaborate. Per-haps the most important are the implicationsarising from a functional average cost curve whichinitially declines and then at some point startsto increase (really a U-shaped curve) or evenfiattens out.

If the cost curve does not continue to fall, ata high level of volume a point will be reached atwhich that producer can retake the functionwithout losing economies. For example, in Fig-ure 1 at volumes up to Ql it will pay the pro-ducer to spin off to middlemen the functionshown by the cost curve. Between Ql and Q2,performance by the producer or spinning it offwill provide the same economies. After volumeQ2, however, it would be beneficial for the pro-ducer to resume performance of the function(unless middlemen split themselves up and formsmaller firms or smaller middlemen are availablein the market, so that a very large volume can

hunctional Spin-OfF: A Key to Anticipating Change in Distribution Structure 21

Cost

Average cost curveof a given function

Q Quantity

FIGURE L A U-shaped average cost curve for a given function faced by an individualproducer or marketing middleman.

be distributed among them). The resumptionprocess can apply to any function. For example,the resumption of the ownership function re-moves the task from a merchant middlemanand results in the producer selling direct to themarket; self-performance of the advertising func-tion means a producer will not use an advertisingagency but relies completely on his own person-nel, and so on.

Functions with continually decreasing averagecost curves. Another extension of the Stiglermodel is that if the functional cost curve con-tinually falls, the middleman industry, and per-haps individual middleman firms which resultfrom this situation, will become bigger andbigger.

Stigler's model can also be reconstructed intoa dynamic rather than a static one. If the chan-nel's costs fall, prices fall; so that given anelastic demand curve, volume will increase. Inother words, the market situation propels itselffurther along the various functional cost curves,with the implications depending on the shapesof these curves. For falling functional cost curves,another round of functional spin-offs, falling costsand prices, and increasing volume will take place.An implication of this last situation is that themiddleman industry would become very large,perhaps creating extremely large firms. Thisprocess may have facilitated the rise of the massmerchandiser.

Extensions of the Functional Spin-off Concept asRelated to Other Structure Dimensions

Multi-channel structures. The concepts em-ployed here can also be useful in explaining therationale behind dual-channel or multi-channel

distribution systems. If the functional cost curvesare analyzed by large retailer versus small re-tailer markets, it is possible for the same func-tion to have different shapes in each market.For example, for the small retailer market theproducer's functional cost curve at a given quan-tity may fall with increasing volume; whereasfor the large retailer market the cost curve forthat same function at the same given quantitymay be flat or even increase with volume. Thiswould occur if there were few economies as-sociated with increasing volume in marketing tobig retailers once any reasonable amount wassold to them; e.g., the selling effort and cost perunit which is required to sell X units to bigretailers is the same as to sell 2X units.

If the above situation held true in a given case,it would be economically beneficial for the pro-ducer to spin off to a middleman the particularfunction involved in selling to the small retailermarket and to sell directly to the large retailermarket. Even if the shapes of the cost curvesin each market were identical, say declining andthen leveling off, the spin-off in the small re-tailer market would be beneficial if the quantitybeing sold was still small enough to be on thedeclining portion of the curve. Of course, if thesituation was reversed, i.e., selling the smallquantity to the large retailer market and large(flat portion of curve) quantity to the small re-tailer market, the spin-off would take place inthe large retailer market. This would lead toindirect distribution to large retailers and directdistribution to small retailers.

Figure 2 portrays the possible situations. Qs(1 or 2) is assumed to be the quantity sold tothe small retailer market and Qb (1 or 2) is the

22 Journal of Marketing, July, 1973

Cost

Average cost curve of a given functionin marketing to large retailers

Average cost curve of agiven function in marketing

to small retailers

Qs Quantity

FIGURE 2. L-shaped average cost curves faced by individual producers or marketingmiddlemen in selling to a large retailer market and a small retailer market for the

same given marketing function.

quantity assumed to be sold to the large retailermarket by a given producer. In the cost curvesshown in this figure, the reason for selling di-rectly to retailers (large or small) at high volumesis shown, i.e., leveling cost curves at Qs2 or Qb2;as well as the reason for selling indirectly, i.e.,falling cost curves at Osl or Obi.

