You are on page 1of 16

Transaction Costs and Slanghter Cattle Procnrement:

Processors' Selection of Supply Otiannels


/ . / / E. Hobbs

This article investigates the hypothesis that different Introduction


supply channels impose different types (and levels of)
transaction costs on beej processors in the Vnlted When purchasing slaughter eattle, meat processors
Kingdom. The choice of supply channel is therefore in the United Kingdom have a uumber of procure-
influenced by these transaction costs. Recent food safety ment alternatives from which to choose. Cur-
legislation and increasing consumer concerns over farm rently, approximately 50% of slaughter cattle are
animal welfare may have altered the transaction costs sold to processors" through live-ring auctions, with
arising from different supply channels. Conjoint the remainder largely purchased direct from indi-
analysis is used to measure the relative importance of vidual cattle producers. The proportion of cattle
selected transaction costs in a processor's procurement sold through live-ring auctions declined from a
decision. A survey of VK beef processing firms is used to level of 72% in 1964. Electronic auction systems
collect data for the conjoint analysis. The results were introduced to the United Kingdom iu 1989
suggest that the monitoring costs arising from the and now account for approximately 8% of slaugh-
traceability of cattle to the farm of origin are ter cattle sales. Cooperative group marketing
particularly important. The implications for vertical schemes account for a further 5% of sales.
coordination in the beef marketing chain are
discussed. ©1996 John Wiley & Sons, Inc.

"Throughout this article, thr terms beef processor, abattoir,


slaughterhouse, or parkier will hv used inteirhanf^'ably.
The use of rattle dealers or agentJ^ is no longer a signiTicant vcrtU'al
alternative.

Tbe author gnilefully acknowledges {he helpful commenis of anonymous reviewers. The research for ihui article was carried out uhile the
author was with the Economics Division, Department of Agriculture, University of Aberdeen and the Scottish .Agricultural College, Aherdeen,
Scotland.
• The author is with the Excellence in the Pacific Research Institute, University of Lethbridge and Faculty of
Management, University of Calgary, 4 219 Village Terrace SW, Calgary T3H 2L4, Canada.

, Vol. 12, No. 6, 509-523 (1996)


1996 by John Wiley & Sons, Inc. CCC 0742-4477/96/060509-15

'509
Hobbs

This atudy investigated the hypothesis that differ- Other words the costs of carrying out an exchange
ent supply channels impose different types (and of goods or services. They arise wherever there is
levels) of transaction costs on heef processors. any form of economic organization.-* Traditional
Consequently, the choice of supply channel is in- neoclassical economic theory implicitly assumes
fluenced hy these transaction costs. Clearly, the transaction costs to be minimized. When modeling
transaction costs that arise from different supply economic hehavior, they are assumed, for sim-
channels do not provide the sole explanation for a plicity, to be zero. The standard neoclassical
processor's choice of supply channel. Other im- transaction involves the exchange between eco-
pi)rtant factors include technological innovations nomic agents (buyers and sellers) of a homogenous
that generated economies of scale in meat process- product: there are no quality variations between
ing. This led to larger plant sizes with lower aver- products and consequently no costs involved in
age total costs of produ<'tion that, in turn, measuring the value of a proiluct. Where products
generated economies in procurement. Longer do exhibit (|uality differences, they are treated as
term, regular supply relationships with producers separate products serving distinct markets. Eco-
he(^ame more attractive for the processor because nomic agents are assumed to possess full informa-
of the need to fully utilize the capacity of the ti()n; hence, there is no uncertainty regar<ling
plant. However, this may he more significant in prices, product characteristics, or the beha^ior of
the North American industry than in the UK in- competitors and trading partners. Neoclassical
dustry. The North American industry has seen economic analysis concentrates on equilibrium
considerable rationalization and concentration on outcomes between firms in a market. There is no
both the processing and production sides, which is direct consideration of how transactions occur.
characterized by relatively large feedlots. This led Transaction cost economics (TCE) focuses instead
to clftser relationships between the produc tion and on reas<)ns for the existence an<l behavior of firms
processing stages of the beef chain, either through and on the nature of business transactions. '••'' It
vertical integration, forward contracting, or pur- suggests that tbe nature and level of transaction
chasing agreements with processors.^ The struc- costs determines the vertical coordination out-
ture of beef profhictii>n in the United Kingdom, come. For example, if the transaction costs of pro-
however, is quite different. Relatively small scale curing cattle through live-ring auctions were to
family farms still (Umiinate the industry (the aver- rise, meat processors may turn to other form of
age beef herd size in England and Wales is around vertical coordination such as procuring cattle di-
20 head of cattle, 41 in Scotland). Hence, a trans- rectly from farmers or through electronic auc-
action cost explanation of the processors' choice of tions.
supply channels bears further investigation. The UK meat processing industry faeces a number
Transaction costs are simply the costs of carrying of external pressures that may alter the transac-
out a transfer of goods between technically separ- tion costs associated with procuring cattle. Two
able pbases of production or distribution, or in examples are the recent Food Safety Act (FSA,
1990) and increasing consumer <'oncerns over ani-
mal welfare on farms. This article investigates the
and Jones vertical coordination to include: importance of transaction costs in the selection of
"*. . . all ihe ways of harmonizing the suci-t^«»ive vertical steptt supply channels. The effects of food safety legisla-
of jirodui-tion uml marketing. The market-price sywtem. verti- tion and animal welfare concerns on these transac-
cal iiilegratiiiii, rontractin^, ciiuperatiun singly or in conihinu- tion costs are (considered.
tion are some of ihc nlternative meant; of eoordination" (p-1)-

Note that in thiif article the broad deHnition of vertical coordination


arrangements, which includes market prices, will he used. Sometimes Transaction Costs
a narrower ilefinitiim of vertical coordination arrangements is used
that is confined to nonmarket arrangements. This problem of defini- In terms of cattle procurement, transaction costs
tions can sometimes lead to confusion. are more than simply the monetary costs associ-

