Yoram Barzel June 2012

1. The exchange of variable-quality commodities Commodity variability is the root of all evil • Yoram Barzel • June 2012

2. Introduction The assumption that commodity quality is uniform is entrenched in economics. But in real life the quality of many commodities is far from uniform. Moreover, buyers are seldom uniform either sellers’ perspective. Two questions arise: • 1. How to incorporate the non-uniformity in the analysis? Indeed, what does “price” mean in this context? • 2. What are the consequences of the non-uniformity?

3. Introduction, cont. It might appear that to incorporate the non-uniformity in the D-S model we can simply add a quality dimension to it. But quality is multi-dimensional. The heart of the problem is the positive cost of information about non-uniform commodity specimens. This cost prevents the clear delineation of rights. The competition to capture these rights is dissipating.

4. Introduction, cont. • A number of major organizational forms have been adopted to reduce these costs. These include guarantees, restrictions on buyers’ opportunities to pick and choose, securitizing and employment for wages. • The rationale for the employment for wages underlies the theory of the firm.

5. Introduction, cont. • I have been thinking and writing about variability for a long time, especially in Barzel (1982). • Akerlof (1970) is the first to introduce the variability problem. • I proceed by first discussing the underlying problem. The rest of the time is given to two distinct issues. One is analyzing how equilibrium is attained in the face of buyers who pick and choose. The other regards arrangements to reduce the effects of variability.

6. The role of information cost • In the purely competitive Walrasian world information cost is zero and commodities are uniform and of known quality. There are no disputes as to what is being exchanged, or at what terms. • When information is not free, resources are required to produce it. • We expect information to be produced by whoever has comparative advantage in doing so. The possessors of the information, which is necessarily asymmetric, might exploit it at others’ expense.

7. The role of information cost, cont. • Two questions arise: • 1. How is the information actually produced? • 2. What are the consequences of it being asymmetric? Akerlof (1970) pioneered the analysis of asymmetric information. His assumptions that sellers possess information about their commodities that buyers do not have preempts the analysis of the first and limits the range of its consequences. I attempt to answer the two questions.

8. The role of information cost, cont. • Sellers may realize the value of their non-uniform commodities by selling every specimen at a price commensurate with its quality. But for most commodities this is too costly. • Making commodities more uniform alleviates the problem, but is costly too, so variability usually remains.

9. Who chooses? • Sellers may retain the right to select specimens for buyers, or let buyers pick and choose. To effect the former requires buyers to trust them. • Sellers sometimes effect the selection. • In most cases, however, sellers allow buyers to pick and choose, but prior to that they can sort their commodities or make other arrangements.

10. Sorting • Sorters who have comparative advantage in the activity, must decide how many attributes to sort by, and which attributes to leave unsorted. • I assume that sorting takes place at a central location where commodities are sorted into “grades” to meet the specifications of customers such as supermarkets. • Customers seek sorting levels such as to equate their sorting costs to the increase in net revenues that the better sorting generates.

11. Sorting, cont. • Buyers’ measurement skill is lower than sorters’; otherwise they would have performed the sorting in the first place. Moreover, sorters effect sorting when and where it is most advantageous. • In spite of their higher sorting cost, buyers may still sort primarily because sorting is designated for the general buyer but not for particular ones.

12. Sorting, cont. • Buyers with taste for particular attributes not sorted by will find it worthwhile to pick and choose. • Buyers make (implicit) measurements when choosing. Others duplicate the measurements of the specimens “rejected” by previous buyers.

13. Pick and choose • Under pick and choose sellers cannot capture the difference between the price they charge and the value of specimens buyers select. • To buyers the quality differential is a free attribute; they select only units valued more than the price and continue to choose until their marginal gain from the positively valued quality is zero.

14. Pick and choose, cont. • What buyers gain is not a pure transfer to them. As with catching fish in the ocean, part of it is dissipated because they compete with each other for the better units by rushing or waiting in line. • I define dissipation as the difference between actual outcomes and the (unattainable) Pareto outcomes.

