Beruflich Dokumente
Kultur Dokumente
subscribe to many of the same good ideas, 2. Four Levels of Social Analysis
a progressive research program requires
more. Kenneth Arrow speaks to the It will be useful for purposes of per-
transformation as follows (1987, p. 734): spective to consider the four levels of
social analysis that are distinguished in
Why did the older institutionalist school fail Figure 1. 3 The solid arrows that con-
so miserably, though it contained such able nect a higher with a lower level signify
analysts as Thorstein Veblen, J. R. Commons,
and W. C. Mitchell? I now think that . . .
that the higher level imposes con-
[one of the answers is in the] important spe- straints on the level immediately below.
cific analyses . . . of the New Institutional The reverse arrows that connect lower
Economics movement. But it does not consist with higher levels are dashed and signal
of giving new answers to the traditional ques- feedback. Although, in the fullness of
tions of economics—resource allocation and
the degree of utilization. Rather, it consists
time, the system is fully interconnected,
of answering new questions, why economic I mainly neglect these feedbacks. The
institutions emerged the way they did and NIE has been concerned principally
not otherwise; it merges into economic his- with levels 2 and 3.
tory, but brings sharper [microanalytic] . . . The top level is the social embedded-
reasoning to bear than had been customary.
ness level. This is where the norms, cus-
There is no question that the NIE toms, mores, traditions, etc. are located.
has grown in stature and influence over Religion plays a large role at this level.
the fourteen years since Matthews’ pro- Although Level 1 analysis is undertaken
nouncement. Initial skepticism has by some economic historians and other
gradually given way to respect—it being social scientists (E. C. Banfield 1958;
the case that economists are very prag- Robert Putnam, Robert Leonardi, and
matic people. Tell them something dif- Raffaella Nanetti 1993; Samuel Hunt-
ferent and consequential about phe- ington 1996; and Victor Nee 1998),
nomena that are of interest to them and Level 1 is taken as given by most insti-
demonstrate that the data are corrobo- tutional economists. Institutions at this
rative: that will get their attention. The level change very slowly—on the order
NIE has progressed not by advancing an of centuries or millennia—whereupon
overarching theory but by uncovering Douglass North poses the query, “What
and explicating the microanalytic features is it about informal constraints that gives
to which Arrow refers and by piling them such a pervasive influence upon
block upon block until the cumulative the long-run character of economies?”
value added cannot be denied. (1991, p. 111).
The NIE, moreover, will not stand North does not have an answer to
still. Even as institutional economics is that perplexing question, nor do I. The
being incorporated within orthodoxy, concept of “embeddedness,” both at the
new opportunities and challenges await. level of society and in the context of on-
Both unfinished business and new proj- going network relations, has been ad-
ects yet to be undertaken await the new vanced to help explicate these issues
millennium. (Granovetter 1985). The vast literature
I begin with a sketch of four levels of on culture (Paul DiMaggio 1994) is also
social analysis, next turn to some of the pertinent. Neil Smelser and Richard
good ideas out of which the NIE works, Swedberg discuss these and related issues
and then examine some of the applica- in their introduction to the Handbook
tions to which the NIE has been put. 3 This framework was first set out in Williamson
Concluding remarks follow. (1998).
Williamson: The New Institutional Economics 597
unavoidably incomplete. But while there flict, mutuality, and order to which I
is near-unanimity that complete contin- referred earlier—is centrally implicated.