510
Cattle Procurement

ated with the purchase and delivery of slaughter According to Ward^ the second type of price dis-
cattle. Instead, they encompass all aspects of the covery process involves adjusting tbe general price
transactional relationship between buyers and sell- level for the specific characteristics of the prod-
ers in the supply chain. It is useful to divide uct. In other words, for dead-weight sales, a car-
transaction costs into three main classifications: cass grading system is used to reward (tr penalize
information costs, negotiation costs, and monitor- carcasses that deviate from some established aver-
ing (or enforcement) costs. age carcass evaluation.
Procuring cattle through a live-ring auction ex-
poses proc^essors to a degree of grade uncertainty
Information Costs because cattle are purchased on a live-weight per
head basis. Hence, the processor bears the risk
Information costs arise prior to a transaction. that the cattle will not grade as expected. Buyers
Economic agents face costs in the search for infor- may have paid a high price for an animal that
mation about products, i)rices, inputs, and buyers looked promising in the auction ring, only to dis-
or sellers. For example, beef processors may incur cover that it does not grade well once slaughtered.
price information costs in deciding the price to of- Research has suggested that the amount of infor-
fer for cattle purchased direct from farmers. mation about cattle quality affects the degree of
When purchasing cattle through the live-ring auc- risk associated with the buyer's pricing deci-
tion, prices are set by the process of competitive sion.**-^ Live-weight sales offer buyers tbe least in-
hidding among hnyers. The subsequent market formation about cattle: buyers know the live
prices become publicly available information. This weight of the cattle, nsually the farm of origin,
reduces price information costs for the ahattoir and they may (or may not) be aware of tbe pro-
because market forces determine prices for each duction methods used on that farm. Direct dead-
transaction. The residtant pri(^e information acts weight sales provide the buyer witb far more in-
as a guide to live-weight auction prices in other re- formation. Price is based on the actual deac!-
gions or later in the week. Furthermore, pro- weigbt carcass grade and should more accurately
cessors in the United Kingdom often hase their reflect the quahty of the animal. Fausti and Feuz^
dead-weight prices on the previous week's live-ring show that, assuming l)uyers to be risk averse, av-
auction price. The price discovery process is de- erage prices paid for cattle sold through live-
scribed by Ward^ as a two stage process. The first weight marketing channels are lower than those
stage involves finding and assessing the overall paid for cattle sold throngh various dead-weight
price level for tbe product in terms of the influ-
marketing channels in which the buyer has more
ences of supply and demand. The use of publicly
information about the quality of the animal. Con-
available price information from live-weigbt mar-
versely, sellers face greater nncertainty when sell-
kets in tbe United Kingdom reduces this cost of
ing animals as dead-weight because they incur the
price discovery for processors. The paradox of
this situation is that as increasingly more pro- risk that the cattle will not grade as expected. Dif-
cessors source their cattle directly from farms, ferent sellers will have different levels of risk
while basing their prices on live-ring auction aversion that will affect their preferences for dif-
prices, the volume of cattle sold through auctions ferent marketing channels.'^ Processors face less
declines. Auction prices inevitably become less grade uncertainty because of the live cattle assess-
representative of slaugbter cattle transactions. ment abilities of procurement staff. To achieve a
This "thin market" problem bas been discussed bigh level of proficiency in procurement staff,
extensively in the literature, see, for example, however, requires investment in human capital.
and
misreporting jirices. However, this may be Icsn nf H concern in the
United Kingdom where national average live-weight prices arc col-
' .A Turlher concern raised by Tomek~ is thai ihin marketN can re- lected and puhhshed hy a separale industry body, the Meal and Ltve-
suli in ilie manipulation of reported prices by selective reporting or Btuck Commission.

511
Hobbs

This is a transaction cost. Alternatively, the pur- one hand and a processor (and sometimes retailer)
chase of cattle direct from farmers or through on the other. Part «f a processor's motivation for
electronic auctions on a dead-weight carcass grade engaging in this type of supply relationship may be
hasis reduces grade uncertainty for processors. to ensure a steady supply of cattle to keep plants
Additional information costs may arise in identi- at full capacity. Plowever, an additional incentive
fying appropriate sellers. If processors purchase may also be to reduce information costs for the
cattle direct from farmers they can develop long- processor. This becomes particularly important as
term supply relationships with individual pro- consumers develop more specific reciuiremcnts for
ducers. Although not <-omniitted to buy/sell from product characteristics such as improved food
one anttther, both parties often value the relative safety standards and animal welfare standards.
security that this type of arrangement can bring.
Under a coo]>erative gronp marketing scheme
(GMS) a group of producers agrees to sell cattle of Negotiation Costs
a particular (]uality to a specific buyer. Tbis often
involves a processor and a retailer in a three-way Negotiation costs arise from the physical act of the
partnership with the farmer group. A GMS allows transaction and are influenced by tbe way in
processors to reduce their supplier information which the transaction is carried out. For example,
costs by providing a regular source of supply for the opportunity cost of the time that procurement
cattle (tf a relatively consistent quality. On the staff take to locate supplies of cattle is a negotia-
other hand, a processor relying on live-ring or tion cost. This cost may he higher for direct dead-
electronic auction markets for a supply of cattle weight pur<-hases from individual producers than
depends on the regular use of this marketing for auction market purchases where many cattle
channel by farmers wbo are known to produce are present at one location. The use of electronic
good (piality cattle. If this is not the case, the in- auctions avoids some of the negotiation costs of
formation costs of de< iding which producer''s cattle procuring cattle because the visual assessment of
to purchase will rise. cattle is carried out by the electronic auction com-
In adtlition to lowering the information costs fac- pany rather than by the processor's own procure-
ing processors, a further motivation to engage in ment staff.
long-term regular supply relationships may be to A GMS will impose different types of negotiation
secure an atiequate supply of slaughter cattle to costs on a processor. Rather than dealing with in-
ensure that slaughter plants are not run consis- dividual producers, with whom the prctcessor has
tently below full capacity. This has become in- considerable bargaining pi)wer, the firm must ne-
creasingly important as the beef processing gotiate the quality and price details of the forth-
industry in the United Kingdom has rationalized coming transactions with representatives of the
into fewer and larger nnits. Average costs rise rap- producer group. The increased bargaining power
idly if a plant is not operated at its optimum tbat tbis gives sellers may be viewed as an increase
rate.2 In the United States in particular, pro- in negotiation costs for buyers, because their rela-
cessors have taken steps to ""capture" their future tive bargaining strength is diminisheil. On the oth-
supplies of slaughter cattle thrcuigh backward inte- er hand, if a group of regular suppliers is
gration into the feeding of cattle, forward contrac- established, the supply of cattle is secured and
ting with large feedlot companies, and purchasing buyers need not spend as mucb time canvasing lo-
agreements with livestock producers whereby tbe cal producers for supplies of cattle.
producer agrees to market a certain number of Another negotiation cost that differs among sup-
cattle to the processor over a specified time peri- ply channels is haulage. The uplift of cattle from
od. While backward integration and forward t-on- farms and delivery to the abattoir is always neces-
tracting have not o<'currcd in the UK industry, sary; tbis is not a transaction (M)st. However, when
group marketing schemes represent purchasing additional transportation is required to facilitate a
agreements between groups of producers on the transaction, such as the extra transportation leg