15. The selection process • The model. • I construct a pick and choose model for a commodity and illustrate it using numerical values. • Suppose the quality of a commodity is uniformly distributed with support $90-110. • The seller initially charges a price p1.

16. p 0.10 commodity value 90 100 0 110 Chart 1 P P 1 2 P

17. The selection process, Cont. Buyers proceed in four steps. • 1. Given their demand (specified in constant quality units ), the price charged, and their rough estimate of the offering, until thy find one where thy wish to shop. • 2. At a cost of $1 they estimate the distribution of quality. • 3. Each randomly picks units one at a time and spends $1 for inspecting each of them. • 4. Each buys when the first unit in his sample whose value exceeds the amount he is willing to pay.

18. The selection process, Cont. • Whereas a buyer would pay, say, no more that $98 for a unit of average quality of the commodity selling at $100, he will nevertheless choose to shop at this particular seller because the chance of getting a unit worth sufficiently more than average is high enough. • Individuals who demand n units will simply repeat the selection process ntimes.

19. The selection process, Cont. • The composition of the seller’s offering deteriorates with each purchase since the units purchased come form the upper end of the distribution. As a result • 1. Fewer individuals will proceed with the inspection process. • 2. The net gain of those who choose to buy declines, and • eventually nobody would be willing to buy from that seller’s collection.

20. The selection process, Cont. • To simplify what comes I assume that the units selected are always the best available. • As support of the distribution shifts up (down), price will increase (fall) but buyers’ decision will follow the same pattern. • As the support widens, picking and choosing becomes more attractive. • As it narrows, eventually it will cease to generate net gains to buyers. Buyers will treat all the units as equal and refrain from picking and choosing. Their net price will fall, and dissipation form that activity will become zero.

21. The selection process, Cont. • Buyers differ in their cost of selection, say due to different time costs. The main effect of this factor is that the low-cost buyers deprive high-cost ones from the best units. Some of the latter even may decide not to visit sellers serving diverse buyers. • Similarly, low time-cost patrons take the best seats in a single price movie theaters, making the events less attractive to high time cost ones. • Sellers may counter by offering time saving services such as shorter lines at the cashiers, and raise prices correspondingly. Such sellers become less attractive to low time cost individuals. • The practice, then, results in separating equilibria.

22. The selection process, Cont. • I now turn to sellers’ behavior. As is obvious, a seller must lower his price when buyers no longer find his commodity worth buying. • As the seller’s act of price change is costly, he will change his price in price changes in discrete steps. • Each step reflects different quality range, and each entails some picking and choosing. • Sellers whose high quality merchandise is depleted put the remaining items on “sale” or sell what remains to “outlets” or to others catering to individuals more willing to buy the low quality units.

23. Dissipation and vertical relations. In the numerical example for simplicity I focus on buyers who are quality-variability neutral. The value of the variable component of the seller’s offering is, on average, $10 per unit. Of the $10, the seller retains about $3, buyers capture about $2 and $5, which is about one half of the value of the variable component of the commodity, is dissipated. The dissipation consists of the seller’s cost of changing price of about $1, of buyers’ selection cost of about $2, and of $2 in buyers’ competition with each other. • The numerical results give an idea of what one might expect in reality, and why the parties are eager to reduce the effects of variability.

24. Dissipation and vertical relations, cont. • A loss of $5 out of $100 may not appear large. But first these costs are over and above the cost of sorting and the cost of other preventive arrangements. More importantly, they are incurred at every one of the vertical stages where the upstream input suppliers and their downstream recipients carry out a market exchange. • For instance, had the Smithian upstream pin producers sold their output to the downstream producers, the dissipation would have occurred between each pair of the 18 successive, and seemingly independent stages of the pin production.

25. Methods of coping with variety • I now turn to methods of coping with variebilty. Besides more thorough sorting or separating buyers, sellers may indirectly reduce the impact of commodity variability or altogether finesse it in a number of ways.

