Home > Supply Chain Management: Make-or-buy decision making
Supply Chain
Management: Make-Or-Buy Decision Making
Stefan Helber
Technical University of Clausthal
Julius-Albert-Str. 2
D-38678 Clausthal-Zellerfeld
Germany
Tel. ++49 5323 953610
Fax. ++49 5323 953699
stefan.helber@tu-clausthal.de
Abstract:
This paper
discusses some of the mechanisms that determine make-or-buy decisions.
The short- and the long-term perspective are treated separately. In
the short term, cost or profit comparisons can be used to determine
what to “make” and what to “buy.” More important, however, is
the long-term question of who should make the often specific investments
required for production, transportation, and inventory processes.
According to Coase and Williamson, the boundaries of the
firm should be determined such that the costs of coordinating economic
activities are minimized. They depend on the nature of the required
assets as well as on the importance, frequency, and uncertainty of the
economic transactions along the supply chain. Based on transaction cost
theory, some rules of thumb can be derived of what to make and what
to buy.
Keywords:
Make-or-buy decisions, vertical integration, transaction cost theory,
Supply Chain Management
What is Supply
Chain Management? It appears to deal with the coordination of tasks
required to make products and to get them to the customers. In Figure
1, this includes production, storage, and transportation activities.
These activities have to be coordinated along the whole chain, possibly
over the boundaries of multiple firms.
Figure
1: Idealized supply chain
Is this a new
phenomenon? Hardly. Supply chains have been managed ever since human
beings discovered that they are more productive when they specialize
and trade the goods they produce. The coordination of the resulting
processes has now been called “logistics” for quite a time, and
it is save to say that many of the logistics problems turned out to
be non-trivial.
How can supply
chains be coordinated? From a very general point of view, one possibility
is to use the market mechanism. The price for, say, a barrel of crude
oil coordinates the activities on oilfields, refineries, and gas stations.
This is Adam Smith’s “invisible hand.” Very often, however, we
think of coordination in the context of organizations that have a hierarchical
structure. We think of firms that coordinate production and logistics
processes internally. When we talk about make-or-buy decisions in the
context of Supply Chain Management, we want to know which of these options
is chosen, i.e. whether a market or a hierarchy coordinates subsequent
stages in a supply chain. This is the question for the boundaries of
the firms. Who should do what, and why? Which processes should be coordinated
internally, and where should the market govern?
Now why is there such a lot of new interest and enthusiasm about a bunch of old and notoriously difficult problems? The conventional wisdom is that the rapid progress of transportation, information and communication technologies makes different organizational arrangements efficient, and this affects the make-or-buy decisions within the supply chain. If transportation costs decrease, it becomes efficient to bundle production quantities for similar or identical goods at some few places in the world due to economies of scale. If it becomes easier to exchange and process information, it is possible to reduce information asymmetries between the acting parties. This should also make coordination over the market more easy. As a result, we might expect that the degree of vertical integration along the supply chain decreases, i.e. that firms start to buy what they used to make. If we want to analyze in more detail what drives these decisions, it is useful to keep the short-term and the long-term aspects of the problem apart.
The short-term
version of the make-or-buy problem arises if we as well as somebody
else can perform a task. This type of problem is usually solved by a
straightforward cost comparison. Assume that making some part would
lead to direct (variable) costs of $100. We could as well buy the part
for $200. This includes the other company’s direct cost, the indirect
costs (overhead) and some profit. The question whether we should “make”
or “buy” depends on our other opportunities to use our resources.
If making the part would take us one day, and we have nothing else to
do this particular day, we should decide to “make,” because the
loss $100 is less than one of $200.
Now assume that we had something else to do: We could make another part, with variable costs of $120, and sell it for $250. However, in this case we would have to buy the first part elsewhere for $200. Now it is worthwhile to buy the first part and make the second as the simple profit comparison in Table 1 shows.
Without alternative job | With alternative job | |||
Make | Buy | Make | Buy | |
Revenue | - | - | 250 | |
Variable cost | 100 | 200 | 100 | 120+200=320 |
Profit | -100 | -200 | -100 | -70 |
Table
1: Example of a short-term profit comparison
Note that the
cost comparison did not include any fixed overhead. When products are
sold, the revenue has to cover some fraction of the overhead. If they
are sold within the firm using some transfer price, the buying part
of the firm might be charged more than the variable cost. If the buying
part is charged say $210 which includes $110 of overhead and profit
for the producing unit, the buying part of the company may be reluctant
to “buy internally,” i.e. to have the own company make the product,
even though this is be the right thing to do for the firm as a whole.