S12
Cattle Procurement

involved in moving animals to and from an auc- firms along the food chain. Monitoring costs for
tion, this additional cost is considered to be a ne- all food firms arc increased because they must
gotiation cost. When using the live-ring auction, monitor the standards and processes of firms fur-
the abattoir is responsible for the transportation ther upstream in the marketing chain, carrying
of stock from the marketplace to the slaughter- out and documenting tests on food (piality and
house. On the other hand, stock purchased pri- safety.'" F'or example, if tbere is a food safety
vately from producers may either be delivered to problem with beef sold in a supermarket that was
the slaughterhouse hy producers or collectetl from caused by a particular on-farm [)roduction prac-
the farms by the processor. When cattle are pur- tice or that was the result of a cattle disease, the
chased through an electronic auction system, pro- retailer (and processor) are now legally responsi-
cessors tend to be responsible for the collection of ble for taking all reasonable precautions to detect
cattle. On the other hand, if the processor is re- this problem.
sponsible for collecting cattle directly from farms, This has two main implications for processors.
it has more control over the timing of delivery to First, they must have procedures in place to moni-
the slaughterhouse. This can be beneficial in terms tor the quality and standards of their supplies.
of optimizing the efficiency of the slaughter line. Second. })rocessors face increased pressure frctm
supermarkets to provide them with information on
the history of their beef supplies, including assur-
Monitoring Costs ances about the production prac tices used on the
farms from which the cattle originated. Easier tra-
Monitoring or enforcement costs arise after a ceability of cattle to the farm of origin greatly re-
transaction. It may be necessary to monitor tbe duces this monitoring cost. Cattle purchased from
quality of goods from a supplier or to monitor the a cooperative GMS or direct from individual farm-
behavior of a supplier (or buyer) to ensure that ers may be more easily traced than cattle pur-
all preagreed terms of the transaction are com- chased through live-ring or electronic auctions. To
plied with. The costs of seeking restitution when a improve traccability and lower the costs of mon-
contract is broken are also considered to he mon- itoring production practices, processors may re-
itttring costs. Some of the key monitoring costs quire clearly documented information ahout the
facing processors arise because retailers need to breeding and production history of the cattle; cat-
ensure that animal welfare and food safety stan- tle from known sources could be pro<-essed in
dards have been adhered to hy farmers. bat(;hes, thereby enhancing traccahility through-
Monitoring and enforcement costs have become out the food supply chain.
more important in the UK food sector since the Two types of monitoring costs arise from animal
revisions to food safety standards embodied in the welfare concerns. The first involves the monitoring
FSA of 1990. The FSA places a new legal respon- of on-farm production |!ractices. Processors in the
sibility »n all those involved in the production, United Kingdom have been encouraged by super-
processing, distribution, and retailing of food to markets to source tbeir cattle supplies from farms
ensure that the UUHI within their possession con- with proven high welfare standards. Conse(|uently,
forms to the provisions of the Act. It does so by the traceability of stock to the farm of origin is
the introduction of a "due diligence" defense important. Processors facing this kind of pressure
clause. This means that, if acxused of an offense from their major customers will ])refer to pur-
under the Act, a defentlant must prove that he or chase cattle with a known background, that is,
she took all reasonable precautions and exercised from farms where the production practices are
all due diligence to ensure the standard and safety recognized as "welfare friendly." Their monitoring
of the food that they sell. In order to invoke the costs will be rediu ed if they can establish long-
due diligence clause, if charged under the Act, a term r(>lationshi|)s with these preferred farmers. It
firm must prove that it undertook adequate mon- may be more difficult for processors to give wel-
itoring of incoming supplies. Tbis applies to all fare assurances to retailers if they are purchasing

513
Hobbs

cattle from a wide variety of occasional suppliers may therefore reduce monitoring costs for pro-
through hve-weight marketing channels/' Vertical cessors because they do not bave to monitor on-
coordination channels that faciUtate these longer farm pro<luction practices that could affect (|uali-
term relationships will be transaction-cost reduc- ty. Instead, dead-weight grade payment provides
ing. farmers with tbe incentive to produce bigb quality
The handling of cattle hetween the farm and the cattle."*'"
abattoir can also impose monitoring costs on pro-
cessors. With increased handling of cattle, there is
a greater risk of carcass damage and weight loss Research Method
due to stress, which reduces the value of the car-
cass. Vertical coordination methods that involve The analysis that follows is based on tbe premise
an additional transportation leg, such as live-ring that each supply channel displays different attri-
auctions, may increase this monitoring cost for butes or characteristics, just as goods and services
processors. Cattle purchased directly from indi- are sometimes viewed by economists as buudles of
vidual producers or through most of the UK elec- attributes.'-^ A processor's preference for a supply
tronic auction systems are transported directly cbannel is determined by the particxdar bundle of
from the farm to the abattoir. Furthermore, as attributes that characterize that channel. Some of
with grade uncertainty, if cattle have been pur- these attributes are transaction costs or they may
chased through an auction on tbe basis of a Uve- be transaction cost reducing. Transaction costs
weigbt price per kilogram, tbe processor bears the vary among supply channels and. bence., affect
risk that weight Io8.s will occur before the cattle processors' preferences for the different channels.
are slaughtered. However, if cattle have been pur- Tbe key transaction cost features of beef supply
chased on a dead-weight carcass grade basis, this channels are first identified. Conjoint analysis is
risk remains with tbe producer. Tbe method of then used to evaluate tbe relative importance of
payment iletermines the incidence of this transac- each feature to a processor's procurement deci-
tion cost. sion.
Processors are increasingly pressured by re- Conjoint analysis has become a popular method
tailers to produce beef of a consistent and reliable of measuring the joint effect of product attributes
<|uahty. Tbis requires tbat the processors' slaugh- on preferences for a product. Tbe technifiiic has
ter cattle are also of a consistent <]uality. The use been applied, for example, to consumer purchase
of a dead-weight carcass grading system encour- decisions,'-* the evaluation of transportation alter-
ages farmers to produce cattle to more consistent natives,'* and the valuation of environmental ame-
quality specifications. It may be more difficult to nities.^'* Although often used to measure
ensure a consistent product quality if cattle are preferences for products or services, conjoint
procured through live-ring or electronic auctions. analysis can also he applied to the choice among
Although the auction system should provide tbe supply cbannel alternatives. Procurement chan-
appropriate price signals to encourage producers nels are characterized as bundles of attributes. A
to sell cattle of a consistently high quality, an indi- processor is assumed to evaltiate tbe "total worth"
vidual processor can never be sure of obtaining a of a channel by combining separate evaluations of
particular farmer's cattle at an acceptable price in the "part wortb" of each attribute. In this analy-
any one week because of the competitive nature of
auction markets. Direct dead-weight purchases
••Many producers, however, still prefer live-weight ]tayment
and distrust the system of dead-weight grades wlit-n selling dircel to
''The recent development of a numlier of national "farm processors. Tbey cannot withdraw cattle if the price is unsatisfactory
srhcmes" through whirh farmers rear their livestock according to and they faee greater risks of fcrade uncertainty, a» dinteuKxed earlier.
rertain animal welfan- cod«'i* iif practice may, however, reduw this For a discussion of LK producers' attitudes toward dead-weight will-
m(init(>rin|( cast. ing, see McLean—Builen."