gent claims contracting is impossible, Feasibility. Students of the NIE es-
the appropriate way to model incom- chew hypothetical ideals—which work
plete contracts remains controversial. off of omniscience, benevolence, zero
Lack of agreement on the definition transaction costs, full credibility, and
and operational import of bounded the like—and deal instead with feasible
rationality is a major obstacle (Ariel organizational alternatives, all of which
Rubinstein 1998; David Kreps 1999). are flawed. Coase (1964) and Demsetz
Contractual incompleteness poses (1969) were among the first to take ex-
added problems when paired with the ception with the asymmetric standards
condition of opportunism—which mani- that were once used in the “market fail-
fests itself as adverse selection, moral ure” literature—according to which
hazard, shirking, subgoal pursuit, and markets are beset with failures whereas
other forms of strategic behavior. Be- “omniscient, omnipotent, benevolent”
cause human actors will not reliably dis- governments (Avinash Dixit 1996, p. 8)
close true conditions upon request or would reliably administer efficacious
self-fulfill all promises, contract as remedies. As we all should have recog-
mere promise, unsupported by credible nized (but needed to be told), all feasi-
commitments, will not be self-enforcing. ble forms of organization—government
But for opportunism, the courts included—are flawed. 4
would simply ask witnesses to “tell us What I have referred to as the reme-
what you know that is germane to our diableness criterion is intended to rec-
decision.” That is not, however, the way tify this asymmetric state of affairs. This
that testimony is taken. Witnesses are criterion holds that an extant mode of
required to take an oath to “tell the organization for which no superior fea-
truth, the whole truth, and nothing but the sible alternative can be described and
truth”: don’t lie; don’t conceal; don’t implemented with expected net gains is
mislead. Inasmuch, moreover, as oaths presumed to be efficient.
are not self-enforcing, penalties for per- To be sure, public policy analysis be-
jury remind witnesses that prevarication comes more complicated when analysts
has consequences. can no longer condemn extant modes
Still a third attribute of human actors because they deviate from a hypotheti-
warrants remark, and that is the capac- cal ideal, full stop. The remediableness
ity for conscious foresight. Indeed, as criterion presses the public policy ana-
Richard Dawkins observes, it is the “ca- lyst to display a superior feasible alter-
pacity to simulate the future in imagina- native. If, moreover, a proposed feasi-
tion . . . [that saves] us from the worst ble alternative cannot be costlessly
consequences of the blind replicators” implemented, then the costs of imple-
(1976, p. 200). Parties to a contract mentation are appropriately included in
who look ahead, recognize potential the net benefit calculus—which has ma-
hazards, work out the contractual rami- jor ramifications for the path depen-
fications, and fold these into the ex ante dency literature. Finally, grounds for
contractual agreement obviously enjoy rebutting the efficiency presumption
advantages over those who are myopic need to be addressed—which brings in
or take their chances and knock on 4 Dixit (1996) counsels the older public finance
wood. The governance of contractual tradition to come to terms with government fail-
relations—the Commons triple of con- ures.
602 Journal of Economic Literature, Vol. XXXVIII (September 2000)
denotes contractual hazards and s de- introduced when transactions are re-
notes safeguards). 5 This can be inter- moved from the market and placed un-
preted as a move from simple to com- der unified ownership (the firm). Be-
plex. We thus begin with autonomous cause added compliance and security
contracting, which is the ideal transac- features always come at a cost, more
tion in both law and economics: “sharp complex modes of governance are re-
in by clear agreement; sharp out by served for those transactions for which
clear performance” (Macneil 1974, p. contractual hazards are especially difficult.
738). This discrete transaction para- The public bureau, in this scheme of
digm comes under strain as contractual things, can be thought of as the organi-
hazards appear. The inability of courts, zation form of last resort: try spot mar-
for example, to verify what is common kets, try incomplete long-term con-
knowledge between the parties to an ex- tracts, try firms, try regulation, and
change (Williamson 1975, p. 30) could reserve recourse to public bureaus for
induce a move from interfirm to in- when all else fails (comparatively). Note
trafirm organization. Other sources of that the common practice of condemn-
contractual hazard include bilateral de- ing public bureaus because they have
pendency (by reason of nonredeploy- lower-powered incentives, more rules
able assets), weak property rights (es- and regulations, and greater job secu-
pecially intellectual property rights), rity than a counterpart firm completely
undisclosed quality, health, and safety misses the point. These features have
hazards, failures of probity, and the like. been deliberately crafted into the pub-
Such hazards compromise contractual lic bureau, thereby to make it better
integrity and give rise to contractual im- suited to govern some (especially diffi-
passes, maladaptations, and investment cult) transactions. 6 Vigilance is none-
distortions. Here, as elsewhere, ineffi- theless needed—lest the public bureau
ciency invites relief. Cost-effective haz- be “overused.”