Assume four possible volume situations:a. Qsl and Qb2b. Qs2 and Qblc. Qs2 and Ob2d. Qsl and Qbl.In the first situation, if the producer is selling

Qsl and Qb2 it can be seen that lower averagecosts for a given marketing function to smallretailers can be obtained at higher volumes (e.g.,Qs2). Hence, it makes economic sense for theproducer to spin off the function of selling tosmall retailers to middlemen who, by combiningthe volume of two or more producers, can achievethese economies and, if competitive, pass themon to the producer. At Qb2 it is obvious thatno further economies (or diseconomies) are avail-able at this and higher volumes to the firm.Therefore, it would not be reasonable to spinoff the function to middlemen in marketing tolarge retailers at that point.

In another situation, if a producer is sellingQbl and Qs2 the reverse channel structure wouldresult. The function would be spun off to middle-men in selling to large retailers and would beretained on a direct basis for small retailers.

Should the relevant volumes be Qb2 and Qs2,there would not be any multi-channels, i.e., therewould be only direct distribution. Should therelevant quantities be Qbl and Qsl, there wouldbe spin-offs in both cases; and should the mid-dlemen be the same firms in both cases, therewould not be any multi-channel distribution, butsimply indirect single-channel distribution. If,on the other hand, distribution to the large re-tailer market and the small retailer market re-quired different types of middlemen, there wouldstill be multi-channel distribution although of adifferent nature.

Middlemen types. The types of marketing in-termediaries that are created will be directly de-termined by the mix of functions spun off tothem, inasmuch as part of the definition of amiddleman depends on the functions he per-forms. For example, if part of the ownershipfunction is spun off, then merchants are created;if a negotiation function is spun off, then agentsare created; if advertising only, then advertisingagencies; if marketing research only, then mar-keting research agencies; and so on.

Horizontal channel structure. Up to this pointthe discussion has dealt with only three of thefour dimensions of structure listed earlier: thenumber of levels; the determination of the num-ber of patterns of distribution—single-, dual-, ormulti-channels; and their type (as defined byfunctions performed). Insight into the determi-nation of one other channel dimension—horizon-

runcTional Spin-Off: A Key to Anticipating Change rn Distribution Structure 23

tal structure, or the number of middlemen at eachchannel level—may also be provided by the func-tional spin-off concept. However, integration ofcertain concepts from the field of industrial or-ganization with the spin-off concept is necessary.

The basic force which will determine the num-ber of channel members at a given level appearsto be the size of the market in relation to theoptimum scale of the firm.^ The greater themarket size is in relation to the optimum scalesize, the greater the number of channel membersthat will evolve, and vice versa. The optimumscale size will itself change from industry to in-dustry and over time as technology changes.

In addition to this fundamental determiningrelationship, there are a number of other factors.Most of these factors are closely associated withscale and market size, which also help to de-termine how many members will exist at a chan-nel level. The forces which tend to create morefirms at a given channel level include: disecono-mies of medium or large scale plants and firms,only small inefficiencies at smaller than optimumscale of operation, growth of market size, highprofit potentials, inability to agree on mergerterms, legal barriers on merger or monopoliza-tion, and buyers wishing the convenience of manyoutlets.

The forces which tend to create fewer firms ata given channel level include; no diseconomiesof large scale operations or diseconomies com-mencing at only extremely high output of plantsand firms; only small inefficiencies at a largerthan optimum scale of operations; decline in mar-ket size; low profit potential; monopolizationpractices brought about through collusion, preda-tory behavior, or barriers to entry; and "out-side" financial considerations in effecting mergers.

Although most theoretical discussions normallyview diseconomies as starting almost immediatelyafter the optimum scale is reached, in practicethere appears to be a rather broad range ofpossible scales—from a minimum optimum scaleto some maximum (which in some cases couldaccommodate a monopoly)—which could providethe same optimum efficiencies.

We will seldom if ever find firms with asingle unique optimal scale. Diseconomies ofvery large scale are typically encountered, ifat all, only at scales substantially greaterthan the minimum optimal scale of a firm.This is in spite of the fact that a prioritheories of pricing and market structure haveusually represented the scale curves of firmsas having a U shape. . . .̂

5. For an excellent detailed discussion of this concept(though not interpreted in a channel context) see, Joe S.Bain. Industrial Organization (New York: John Wileyand Sons, Inc., 1968). Chapter 6.

6. Same reference as footnote 5, p. 175.

. . . the bulk of evidence is consistent withthe hypothesis that the gigantic firms are ingeneral neither more nor less efficient thanfirms which are simply large.'

Hence, since diseconomies come slowly, if atall, a basic underlying trend could be towardfewer firms and, at the extreme, one (monopoly)channel member at a given level of a channel.In other words, it is conceivable that a channellevel not only can be an oligopoly, duopoly, oreven a monopoly and still operate, but musthave that market structure in order to operateat an optimum scale.