514
Cattle Procurement

sis, the channel attributes are the transaction nuity of supply, tbe degree to wbicb animals were
costs of using the channel. The underlying model bandied when being transporteil from tbe farm to
is, therefore. tbe abattoir, tbe method of payment (hve-weigbt
or dead-weight), and the degree of traceability.
total evaluation of a channel; = pHrt worth of level i for transaetion Continuity of supply is important for two rea-
rust attribute l--' +
part worth of level_/ for transaction sons. If cattle are procured from long-term regu-
cost attribute 2--- + lar suppliers, the processor faces lower
part worth of level n for transaction information costs in searching out new suppliers.
cost attribute m.
Monitoring costs are also reduced because there
should be greater consistency in the quality of ani-
where the supply channel has m attributes, each mals entering the abattoir.
having two or more levels. Each scenario consists The transportation of cattle from farm to ahat-
of level i of attribute 1, level j of attribute 2, and toir is important for two reasons. First, the pro-
so on up to level n for attribute m. Tbe total eval- cessor's monitoring costs are reduced if animals
tiation (or total wortb) of a cbannel is therefore are transported directly from the farm to the
the Hum of the part-worth evaluations of each con- ahattoir. Less bandling of cattle means less stress,
stituent attribute level. Comparisons of tbe total less potential for bruising, and may also be more
wortb of different cbannel scenarios can then be acceptable to tbe processor's customers on animal
ma<le. welfare grounds. Second, if cattle are transported
Data for a conjoint analysis are collected through directly to the abattoir from farms, abattoir man-
a survey in which respondents evaluate a number agers tend to have more control over the timing of
of hypothetical scenarios. The scenarios are hased delivery to the plant. Tbis reduces their negotia-
on combinations of the product (or supply chan- tion costs.
nel) attributes. The contribution tbat each attri- The basis of payment affects the grade uncertain-
bute level makes to the respondent's evaluation of ty that processors face. Payment on a live-weigbt
the scenarios is calculated. A conjoint experiment basis may involve greater grade uncertainty than
c(tnsists of three phases: designing the scenarios, payment on a dead-weight carcass grade. Supply
collecting the data, and analyzing the results."* channels that require live-weight payment impose
the information cost of grade uncertainty on pro-
cessors.
Designing Channel Scenarios Finally, the traceability of cattle was identified as
an important factor in tbe cboice of supply cban-
The first step involves choosing tbe supply channel nel. Ease of traceability reduces the monitoring
attributes. It is necessary to limit tbe number of costs facing processors because they can more eas-
attributes to avoid overcomplicating the data col- ily check on-farm production standards. These
lection procedure. Problems occur if too many at- standards may be of concern to tbe processor's
tributes are used, altbougb there is a trade-off customers.B
between tbe number of attributes and the extent Tbe next step was to assign levels to the attri-
to which the scenarios are "realistic" for respon- butes. Hair et al."' suggest that each attrihute be
dents.'" Through a series of discussions with abat- assigned tbe same number of levels. The estimated
toir managers, the Ust of potential transaction cost importance of an attribute may increase as tbe
attributes was reduced to four.*^ Tbese were conti-
•^The diseussions with industry representatives were vital i>ecaiise four attributes, therefore, alt represent distinct aspects of the
the seleetUin of attributes is critical to the success of conjoint analy- processors' purchasing decisions. For examjilc, ciintinuity of HuppU
sis. The four attributes were chosen because they represented impor- focuses on whether suppliers are occasional or long-term regular sup-
tant aspects of processors' supply decisions, were easily pliers whereas tracealiility refers tn hnw easy or difficult it is for pro-
<ominunicated to respondents, and could be used to distinguish be- cessors to link specinc cattle with specific suppliers and how much
tween marketing channels.'^ information they have about the supplier.