This shows that it is not trivial to find transfer prices that induce
the “right” decisions.
This type of make-or-buy problem arises only if both the own and another firm can perform a task. While in the short-term only variable costs are relevant (which does not include, for example, any depreciation for long-term investments like machines or trucks), the long-term picture may be completely different. Here the question is: Should we build up capacities for some type of process, or should this be done buy some other firm? And why?
The long-term
make-or-buy problem is the question of the boundaries of the firm: What
should we do, what should somebody else do, and why? To uncover some
of the forces that drive make-or-buy decisions, it is useful to start
with some examples that may at the first glance look childish.
Imagine the
situation of a scientist planning to attend a workshop in a foreign
country. Assume that the only relevant way for our scientist to travel
to the foreign country is to fly. In many cases, the scientist will
chose to “buy” (a ticket), without ever considering to “make”,
i.e. to fly in a plane she owns. Why does the scientist not own an airplane,
or a (small) fraction of an airplane? Why is it more efficient to have
airlines organize air traffic?
Now assume
that our scientist is traveling back. On her way from the airport to
her home she might use a car of her own. Why is she using a car of her
own? If we observe airline that own airplanes, why are most cars
not operated by “car-lines” (even though some are, through car
rentals and taxi companies). Why might it be more efficient not to have
“car-lines?”
Let us consider
a more serious example. Assume that a machining company uses, among
other things, bolts and a component that is used only in the machines
this particular firm produces. Assume furthermore, that special production
facilities are required to make these components. What will the firm
buy, and what will it make? One might argue that the firm will buy the
bolts because they are available on the market whereas the special components
are not and have hence to be made. A slightly more sophisticated answer
might be that large economies of scale can be realized in the production
of bolts and that it is therefore efficient to have “somebody else”
make the bolts. However, putting the monopoly problem aside, one could
image to merge all the bolt-producing companies to realize the maximum
possible economies of scale (as in the large factories of the former
Soviet Union) and furthermore all the companies that use bolts. With
respect to economies of scale, having large firms that produces (almost)
everything looks like a very attractive way to organize production.
Eventually one has to ask with the 1991 Nobel Prize winner Ronald
Coase: “Why is not all production carried out in one big firm?”2
The answer
to this question has been given by Commons, Coase, and Williamson.3
They argue that in a world with division of labor, the economic activities
of many individuals have to be coordinated. This includes the motivation
of the involved people. This coordination and motivation4
causes costs that can be compared to the phenomenon of friction
in mechanical systems. Just as mechanical systems are designed to minimize
friction, economic systems evolve in a way that the costs of drafting,
negotiating, and safeguarding agreements about production processes
are minimized. These costs are called transaction costs5
as they are related to the transfer of goods and services within the
supply chain. Note that these so-called “costs” are rather intangible,
compared, for example, to direct labor. For this reason, a transaction
cost analysis will usually not be a quantitative comparison like
the cost comparison for the short-term make-or-buy problem. Instead,
the economist usually compares different institutional arrangements
that describe how transactions could be assigned to governance structures.6
With a theory about what determines transaction costs one can develop
rules of thumb about what to make and what to buy, i.e. where the boundaries
of the firm should be.
The basic question is for the governance structure that coordinate economic activity. There are two extremes: One is the market system where the price mechanism serves to coordinate plans. This affects the buy-part of the problem. On the other hand, however, plans can be coordinated through hierarchies, i.e. within firms (or households). The question is: What should be coordinated how, and why? The answer depends on our assumptions about human behavior on the one hand and on the nature of the involved assets on the other. Their nature determines how transactions can or have to be organized.
In many production
or transportation processes in the supply chain, specialized equipment
is used. Once this equipment has been installed, it can be used for
a long time, but often only for a very specific purpose. Consider the
supply chain in the oil industry: One important means of transportation
are pipelines. If they are given proper maintenance, they can be used
for many years, but only for one particular purpose. It might
not be economically worthwhile to de-install a pipeline somewhere and
re-install it elsewhere, even if this were technically feasible. Now
consider the cars that the company uses. If they are no longer needed,
somebody else might use them. Cars can be rented as well as bought and
sold. There is a large market for used cars, but probably not for used
pipelines. The investment in the pipeline is therefore specific, whereas
the investment in cars is not. The first question a manager dealing
with a make-or-buy problem, whether in a supply chain context or elsewhere,
is hence whether this part of the chain requires specific assets.
The reason is that whenever specific assets are needed, the coordination
problems are more difficult than when the required assets are not specific.
Now why is this? The answer comes with some hard-boiled assumptions
about human behavior.