26. 1. Guaranteeing output • Guarantees, from buyers’ perspective, make commodities essentially uniform. They dispenses with picking and choosing as buyers are entitled to one good unit. To buyers guarantees transform the variable commodity into a fixed one. • By guaranteeing, sellers retain ownership over commodities’ variable components. They will lose when providing low quality as they face a high level of guarantee claims and they are induced to produce the optimal quality mix. • Akerlof(1970) was the first to point out the use of guarantees to avoid lemons. See also Barzel (1982).

27. 2. Restricting buyers’ opportunity to pick and choose. • 2a. Bundling. Bundling sits half way between sellers’ selection and buyers’ picking and choosing. I hypothesize that bundling serves to discourage buyers from picking and choosing. But then buyers must trust sellers not to load the bundles with low quality specimens. An implications of the bundling hypothesis is that in areas with significant wage dispersion, and thus wide dispersion in the cost of picking and choosing, the fraction of bundled commodities is larger than in areas with more uniform incomes.

28. 2. Restricting buyers’ opportunity to pick andchoose.cont. • 2b. Restricting buyers. Barzel, Habib and Johnsen (2006) show that the attempt to dissuade speculators from picking and choosing explains many of the restrictions associated with IPO’s such as the number of shares one may buy. Two other cases where sellers restrict buyers’ ability to pick and choose are: • First in some wholesale used cars auctions. • Second is De Beers sightings where buyers’ picking and choosing is severely restricted. In Barzel 1977 (preceding Kenny and Klein) I argue that the purpose of the practice is to prevent dealers from picking and choosing.

29. 3. Sorting individuals into homogeneous groups to finance commodity purchases • The variability of the individuals engaged in exchange is another source of exchange problems. Consider financing assets posted as collateral. Securitizing the loans reduces the financing cost, but variability in borrowers’ default rate facilitates adverse selection. Lenders impose restrictions both in terms of the assets and of the borrowers such that borrowers become largely uniform, needing only modest scrutiny.

30. 3. Sorting individuals into homogeneous groups to finance commodity purchases cont. • Consider new car buyers. New cars become less uniform as they age. The down payment and a repayment schedule lenders require as well as minimal borrowers’ income-level requirement makes the latter’s default rate more uniform, reducing the need to thoroughly scrutinize them. The standardized lenders’ assets is what accommodates the securitization of the loans. • Borrowers become the owners of the cars’ variable component, giving them a strong incentive to protect the equity in their cars and avoid default. • To prevent lenders from using excessively lax selection criteria, we expect securitizers to require lenders to hold the “toxic” tranches of the loans they initiate. • Similarly, securitizers are expected to require the mortgages’ originators to hold the equity, i. e., the most junior, (“toxic”) tranche.

31. 4. Wage contracting and its rationale for the firm. • The wage contract radically alters the form of exchange as it altogether eschews the commodity exchange problem. Two major reasons for exchanging labor services for wages are: • 1. Bypassing the need to measure workers’ product. • 2. Have workers cooperate with capitalists who guaranteeing their output in exchange for employing and controlling them. These reasons for the use of wage-labor contribute to the theory of the firm.

32. 1. Bypassing the need to measure workers’ products. • Owners of labor services may operate as independent contractors producing and selling their output in the market, or sell their services for wages. In the latter case upstream wageworkers transfer their output to downstream ones. Employing workers for wages serves as a substitute for exchanging commodities in the market.