515
Hobbs

number of levels increases because respondents fo- attributes and levels through close consultation
cus disproportionately on that attribute. For this with industry representatives.
reason, and to keep the presentation of alterna- In a survey of commercial users of conjoint
tives as simple as possible, each attribute was as- methodology, Cattin and Wittink'" raise a number
signed two levels. The levels were: for supplier of caveats to the use of this technique. Among
type, long-term reftular and orc-asional; for rattle them is the need to check the validity and re-
handlin<r, handled once between farm and abattoir Uabihty of the results. Some resean'hers test for
and handled more than once; for payment meth- reliabihty by asking respondents to evaluate one
od, dead-weight carcass grade and per kilo live or more of the scenarios twice at different times.
weight; for traceability, easy and diffi<'ult. The nature of this mail survey precluded tliis type
Different levels of these four transaction cost at- of reliability test. However, the validity of tbe re-
tributes were combined to prodnce hypothetical sults, in terms of the ability of the conjoint model
supply channel scenarios. The number of scenar- to predict individual preferenires accurately, was
ios depends on the number of attributes and the tested using hold-out scenarios. There were two
number of levels. A full factorial design inc hides additional supply channel scenarios that respon-
all possible combinations of attributes and levels. dents evaluated along with the other eight scenar-
This would have generated 16 scenarios (because ios. Data from the hold-out scenarios were not
there are four attributes with two levels each, 2^ used in the computation of the part worths for the
~ 16). It was felt that this woidd be too many for model. The validity of the model results for each
respon<lents to evaluate accurately. Instead, a individual was then tested by comparing each
fractional factorial design was generated using the individuaPs preference score for the two bold-
Categories procetlure of the SPSS computer pack- out scenarios witb the individual's actual prefer-
age,'^ This resulted in eight scenarios designed to ence scores. Tbe predictive accuracy of the aggre-
capture the main effects of each attribute. gate model with regard to the twa hold-out
An example of one of the supply channel scenar- scenarios t^ould also be tested. These tests gener-
ios is: type of supplier, regular; handling of cattle, ated Kendall's T and Pearson's R statistics that
once (direct from farm to processor); payment report the correlation between observed and j)re-
method, dead-weight carcass grade; and tra- dicted preference ratings for the two hold-out
ceability to farm of origin, easy. This might de- scenarios.
scribe purehases from individual farmers or from
a GMS. The scenarios do not, however, bave to
describe existing supply channels. Conjoint analy- Collecting Data
sis allows the presentation of hypothetical sceuar-
ios in order to measure tbe extent to which A postal survey of UK beef processing firms was
respondents will trade off one attribute for anoth- carried out in May 1994. A short ([uestionnaire
er. This can sometimes create credibility prob- collected data for the conjoint analysis together
lems. Because tbe scenarios are hypothetical, with summary information about the company. A
attribute levels must be plausible and capable of total of 255 companies were contacted. Ninety-
being traded off by respondents. They must be three usable responses were received, a response
communicated easily to respondents and realis- rate of 36%.
^jp 16,17 Eaph scenario was checked to ensure that The respondents represented a range of different
the combinations of attribute levels, although hy- types of com])anies in terms of size, ownership, lo-
pothetical, were realistic. The realism of the eval- cation, use of supply channel, product range, and
uation task required of respondents is crucial in customer type. Respondents were presented with
determining tbe response rate received from the the 10 scenarios (including the two hold-out sce-
survey and in determining the extent to which tbe narios) and asked to evaluate each scenario on a
results truly reflect individuals'' preferences. This preference rating scale of 1 to 9, wbere 1 repre-
also underlies the importance of carefully choosing sented least preference and 9 the highest prefer-

•616
-a -a •a •a T3 -o T3 V

efe rrei
•I

:rre

rre.
rre

efe rre

efe rre

rre
«

0) o J ©
:= «

tuatii
ires.
O- c_

• the
— c- c_
V) VI VI
o O O o o o c

'ferent
CN CS c^ CN CN CN CN CN CN CN
3
W CO OO oo oo 00 oo oo
1,

rat4
OO OO OO
•"^^
c o
r- r- r- r- r- r- r- r-
o so so so so c - 5
an so so NO so SO

>^

a ^ ^ I *i *
P 3 1" ^
k. m O — - I
o u ^ C •—'
"** e
(N fN CN CN OJ OJ fN (N O4 3} ^ M C
S at) ^ O
M S CO —
o XJ -a —
-o —
z XJ X © i i!
3.1

3.1

0.1

O.I
OJ
Si P P
:;^ ^ = 3

JOJ
u -^ ^ ? ">
D o tJ u u
c c- CL. 2-
3 to to t; V) to VI VI V)

h 1) re
1
— S - £
G " B
I- u: _
H g; -2 Si)
• - fa 3 "* Co
3 CO u

do ^X I
<s —
S ^S .=: -
i c *"
1101
Diff Icn

cj "^ r* ^ -= !5
Dili"

j*" fa
^^ ^ j Cii t/i VI VI
re re ." = o
H HC — u: LJ

CS • 3 _ o u JJ -J 3 I g «
O "cb 2 "cb "cJb t^ re TZ "cb "re O *^ bD fc
L.
fa 'S "5 « S X *a;
wei

V3 CO CO
^ t. O 7 L<
H o re VI
CQ QG U 0,1

Z y o > > >


re
VI
>
V)
re
> ra o fj (J p —• L> o,' S
SH o CO =0 CO re re re SO c5 ^ _o X E
Q
> H k-
;^ * Z. II
•B X X II
ON O — c a Q £ ^ bC _ .
2 « o _
plift]

plift)

plift)

§' o u C9 C
o o B CB
Oil

o o c c o o
u t) <u _ ,_
.— re
J H CO re
•~ "3
re '>* -J P-^ ^^

Q < 3 s "5-J c 1_
E0 C8
O w •a CJ D
r
.10

.10

•<I fa c o o c o
s o ° O i2 3
CQ
CO ^ r
'S
a;
1

II
1

L.

— - — -3 X
to W C3
-c _C
c "^ •J: c
,200 .2 _re .2 re ^5 re ca
1—
^ ^
C
X E
cn 0 L. u
c*3 C3 .—_'
(J re =0 3 3
> D to o
CO O oo C
•u o c
o O O o t, u
' 1

c
1

'^ CB 1)
tt
3
. 3 13T.
CASl

>•

o CO
00 ON * S
c
Hobbs

Figure 1 displays the survey question used basis, zero if on a per kilo live-weight basis; T
to collect the conjoint data and describes the 10 1 if traceability of cattle is easy, zero if diffi<rult.
scenarios. The regression equation is, therefore.