The first assumption
about human behavior is that while humans try to act rational
in the sense of maximizing measure of individual utility, their ability
to do so is bounded as they can process only a limited amount
of information.7 Processing information requires time
and effort, so problem-solving is imperfect and costly (as any operations
researcher knows). If human rationality is bounded, it is not possible
to make arrangement for all possible events and actions. In this case,
negotiating a long-term investment like a pipeline is surely a difficult
task. However, humans do make long-term decisions, for example when
they marry. When people marry, they basically “agree to agree” without
writing a contract about all possible future courses of action. They
agree to “work things out.”8
The second
central assumption about human behavior is that humans act opportunistic.
Whenever it is possible to pursue the own interest at somebody else’s
expense, at least some people might do so. While this includes behavior
such as lying, stealing, and cheating, the more subtle forms are no
less important. Will people drive more careful with their own car or
with a company car? Will they be more careful about an apartment they
own or about one they rent? Will they take more risks when they know
that their insurance pays if necessary? Will the trainee stay with his
company after his training is over, or will he change to a competing
firm that can pay higher salaries to already trained employees because
it does not offer any training itself? Very often, information asymmetry
is involved: one party knows more than the other, and it is willing
to use this superior knowledge. In this case, a general “agreement
to agree” will not work.
Finally, it
is important whether the transaction along the supply chain is frequent
or only occasional. We can furthermore assume that markets provide very
strong incentives, whereas in hierarchies incentives to be efficient
tend to be weaker. Most people will not fire an employee as quickly
as they may turn to a cheaper gas station. Coming back to the make-or-buy
problem, some preliminary results can be summarized in Figure 2.
Figure
2: Efficient governance
If the process
for which we have to solve the make-or-buy problem requires only nonspecific
assets, it may be a good idea to use the market to coordinate economic
activities, i.e. to buy the good or service. Computer programs for word
processing are an example. They can be used in many places, different
comparable programs are available, and there are numerous tests providing
information about what the program can do and what not. Bolts are another
example. Their properties are relatively clear, and in both cases there
are strong economies of scale to realize if specialized producers make
word processors and bolts and the strong incentives of the market system
are used to reward or punish them.
If, on the
other hand, the investment is highly specific, like the oil pipeline
mentioned above, it may be extremely difficult to find a partner who
invests in the pipeline. The reason is that all he can after
he has invested the money is to transport our oil. Such a partner
would run into a “hold-up” situation: After he has build the pipeline,
we might tend to unilaterally reduce his revenues for transporting our
oil, and there would be little he could do about it because there would
be no other way to use this pipeline. He could not even credibly threaten
to end the relationship. Since any partner would foresee these problems,
negotiating and enforcing an agreement would be very difficult.
For this reason,
it might be more efficient to have the extraction and the transportation
of the petroleum under unified governance, i.e. the oil-extracting company
should own the pipeline. Market mechanisms will not work well in this
case: We have to make, i.e. to transport the oil ourselves.
There is an
area in the middle where the investment characteristics are mixed and
transactions occur only occasional. In this case a so-called neoclassical
contract law can provide for third-party assistance to solve disputes
without breaking off the relationship.9
Another example from the context of Supply Chain Management is that while the investment in a highly automated inventory storage system close to the production line is extremely site-specific, the investment in trucks required for transportation processes is not. For this reason, it might be more difficult to outsource inventory storage activities than transportation activities.
Though it is
usually not possible to quantify transaction costs exactly, one should
nevertheless try to find an institutional arrangement that minimizes
the sum of production and transaction costs. Production costs depend
on economies of scale: If a good can be produced in large numbers, the
per-unit cost will tend to be low.
Let G
denote the cost advantage of buying the good that is due to economies
of scale. Since the other producer makes larger quantities than we would
make, he can produce cheaper, so this cost advantage of buying G
is always positive. However, it decreases as the good becomes more specific,
i.e. the measure k grows in Figure 3. Now consider the transaction
cost advantage of buying the product. If the good is completely unspecific
(k=0), coordinating plans using the market mechanism is less
costly than the planning process we would need to make the product ourselves.
That is, there is a positive transaction cost advantage C
for (k=0).
Figure
3: Production and transaction cost advantages (Williamson 1985, p. 93)
However, as
the good or service becomes increasingly specific, the transaction cost
advantage of buying instead of making in Figure 3 decreases and eventually
becomes negative. So does the sum of production and transaction cost
advantages G
+ C.
At some critical point k*, this sum is zero. If a good is very
specific (k>k*), we should make it ourselves, if it
is not (k<k*), we should buy it.