33. Employment for wages, including employment in vertically integrated firms, requires less measurement of the commodities being transferred than where workers sell their output in the market. Within firms employers instruct employees what and how to produce andto whom to transfer their output, reducing employees’ gain from picking and choosing, thus reducing the need to measure the commodities (Barzel, 1982). The employment for wages, then, finesses the problem of exchanging non-uniform commodities in the market. Wage-employees’, however, have to be induced to produce, and this requires costly supervision. Alchian and Demsetz (1972) explicitly point to entrepreneurs’ need to employ workers for wages and supervise their effort when measuring workers’ output is (prohibitively) expensive. These considerations apply to every vertical production stage except when commodities are sold to final consumers

34. 2. Guaranteeing output • The expected value of workers’ output may be positive yet in some cases the specimens they produce or their activities may have a large negative value (exploding products; damaging fellow workers). Buyers or fellow workers require guarantees to protect themselves from abuse. But when losses are large, workers are unlikely to possess the means to cover them. • Capitalists may provide the guarantees accompanied with the wage contract along with the supervision that goes with it. • Thus solutions to the problem that product variability create lead to the wage contract.

35. 3. The firm. • Employer-employee relationships are conducted within firms. The explanations given here for these relationships contribute two major ingredients to the theory of the firm. One contribution regards the scope of vertically integrated firms that emerge in response to measuring variable-quality products (Barzel,1982). Alchian and Demsetz (1972) is a special case here, • The other contribution regards firms guaranteeing workers’ output; a force that is enhanced by scale economies to assembling guaranteeing capital. This force may lead to large, integrated firms with clear boundaries (Barzel 1997).

36. 5. Exchanging services • Like commodities, services may be exchanged directly under explicit or implicit guarantee or via the wage contract. Whereas some services are sold by output, selling others indirectly via the wage-employment is a substitute for the more costly exchange by output.

37. 5a. Services produced by wage workers. • There must always be some connection between input and output. But then service-workers’ output is measurable, even if imperfectly. One may rather conclude then that services differ from commodities in that their output is more costly to measure directly than that of commodities so they are often measured indirectly via the input intensity. (Services can’t be returned.)

38. 5b. Guaranteed services. • Some services are readily measurable and given the variability in their output, they are often guaranteed. One such services is advice. • One type of advice is how to minimize the net losses from fire, this service has measurable financial effects. Advice seekers are obviously ignorant of the quality of the advice they seek. The exchange of advice implies asymmetric information, and the advisor could overcharge for the quality he or she provides. To reduce that we expect the advisor to become the residual claimant to the effect of the advice, which . implies here guaranteeing it (Barzel 1997).

39. Guaranteeing advice cont. • Here too guarantors means may be insufficient for guaranteeing. The advisers may abet clients’ fear of not receiving the benefits they are entitled to by joining fire insurance firms that back the advice of these specialized employees.

40. Summary • This paper explores the effects of commodity (and service) quality-variability. The costliness of measuring commodities results in not well-delineated rights, and dissipation occurs in the competition for them. • Pricing each of the specimens and selling them individually is too expensive for most commodities • Sellers then must sell most of their commodities at uniform prices. They may select specimens for their buyers, but the cost of creating the trust needed for that purpose is usually too high. • So sellers usually let buyers pick and choose from uniformly priced but non-uniform commodities. • Sellers, however, first sort commodities and adopt arrangements to reduce their losses from buyers’ picking and choosing. • How the market equilibrates in the face of picking and choosing is analyzed, and Chart 2 summarizes the results.

41. In Chart 2 D1 is the conventional demand curve and D2 is the demand curve net of buyers cost of picking and choosing and of competing with each other. S1 is the conventional supply curve and S2 reflects the added costs of sorting and changing prices. The actual market quantity is Q2 and Price is P2. • While not shown, it is easy to see the associated dissipation consisting of the triangle between Q1 and Q2 and the areas between the two pairs of curves to quantity Q2.

42. S D Q

43. Several methods are used to lower the cost of dealing with variability, each presumably chosen when it reduces the dissipation more than any of the others. Of these methods one is making borrowers more uniform from sellers’ perspective, thus reducing lenders’ cost of selecting them when their loans are securitized, and one is measuring wage employees’ effort which is a substitute for measuring goods and services. Organizing the exchange via the use of the wage contracts means that workers are placed in firms. The factors that explain the use of wage-workers also explain vertical as well as horizontal integration, and thus contribute to the theory of the firm.

44. Thank you!