= C + o,S 4- + + +
Results
where R is the respondents' 1 to 9 ratings of ihe
scenarios, C is the constant, and |i, is the error
Averaged Results for Whole Sample
term.
Ordinary least squares (OLS) analysis was used to There were 744 observations in the data set be-
estimate the model. The dependent variable was cause the 93 respondents each provided eight ob-
the respondents' rating of each scenario. The in- servations (the two hoid-<tut strenarios were not
dependent variables (the attrihute levels) were used to estimate the model). Initially, the model
treated as dummy variables. The explanatory was estimated for each individual. This was pos-
variahles are defined as 8 — I if the supplier type sible because there were eight ohservations per re-
is long-term regular, zero if occasional; H — I if spondent. The individual results were averaged to
cattle are handled once, zero if more than once; P give results for the entire sample.'^'"^'
= 1 if payment is on a dead-weight carcass grade The OLS coefficients produ<:ed by this analysis
were transformed into part-worth evaluations of
the attribute levels hy the conjoint procedure in
weri- asked tn fvaltial<- rarh st-eiiurui on a jircfcr-
SPSS."'••** If an attribute was generally preferred
ence ratiii); HCUU* ratlicr than Ut'ing asked to rank thi^ srenarioh in ur- over another, it had larger part-worth estimates.''*
di-r af preference lieraiisr it was recof^ized that the use of ranking Table I rejiorts the averaged part-worth evalua-
sralcw ill mail surveys van make the survey instrument extremely dif- tions and relative importance of each attribute for
ficult to ut the whole sample, together with Pearson's R and

Tdile 1. Avcraircd Ciinjuinl Results for Whole Sample

Ri>lative Impiirlanie
Allribute Level Part-Horlh Kvalualiun of Attribule

Constant 4.7030
Supplier 17.90%
Long-term relationship with regular supplier 0.6411
Orcasinnal stipplier -0.6411
Handling 23.92%
(battle handled once from farm to abattoir 0.8535
(battle handled more than once from farm to abattoir -0.8535
Basis of Payment 19.44%
Df'ad-weight carcass grade 0.6949
Per kilo live weight -0.6949
Kase of Traceability 38.74%
Easy 1.3831
Diffuuh -1.3831
Pearson's R 0.995
Kt-ndall's T for two holdout;^ 1.0

Recall that results from individual respondents are averaged Io give the part-worth evaluationH fur the whole sample.

518
Cattle Procurement

T statistics for the model. Statistics close


to 1.0 indicate that the model is a good fit. The Table 1 . Predicted Prelerence Scores and Hanks.
Pearson's R correlation coefficient (0.995) mea-
sures the correlation between all observed and es- Predicted Prcdicled
timated preferences. Tbe accuracy with which tbe Scenarios Prpferencf Score Raik
model predicts eacb respondent's preference score
for the bold-out scenarios is compared to their ac- 1 3.9 5
tual preference scores. Tbe Kendall's T correlation 2 4.2 4
coefficient (l.O) reports tbe extent of this correla- 3 6.9 2
tion and confirms tbe vali<lity of the model. 4 2.8 7
The part-worth evaluations in Table I can be 5 2.4 8
6 8.3 1
Slimmed to give a total evaluation for any combi- 7 5.3 3
nation of tbe attribute levels. Hence, scenario 1
(occasional supplier., cattle bandied more tbau
a 3.8 fi

once, payment on a per kilo live-weigbt basis and TIu' sffnarios are as follows:
easy traceability of cattle to the farm of origin) t. Siipplipr, occaNiiinal: handling, more than onre:
would have a total evaluation (or total worth) of [laymenl, live weight; traifabiiity, <"asy.
2. Supplier, oecaaiona); handling, once; payment, deiid-
weight gra<li*: trarealiility. difrinilt.
(4.7030) + (-0.6411) + (-0.8535) + ( 0.6949)
3. Supplier, regular; liandling, once; payment, live weight;
+ (1.3831) - 3.9, traceability, easy.
4. Supplier. occaNiiinul: haniiling. onec; payment. live
which represents the sum of tbe attribute level weight; traceahility. diffiriill.
part worths and the constant term. On the other 5. Supplier, regular; handling, more than onee; payment,
band., scenario 6, which consisted of a regular tivt' weight; trareabilily. diffti-ult.
supplier, cattle handled just once from farm to 6. Supplier, regular; handling, once; payment, dead-wright
abattoir, payment on a dead-weight carcass grade grade; traceability. easy.
basis and easy traceability of cattle to tbe farm of 7. Supplier. orca«ii>nai; handling, more then once;
origin, has a total evaluation of payment, dead-weight grade: lra<'t'ability., easy.
8. Supplier, regular; handling, more than once; payment,
dead-weight grade: tmreability. dirficult.
(4.7030) + (0.6411) + (0.8535) (0.6949) +
Predicted rank calculated from the predicted preference
(1.3831) = 8.28.
geores.

Scenario 6 was, therefore, preferred to scenario 1.


Scenario 6 combined attribute levels that most re- ment and difficult traceabiUty of cattle. Tbey dif-
spondents said they would prefer. If we assume fer in tbat scenario 2 bas an occasional supplier
that the expected value of the transaction costs as- and direct transportation of cattle from the farm
sociated with each attribute were equal across all to the abattoir, while scenario 8 bas a regular sup-
levels, a risk averse firm would naturally choose plier but cattle are handled more than once. The
scenario 6. predicted preference scores for scenarios 2 and 8
Table II reports tbe predicted preference scores are 4.2 and 3.8 placing them fourth and sixtb, re-
for each of the eight scenarios (omitting tbe two spectively, in tbe ranking of scenarios. Tbis illus-
bold-out scenarios). Tbese were subsequently con- trates bow respondents traded off an occasional
verted to a predicted ranking of the eight scenar- supplier against additional bandling of cattle. It
ios. The predicted preference scores and rankings implies that respondents would be willing to give
demonstrate the way in which processors trade off up a degree of regularity of supply from a supply
between attribute levels wben making their pro- channel in return for a reduction in the amount
curement decisions. For example, scenarios 2 and that cattle are handled. The monitoring and nego-
8 bave in common dead-weight carcass grade pay- tiation costs that arise from the transportation of