ard mitigation through added governance If transaction cost economics works
ensues. out of variations on a few key themes,
Moving from less to more complex then this schema, which was initially
governance entails introducing added devised to help explicate the gover-
security features, reducing incentive in- nance of contractual relations in inter-
tensity, and incurring added bureau- mediate product market transactions,
cratic costs. Moving from simple (dis- should also apply, with variation, to
crete) contracts to complex (incomplete other classes of transactions. It does.
long term) contracts is thus attended by The four nodes are interpreted with
a whole series of features: the length of reference to intermediate product
the contract increases, penalties to de- transactions above. Consider final goods
ter breach are introduced, provision is markets and government procurement
made for added information disclosure transactions.
and processing, and specialized dispute Transactions in final goods markets,
settlement mechanisms appear. where individual consumers are the
Additional mechanisms, to include
the use of hierarchy to effect coordi- 6 This is a recurrent theme not only of the trans-
nation and decide disputes by fiat, are action cost economics literature but also of parts
of the agency theory literature. See especially
5 A variant of Figure 2 originally appeared in my Holmstrom (1989) and Holmstrom and Milgrom
paper on “Public and Private Bureaucracies” (1991, 1994), where the benefits of low-powered
(1999). incentives in firms are featured.
604 Journal of Economic Literature, Vol. XXXVIII (September 2000)
buyers, are similar but different. Node are initially expressed as tautologies,
A transactions are generic and competi- which Coase has wryly defined as “a
tively organized. Node B transactions proposition that is clearly right” (1988,
are rare. These correspond to “P. T. p. 19). Because good tautologies expand
Barnum”—there’s a sucker born every the mind and are hard to come by, they
minute—and other fly-by-night transac- deserve respect. Lest, however, we slip
tions. Node C is the credible commit- into the speculations to which Wesley
ment node. Branding in combination Mitchell once referred 7—which is a fate
with reputation effects and product that beset the older style institutional
warranties appear. Also, for some natu- economics as well as the American Le-
ral monopoly transactions, public utility gal Realism movement—we need to ask
regulation serves credibility purposes. what are the mechanisms through
Node D is a nearly empty set. Econo- which a proposed theory operates and
mies of scale and of specialization are what are the refutable implications.
impediments to own-supply by consum- The effort to operationalize promis-
ers, although collective organization ing ideas has both theoretical and em-
(consumer cooperatives) can be used pirical parts. The theoretical often takes
to manage some transactions. (Many the form of a progression from informal
household services can be thought of as to preformal, semi-formal, and fully for-
own-supply, but few fit comfortably mal modes of analysis—ideally acquir-
within the schema.) ing value added in the process. Such an
Government procurement transac- effort helps to sort the sheep from the
tions are also similar but different. goats. Nicholas Georgescu-Roegen had
Node A describes generic transactions a felicitous way of putting it: although
to which, often, tedious technical speci- the “purpose of science is not predic-
fications apply. Very few government tion, but knowledge for its own sake,”
transactions are of a Node B kind. prediction is nevertheless “the touch-
Credibility mechanisms at Node C in- stone of scientific knowledge” (1971, p.
clude the elaborate machinery of “ad- 37). Would-be theories for which pre-
ministered contracting,” as with defense dictive content is lacking must eventu-
procurement (which transactions, how- ally step aside (be set aside) for those
ever, are sometimes compromised by for which the hard work of formalization
the shared interests of the government and empirical testing are undertaken.
agency and the private supplier). And Theory Development. Formalization
Node D is the public bureau, where for is vital to a progressive research agenda,
probity or political reasons the govern- but it sometimes comes at a cost. Thus
ment chooses to manage the transaction although Simon once argued that
itself. “mathematical translation is itself a sub-
Other applications of the schema in- stantive contribution to theory . . . be-
clude the employment relation (James cause it permits clear and rigorous rea-
Baron and Kreps 1999, ch. 4) and cor- soning about phenomena too complex
porate finance (the choice between to be handled in words” (1957, p. 89)
debt and equity). Some transactions, 7 “Speculative systems can be quickly excogi-
such as alliances and joint ventures, tated precisely because they do not require the
pose complications of a disequilibrium economist to collect and analyze masses of data, to
contracting kind (Williamson 1991) that test hypotheses for conformity to fact, to discard
those which do not fit, to invent new ones and test
are beyond the reach of the schema. them until, at long last, he has established a factu-
Operationalization. Many good ideas ally valid theory” (Mitchell 1945, p. 2).