This concept of a wide range of optimum scaleswhen combined with the concept of functionalspin-off has important implications for channelstructure. It suggests the possibility of a struc-ture which consists of all producers using directchannels, all using indirect channels, or someusing direct and some using indirect. The firstcase is probable when all producers reach theminimum optimum point; the second is probablewhen none reach the minimum optimum point.The last case (of mixed structures) is most proba-ble when some do and some do not reach thispoint; it is also optimally compatible when allreach the minimum optimum point, as in thefirst case.

Consider a hypothetical example of a markethaving a total volume of $10 million and a mini-mum optimum scale for distribution functions(whether undertaken by the producer or a mid-dleman) of $1 million, with no diseconomies upto the total market. If no producer reaches thisminimum, then it is probable that all will dis-tribute indirectly with a total of one to ten mid-dlemen in the market. If one producer reaches100% of the market, he can distribute directlywith no middlemen in the market, he can dis-tribute completely through one to ten middle-men, or he can distribute directly in part andindirectly in part through one to nine middle-men (nine, if he lets S9 million go indirectly). Ifonly one producer reaches SI million in output,he may distribute directly or indirectly. If hedistributes indirectly, then the other producers,who will probably distribute indirectly in anycase, will do so through one to nine middlemen.If there are ten producers, and eight producerseach reach at least $1 million in sales for a totalof $9 million, some of the eight may distributedirectly and others indirectly, or all may dis-tribute one way or the other. If all decide todistribute directly, then the remaining two pro-ducers will probably distribute indirectly throughone middleman.

Thus the minimum optimum scale point onlyindicates the maximum number of channel mem-

7. Same reference as footnote 5, p. 173.

24 Journal of Marketing, July, 1973

bers that are compatible with maximum efficien-cy, and so leaves room for a number of otherforces (not discussed here) which will determinethe actual point in the optimum range that willbe utilized by the firms in the industry, i.e., theactual outcome in terms of number of channelmembers at different levels. Further, firms maynot even choose to (or cannot) operate withinthe optimum range because of some of the fac-tors listed earlier. These factors include: onlysmall inefficiencies at nonoptimum scales, a de-cline in market size, a high profit potential evenwithout optimum operation, monopolization prac-tices, inability to merge, and a desire for outletconvenience. In spite of these possibilities, onecannot ignore what is a very key determinantof the number of channel members that willexist at each level—market size in relation tothe optimum scale of firm and the spin-off ability.

Hypotheses

The key hypotheses generated by the functionalspin-off and industrial organization concepts fordistribution structure are as follows:

As Related to "Number of Levels" Dimension

1. A producer will spin off a marketing func-tion to a marketing intermediary(s) if thelatter can perform the function more effi-ciently than the former. This will logicallybe the case when economies can be effectedfor that function by a change in volumefrom that of the producer. The greater theeconomies, the greater will be the incentiveto spin off. If the majority of producersin a given industry are in or will be in asimilar position, then the use of marketingintermediaries will characterize, or come tocharacterize, that industry.

2. If there are continual economies to beobtained within a wide range of volumechanges, the middleman portion of the in-dustry (and perhaps individual middlemen)will become bigger and bigger.

3. A producer will keep or resume a marketingfunction from a marketing intermediary(s)if the former can perform the function atleast as efficiently as the latter. This willlogically be the case when no economiescan be effected for that function by a changein volume from that of the producer. If themajority of producers in a given industryare in, or will be in, a similar position, thenthe nonuse of marketing intermediaries (di-rect distribution) will characterize, or cometo characterize, that industry.

4. If in performing a marketing function amarketing intermediary finds that for a partof that function (i.e., a subfunction) an-

other perhaps more specialized marketingintermediary can perform it more efficiently,then he will spin off that subfunction to thelatter. This will occur for the same reason-ing presented in hypothesis 1 above. Simi-larly, the first marketing intermediary willkeep or resume a subfunction if there areno economies to be effected by a spin-off.

-45 Related to "Number of Channels" Dimension

5. If a producer finds that in marketing toone (or more) of his markets a middlemancan perform a given marketing functionmore efficiently for the reasons noted inhypothesis 1 above and for another (orothers) of his markets he can perform thesame function at least as efficiently for thereasons noted in hypothesis 3 above, he willspin off that function in marketing to thefirst market(s) and keep or resume thefunction in marketing to the second. Ifthe majority of producers in a given in-dustry are in, or will be in, a similar po-sition; then the use of dual- or multiple-channels will characterize, or come to char-acterize, that industry.