519
Hobbs

Full Sample

Large Processors

Supplier Handling Basis of Ease of


Type Payment Traceability

Supply Channel Attributes


2. Relative importance of supply rhannel attributes for full of
processors compared with subsample of large processors.

cattle to the ahattoir may he more important tban rectly when heing transported to the slaughter-
the information anH monitoring costs that arise house, and in controlling the timing of
from locating suppliers. delivery to the slaughterhouse., are important. The
The attrihute with the widest range (from bigb to information costs arising from different payment
low) of part-worth estimates is the most important. methods (for exam])Ie, grade un( ertainty when
Figure 2 compares the relative importance of each huying cattle on a per kilo live-weigbt basis) are
of tbe four attributes. Clearly, for tbe full sample, slightly less important. Finally, the information
the most important attrihute in the choice of sup- and monitoring costs associated with searching out
ply channel scenarios was ease of traceahility new suppliers and ensuring a consistency in the
(39%) followed hy handhng of cattle (24%), meth- quality of suppbes appear to he tbe least impor-
od of payment (19%), and, finally, the type of re- tant of these transaction costs.
lationship witb tbe supplier (18%). Tbis suggests
tbat tbe monitoring costs arising from tbe tracing
of cattle to tbe farm of origin are the most impor- Results for Largest Processors
tant determinants of the choice of supply channel.
A fnrther reason for tbe importance of traceahility The UK slaughtering and processing sector bas
could he that processors fare higher uncertainty undergone con.siderahle rationalization in recent
regarding monitoring costs that arose as a result years. The numher of abattoirs fell from 1890 in
of the 1990 FSA. While processors bave heen deal- 1971 to 1972 to 552 in 1993, with further rational-
ing, for some time witb tbe issues of payment meth- ization occurring.'** There has heen a correspond-
od, regularity of supply, and handling of cattle, ing rise in the scale of production. Consequently,
traceahility is a relatively new concern for tbe UK industrial concentration in this sector has in-
heef industry. Tbere has heen considerahle uncer- creased. Processors that slaughter over 30000 cat-
tainty surrounding the bkely outcomes of the FSA tle a year, while representing only 4% of total
for food firms, and few legal precedents existed ahattoirs, account for approximately 4 1 % of na-
when the Act was first passed, i*' This may partly tional slaughterings. These are the most important
explain the increased importance given the tra- firms in the industry in terms of their influence on
ceahility attrihute over the other three attributes. procurement trends. To reflect this, the sample
Second, the monitoring and negotiation costs as- was segmented according to the size of the ahatioir
sociated witb ensuring that cattle are handled cor- as measured hy average weekly slaughterings. A

520
Cattle Procurement

Table III. .Averaged lonjuinl Rt^sulU f( r Subsample of Larjfi' Pruressors

Rflalivi' Impiirlanfe
Atlribnto Level Parl-Wortb l*]valaalioii (if Altrjbuli'

Constant 4.6071
Supplirr 12.36%
Loiifi-term relationship with regular supplier 0.5119
Ocrasional sui)plier -0.5119
Han<lliii<: 20.69%
Cattle handled once from farm to ahalloir 0.8571
Cattle handled more than onee frtim farm to abattoir -0.8571
Basis of Payment 23.85%
Deatl-weipht carcass grade 0.9881
Per kilo live weight -0.9881
Ease of Traceahility 43.14%
Easy 1.7857
Difficult -1.7857
Pearson's R 0.994
KenddlTs T for two holdoiit^^ 1.0

conjoint analysis was carried out on the data from handling of cattle attribute. Tbe information costs
the 21 largest processors in the sample (slaughter- arising from grade uncertainty appear to be more
ing over 1000 (tattle, on average, per week). important to larger processors than to processors
Table III reports the averaged results for tbe as a whole.
subsample of large companies. The Pearson's R
and Kendall's T statistics indicate that tbe model is
a good fit. Clearly, the major difference between Discussion
the subsample and tbe full sample results is tbe in-
creased importance placed on tbe traceability at- The ease with which processors could trace cattle
tribute by the larger processors. This can also be back to the farm of origin was clearly the most im-
seen in Figure 2. The increased importance of tra- portant of the tran.saction cost attributes. This
ceahility appears to have been traded off against a was the case for the full sample and for the sub-
reduced importance for supplier type. Super- sample of larger processors. Ease of traceabiiity
markets in the United Kingdom now require more reduces monitoring costs for processors. This has
information from beef pro<'essors about tbe origin become particularly important in the United King-
of cattle and the conditions under which animals dom sint-e the introduction of the 1990 FSA.
were reared. Hence, traceability of cattle is likely Shortly after the FSA came into force, indications
to be important for tbose processors who supply that the monitoring costs arising from traceahility
supermarkets. Twenty of the 21 large processors in might alter vertical coordination in the beef mar-
the suhsample sold products to supermarkets. For keting chain began to emerge^": "According to the
all but tbree of tbese processors, supermarkets ac- Meat Industry Liaison Group (MLG), auction mar-
counted for at least 50% of their beef sales. kets are in danger of being declared a "no-buy"
Wben compared witb the results for the full sam- area hy powerfnl supermarket companies as they
ple, a second change is the increased importance prepare to meet the supply (-hain audits demanded
of the method of payment attribute relative to tbe hy the 1990 Food Act . . . Supermarket buyers