To solve real
make-or-buy problems, one needs tools that help to structure the relevant
alternatives. In the given context, we are trying to assign a set of
processes to some kind of make-or-buy strategies. One popular type of
tool is the two-dimensional portfolio representation that helps to develop
idealized strategies. The first dimension is the “strategic relevance”
of the process under study. The idea is that only the strategically
relevant processes should be performed by the own firm. Strategically
relevant processes tend to be highly specific, be repeated frequently,
and be subject to considerable uncertainty. They may require knowledge
a firm does not want to share, knowledge that constitutes a competitive
edge. Other processes, on the other hand, may be of little strategic
relevance, for example the services of the janitors or cleaners who
take care of the firm’s office building. A specialized facility management
firm should be able to do this better than any firm in a different business.
These processes are rather unspecific, and they should be bought. In
a real-world context, however, there may be barriers that make in- or
outsourcing of any given process difficult.
If we try to
outsource a process of little strategic relevance, we may not find a
partner with the required knowledge and capital. The employees related
to this process will usually not appreciate to be outsourced. Similar
problems can occur with respect to insourcing a process.
Figure
4: Idealized strategies for previously external processes (Picot 1991)
Figure
5: Idealized strategies for previously internal processes (Picot (1991),
Gerhard et al. (1992))
In Figure 4
we consider insourcing. If a process is highly specific, uncertain and
of strategic relevance, and the difficulty of insourcing the process
is low, the firm should choose to “make.” We might choose to buy
the respective firm. If this is not possible, i.e. the difficulties
of insourcing are high, we may try to form some type of strategic alliance
or to exchange shares in order to make future negotiations less difficult
due to a common interest. If the process is of little strategic relevance,
we might try to develop additional suppliers to increase the competition
in this area.
The options for outsourcing are displayed in Figure 5. If an unspecific process of little relevance can be easily outsourced, we should decide to “buy.” If outsourcing is difficult in the short term, we should try to develop suppliers. If a process is highly important and specific, we will not want to outsource the process. This process constitutes the core of our business. If outsourcing this process would be difficult, we should continue therefore to “make.” If, however, outsourcing would be easy, we should seriously reconsider our competitive position. A competitor or a firm up- or downstream in the supply chain might decide to take this process over, therefore our competitive advantage might not be sustainable.
Make-or-buy
decisions in the supply chain context should take the effort of coordinating
activities along the chain as well as the incentives for the acting
individuals into consideration. These transaction costs depend on the
specificity of the required assets. The rule of thumb for the manager
of a supply chain is: “Buy” the things non-specific and less important,
and “make” those that are highly specific and strategically important.
For those processes between these to extremes, consider long-term contracts,
the exchange of share, or other types of strategic alliances.
References
Coase, R.: The Nature of the Firm, in: Economica 1937, S. 386-405
Commons, J. R.: Legal Foundations of Capitalism, New York 1924
Commons, J. R.: Institutional Economics, in: American Economic Review 1931, S. 648-657
Furobotn, E. G./Richter, R.(Hrsg.): The New Institutional Economics, Tübingen 1991
Gerhard, T./Nippa, M./Picot, A.: Die Optimierung der Leistungstiefe, in: Harvard Manager 1992, S. 136-142 (in German)
Macneil, I. R.: Contracts: Adjustment of Long-Term Economic Relations under Classical, Neoclassical and Relational Contract Law, in: Northwestern University Law Review 1978, S. 854-905
Milgrom, P./Roberts, J.: Economics, Organization and Management, Englewood Cliffs/NJ 1992
Picot, A.: Ein neuer Ansatz zur Gestaltung der Leistungstiefe, in: Zeitschrift für betriebswirtschaftliche Forschung 1991, S. 336-357 (in German)
March, J./Simon, H.: Organizations, 2. Aufl., Cambridge/MA et al. 1993
Williamson,
O. E.: The Economic Institutions of Capitalism, New York et al. 1985
1 See Williamson (1985)
2 Cited according to Williamson (1985), p. 133.
3 Commons (1924, 1931), Coase (1937), Williamson (1985)
4 Organization can be interpreted as the task of coordinating and motivating simultaneously, see Milgrom/Roberts (1992). See Furobotn/Richter (1991) for an overview of the New Institutional Economics.
5 Williamson (1985), p. 20.
6 Williamson (1985), p. 41.
7 The concept of bounded rationality is due to Simon, see March/Simon (1993).
8 Williamson 1985, p. 60
9 MacNeal (1978)
All Rights Reserved Powered by Free Document Search and Download
Copyright © 2011