Williamson: The New Institutional Economics 605
and subsequently asserted that the for the property rights theory of the
“poverty of mathematics is an honest firm, which is a major intellectual
poverty that does not parade imaginary achievement that has spawned a grow-
riches before the world” (1957, p. 90), ing literature on the formal modelling
provision also needs to be made for the of incomplete contracts. 8 As Kreps sug-
possibility that core features of the the- gests, however, valuable chunks are
ory are left out or obscured by the missing. In the spirit of full disclosure
translation. There is, after all, such a (honest poverty), I focus on these.
thing as prematurely formal theory. The most consequential difference
Kreps speaks to the issues as follows between the TCE and GHM setups is
(1999, p. 122): that the former holds that maladapta-
tion in the contract execution interval is
If Markets and Hierarchies has been transla-
the principal source of inefficiency,
ted into game theory using notions of infor-
mation economics, it is a very poor transla- whereas GHM vaporize ex post
tion . . . In particular, mathematics-based maladaptation by their assumptions of
theory still lacks the language needed to cap- common knowledge and costless ex post
ture essential ideas of bounded rationality, bargaining. The upshot is that all of the
which are central to . . . transaction costs
inefficiency in GHM is concentrated in
and contractual form. Anyone who relies on
the translations alone misses large and valuable the ex ante investments in human assets
chunks of the original. (which are conditional on the ownership
of physical assets). 9
What is referred to as the “property This shift from ex post maladaptation
rights theory of the firm,” which had its (the hazards from which vary with the
origins with Sanford Grossman and condition of asset specificity and the
Oliver Hart (1986) and has subsequently disturbances to which a transaction is
been developed by Hart and John Moore subject) to ex ante investment distor-
(hence the reference to the GHM model), tions matters. For one thing, GHM
relates to but differs significantly from makes very limited contact with the
the transaction cost economics setup (as data 10 whereas (as discussed below)
presented, for example, in Williamson TCE is an empirical success story. Re-
1985, 1991). It is similar in that it deals lated (ex post) governance and (ex ante)
with the make-or-buy decision through investment differences are the following:
a setup where contracts are incomplete
8 The January 1999 issue of the Review of Eco-
(by reason of bounded rationality),
nomic Studies is entirely devoted to recent contri-
mere promise cannot be used to over- butions, critiques, responses, and extensions upon
come noncontractibility (by reason of the GHM model.
9 Bounded rationality enters this setup in a very
opportunism), and parties to the con-
peculiar way: parties who are unable to write com-
tract are bilaterally dependent (by rea- plete contracts ex ante are nevertheless able to an-
son of asset specificity). These com- ticipate ex ante what decisions will be taken ex
monalities notwithstanding, there are post, contingent on state realizations (Kreps 1999,
pp. 123–25). In effect, the GHM setup is one of
also major differences. selective unbounded rationality: “not everything
Some of these differences are attrib- that is logically consistent is credulous” (Kreps
utable to simplifications that invariably 1999, p. 125).
10 That the data relevant to GHM are so limited
attend formal modelling. Ideally, core and inaccessible explains why there has been “no
features of the verbal argument and the formal testing of the property rights approach”
mechanisms through which they work (Hart 1995, p. 49)—although the “inside contract-
ing system” (John Buttrick 1952) is an approxima-
are made more precise in the process of tion to (and its failures could be interpreted as a
formalization. Such a case can be made partial contradiction of) GHM.
606 Journal of Economic Literature, Vol. XXXVIII (September 2000)
(1) The TCE rendition of the make- ing). GHM assume that incentive
or-buy decision between succes- intensity, administrative controls,
sive stages (A and B) asks whether and informal organization are un-
A and B should be separately changed by ownership and that
owned and operated or if the own- courts are irrelevant (because of
ership and operation of these two costless renegotiation). None of
stages should be unified. If in- the physical asset utilization and
dependent, then each stage appro- transfer pricing distortions that I
priates its net receipts (high-pow- associate with the “impossibility of
ered incentives obtain) but selective intervention” (Williamson
maladaptation problems can arise 1985, pp. 135–40) thus occur
during contract execution. If uni- under the GHM setup.