521
Hobbs

. . . biggest objection is the way animals sold un- tant attribute for the sample as a whole (the
der the hammer lose their identity." (N/5) second most important for the larger processors).
Furthermore, one UK supermarket meat buyer Responses to other survey questions revealed that
was quoted as saying^**: "This means we have to dead-weight carcass grades were by far the pre-
know where our animals have come from and how ferred method of payment by processors. Payment
they were managed. This cannot he done through on a live-weight basis places the risk <tf grade un-
the auction system. As soon as we can establisb a certainty on the pnM-essor. Live-ring aui'tions re-
network of three-cornered quality assurance part- quire payment on a live-weight hasis and,
nersbips with farm-groups, abattoirs and our- therefore, impose higher information costs on the
selves, we will refu,se to handle any auction processor. Direct purchases from individual farm-
animals. We will eventually tell abattoirs tbey ers through group marketing schemes and through
must not give us auction stock." (N/6) This sug- most UK electronic auction systems use a dead-
gests that auction companies should take steps to weight carcass grade payment system.
improve the traceability of cattle in order to an- Finally, the relationship with the supplier, al-
swer these concerns. The supply channel scenarios though not as important as the other three trans-
were hypothetical and were designed to elicit pro- action costs, did influence the selection of supply
cessors' preferences toward different supply chan- channels. The survey revealed that long-term reg-
nel chara<'teristics. Whether the behavior of ular supi)liers were preferred to oc<'asional sup-
processors changes in response to the traceahility pliers. Processors incur lower information costs if
requirements of the FSA and consumer animal they do not constantly have to seek out and assess
welfare concerns may not be clear until the uega- new suppliers. The quality of cattle entering the
tive consequences of not changing behavior be- abattoir sbould also be more consistent. Generally
come known. The consequences for vertical speaking, direct purchases from individual pro-
coordination in the UK beef marketing chain have ducers and from GMSs are tbe supply cbannels
yet to be fully realized. characterized by long-term regular suppliers.
The extent to which cattle were handled during Live-ring and electronic auctions create occasional
transportation from the farm to the abattoir was supply relationships as far as an individual pro-
the second most important attribute for the full cessor is concerned. There is no guarantee that a
sample (the third most important for the larger processor will consistently be able to purchase cat-
processors). Minimized handling of cattle reduces tle from preferred producers each week at an ac-
the potential for carcass damage and shrinkage re- ceptable price.
lated to stress; this may also he more acceptahle
to consumers on animal welfare grounds. The re-
sult is lower monitoring costs for processors. Cat- Conclusions
tle procured direct from individual farmers or
through a GMS tend to he handled less than those TCE provides many useful insights into the struc-
sold through live-ring auctions. Electronic auction ture and organization of industries, suggesting
systems differ in the extent to which cattle are that, all things being equal, industries will be or-
handled. Under most of the UK electronic auction ganized to minimize the cost of carrying out trans-
systems, cattle are transported dire( tly from the actions. This article provided a transaction cost
farm to the abattoir, thcrehy reducing this mon- explanation of the choice of supply channel by
itoring cost. Those electronic anction systems re- beef processors in the United Kingdom.
quiring farmers to deliver cattle to a central The results of the conjoint analysis indicate that
collection point before transportation to the buyer's the monitoring costs arising from the traceability
plant impose higher monitoring and negotiation of cattle are important to tbe selection of beef sup-
costs on the buyer. Hence, they would be less ap- ply cbannel(s). In the United Kingdom, pressures
peaUng as vertical coordination alternatives. for greater traceability of cattle and tbe need to
The basis of payment was the third most impor- provide consumers with farm and quality assur-

•S22
Cattle Procurement

ance guarantees suggest that the monitoring costs respond to processors* concerns, the long-term
that processors (and retailers) incur through auc- trend of decUne in the share of slaughter cattle
tions and i)ccasional supply relationships with in- marketing through live-ring auctions is likely to
dividual producers may become prohihitive. continue. Long-term stable relationships between
Instead, there will he increasing pressure from farmer marketing groups, processors, and re-
downstream firms to move toward closer forms of tailers (such as strategic alliances or formal con-
vertical coordination. Unless auction markets can tractual relationships) may emerge.

References

1. R.L. Mighell and L..4. Jones., Vertical Coordination in 1990?" Agribusiness: An International Journal, 8, 575
Agricnlturp, USDA ERS-19. Washington. DC, 1963. (1992).
2. A.D. Barkema and M. Drahenstott. "A CroBHroads for 11. A. MeLean-BuUen. "Selling Livestock—By Liveweight or
the Catlle Industry." Economic Rei'ieu: [Federal Reserve Deadweight." Farm Management, .5. 169 (1984).
Bank of Kant^aHJ, November-December, 47 (1990). 12. K.J. Lancaster. "A New Approach to Consumer Theory,"
3. S.N.S. Cheung, "On the New Institutional Erononiirs," in Journal of I'olitical Economy, 62, 132 (1966).
Contract Economics, L. Werin and K. Wijkander, Eds., 13. J.B.E.M. Steenkanip. "Conjoint Measurement in Ham
Blarkwell, Oxford, UK. 1992, p. 48. Quality Evaluation/' Journnl of Agricultural Economics,
4. R.H. Coaae, "The Nature of the Firm/" EconomM^a, 4, 38, 473(1987).
3«6 (1937). 14. B.E. Prentice and D. BennelK "Determinants of Returns
5. O.E. Williamson, Economic Organization — Firms, Mar- hy U.S. Refrigerated Trucks: Conjoint Analysis Ap-
kets and Policy Control, Harvester Wheatsheaf, Heniel proach," Canadian Journal of Agricultural Economics,
Hrmpsteail. Hertfordshire. UK. 1986. 40, 109(1992).
6. C E . Ward. "Market Strurture Dynamics in the Live- 15. W.L. Adamowicz, P.C Boxall. J.J. Louviere. J. Swail,
stock-Meat Subsertor; Implications for Pricing and Price and M. Williams. "Stated Preference Methods for Valuing
Reporting," in W, Purcell and J. Rowsell. Eds.. Key Is- Environmental Preferences." in D.I. Bateman and K.
sues in Livestock Pricing: A Perspective for the 1990s, Willis. Eds., Valuing Environmental Preferences: Theory
Research Institute on Livestock Pricing, Btack^burg. VA, and Practice of the Contingent Valuation Method in the
1987, p. 8. L'S, EC and Developing Countries. Oxford University
7. W.G. Tomek. "Price Behavior on a Declining Terminal Press. Oxford. UK. 1995.
Market." American Journal of Agricultural Economifs, 16. J.F. Hair. Jr., R.E. Anderson. R.L. Tatham, and W.C.
62,434(1980). Black. Multivarinte Data .Analysis with Readings, 3rd
8. D.M. Feuz, S.W. Fausti. and J.J. Wagner, "Analysis of ed., Macmillan, [\ew York, 1992.
the Efficiency of Four Marketing Methods for Slaughter 17. P. Cattin and D.R. Wittink, "Commercial Use of Con-
Caltlc." Agribusiness, 9, 453 (1993). joint Analysis: A Sut^ey/' Journal of Marketing, 46, 44
9. S.W. Fausti and D.M. Feuz, "Production Uncertainty (1982).
and Factor Price Disparity in the Slaughter Cattle Mar- 18. SPSS, SPSS Categories, SPSS Inc., Chicago. 1990.
ket: Theory and Evidence." American Journal of Agri- 19. MLC, The Abattoir Industry in Great Hritain. 1994 edi-
cultural Economics, August. 533 (1995). tion. Meal and Livestock Commission, Milton Keynes,
10. J.E. Hohhs and W.A. Kerr. "The Coat of Monitoring UK, 1994.
Food Safety and Vertical Coordination in Agrihusiness: 20. Agra Europe, "UK Supermarkets Want Meat Husbandry
What Can Be Lt-arned from the British Food Safety Act Warranty". Agra Europe, May 10, N/5 (1991).

523

You might also like