As Related to "Middlemen Types" Dimension

6. If marketing intermediaries characterize anindustry, their nature will be determinedby the mix of functions and subfunctionsspun off. For example, if the ownershipfunction is a prevalent spin-off function,then the merchant will be a prevalent typeof marketing intermediary in the industry.

As Related to "Number of Middlemen" Dimension

7. The greater the market size is in relationto optimum scale size (at each channellevel), the greater the number of channelmembers that will come into being. Withthe growth of market size, and especiallyif there exist diseconomies or only verysmall economies of larger scale, more firmsmay be expected to enter the channel. Witha decline in market size, and especially ifthere exist economies of larger scale, firmsmay be expected to leave the channel.

8. With a change in technology and the growthof optimum scale size, firms may be ex-pected to leave the channel if there is nocorresponding change in market size andvice versa.

Implications

Using the eight relationships hypothesizedabove, the channel planner is in a better positionto anticipate trends in his industry by estimatingrelevant cost and market data. (It is beyond the

hunctional Spln-Off: A Key to Anticipating Change in Distribution Structure 25

scope of this article to describe a program ofdata collection or to suggest how to overcomethe admittedly difficult, but not impossible, taskof collecting competitive cost information.)

The channel planner must estimate the presentand future total market volume, new technologi-cal changes which affect the optimum firm scale,and the volume and shape of the average costcurve for the key marketing functions of arepresentative sample of firms (producers andmarketing intermediaries) in his industry. For-tunately, because his purpose is to gauge broadunderlying trends, the data can be fairly roughwithout losing its usefulness. Trade associationsoften collect competitive cost data for use bytheir members. However, further estimates ofdifferent types of marketing costs by marketbeing served would probably have to be madefrom such sources. He then can apply to hisfindings the structural results predicted by theeight hypotheses listed above.

For example, if he finds that producing firmsin his industry in general have faced a decliningaverage cost curve for the ownership functionin the past but are now approaching a flat por-tion of the curve where no economies of scaleare forthcoming, and further he predicts an in-crease in total market volume, he will anticipatein his channel strategy a general move by hisindustry to more direct ownership channels (seehypothesis 3 above). He will also anticipatemore firms entering the market at the producerlevel (see hypothesis 7 above: note in this casethat the market size may actually decline at the

merchant middleman level because of the firstconclusion). The exact change in number ofchannel members will depend on factors dis-cussed earlier under the subtitle "horizontal chan-nel structure." If, in the same example, thechannel planner also finds that with the samevolume increase the negotiation function costcurve declines sharply, he will also anticipate anincrease in the number of agent middlemen.

The possible combination of cost and marketsize changes are numerous. The channel plannerby collecting the data and applying the conceptsdiscussed here should be able to anticipate dis-tribution structure changes before they becomeobvious trends.

The functional spin-off concept is a powerfulconceptual tool for the marketer in understand-ing many of the aspects of channel structure(which itself is one of the most and perhapsthe most fundamental contribution of marketingas a discipline)^ and in predicting structuraloutcomes in specific industries. It is also useful,in a micro or managerial sense, to the channelselector or designer who is seeking to concept-ually organize and understand the framework ofunderlying economic forces within which he mustoperate and make his decisions and to which hemust adapt.

8. R. Ferber, D. Blankertz, and S. Hollander, MarketingResearch (New York: Ronald Press Company, 1964), p.471. See also, Michael Halbert, The Meaning and Sourcesof Marketing Theory (New York: McGraw-Hill. 1965),p. 10.

'MARKETING MEMO

The Consumer Isn't Always R i g h t . . . A Lesson Perhaps For Marketers . . .

How should good teaching be measured? The major defense for defining goodteaching in terms of good scores on the student evaluation forms is based on ananalogy between the student and the consumer—the student, as the primaryconsumer of the teaching product, is in the best position to evaluate its worth.However, the present data indicate that students are less than perfect judges ofteaching effectiveness if the latter is measured by how much they have learned.If how much students learn is considered to be a major component of goodteaching, it must be concluded that good teaching is not validly measured bystudent evaluations in their current form.

—Miriam Rodin and Burton Efldin."Student Evaluations of Teachers,"Science, Vol. 177 (September 29,1972), pp. 1164-1166, at p. 1166. Copy-right 1972 by the American Associa-tion for the Advancement of Science.