fied, then the two stages are man- (3) TCE examines a wide range of ex
aged coordinately through hierar- post devices for infusing credible
chy. (Maladaptation problems are commitments into contracts and
thereby relieved; incentives are applies this reasoning to a wide set
lower-powered; and added bureau- of transactions. Variations on this
cratic costs arise.) By contrast, theme include hybrid modes of or-
GHM view vertical integration in a ganization (Scott Masten 1996,
directional way: either A buys B or Part III), exchange agreements
B buys A, and it matters which way and other uses of hostages to sup-
this is done. That is because com- port exchange, the organization of
mon ownership under GHM does work, the organization of labor and
not imply unified management. In- human resources more generally,
stead, each stage (in all configura- corporate governance, regulation
tions—A and B are independent; A (and deregulation), public bu-
buys B; B buys A) appropriates its reaus, and project financing. Be-
net receipts. This last is a very un- cause GHM is a property rights
usual condition, in that unified and property rights only construc-
ownership is normally thought of tion (Holmstrom 1999), it relates
as a means by which to effect to some of these issues not at all
cooperation.11 and others very selectively (Hart
(2) TCE maintains that each generic 1995; Hart, Andrei Shleifer, and
mode of governance—spot market, Robert Vishny 1997).
incomplete long-term contract,
firm, bureau, etc.—is defined by a GHM is nonetheless a pathbreaking
syndrome of attributes to which contribution and has set the formal
distinctive strengths and weak- modelling of incomplete contracting in
nesses accrue. Specifically, TCE motion. New formal models of incom-
holds that alternative modes differ plete contracts which are closer in spirit
in incentive intensity, administra- to TCE include the treatment of pro-
tive controls (to include auditing, curement by Patrick Bajari and Steven
accounting, and transfer pricing), Tadelis (1999), which focuses on the in-
access to the courts, and informal centive and ex post adaptation differ-
organization (to include politick- ences between fixed price and cost plus
contracting. Also, the recent paper by
11 The inside contracting system referred to in Susheng Wang and Tian Zhu (2000)
note 7 supra is in the spirit of GHM organization. employs the idea that alternative modes
Williamson: The New Institutional Economics 607
Thus even if we are confident that “poli- This political prescription for massive
ties significantly shape economic perfor- and rapid privatization was reinforced
mance because they define and enforce by the economic theory of the firm on
the economic rules,” whereupon “an es- which the Boycko et al. team relied.
sential part of development policy is the Specifically, they appealed to the afore-
creation of polities that will create and mentioned work by Grossman and Hart
enforce efficient property rights,” there is (1986), which views ownership as a sys-
the further problem that “we know very tem of control rights and treats the ap-
little about how to create such polities” propriate assignment of property rights
(North 1994, p. 366). as determinative (Boycko, Shleifer, and
Real-time events, however, cannot be Vishny 1995, p. 13). Upon privatizing
put on hold. Hard choices have to be state-owned enterprises, therefore, ef-
made. Economic reform in Russia is an fective restructuring by the new stake-
example. holders would presumably follow (op
The team of Maxim Boycko, Andrei cit, p. 150). In the confidence that the
Shleifer, and Robert Vishny responded future would take care of itself, the
to the perceived need to give shape to mass privatization program that was be-
the reform with the recommendation gun in the spring of 1992 had purport-
that the Russian economy should be edly reached a “triumphant completion”
privatized quickly and massively. Con- in June 1994 (op cit, p. 8), by which
siderations of both Realpolitik and eco- date two-thirds of Russian industry was
nomic theory were invoked in support privately owned.
of this recommendation. Had the Boycko et al. team consulted
There being widespread agreement that the new institutional economics, a more
“political influence over economic life cautious and selective program of priva-
was the fundamental cause of economic tization with greater attention to imple-
inefficiency” [in Russia], Boycko, Shleifer, mentation would have resulted. Consider
and Vishny (1995, p. 11) declared that: first the literature on franchise bidding
for natural monopoly, where the prop-
. . . the principal objective of reform was . . .
to depoliticize economic life. . . . Privatiza-
erty rights approach and the gover-
tion fosters depoliticization because it de- nance approach reach very different
prives politicians of the opportunity to allo- conclusions.
cate goods. . . . The goal of privatization was The property rights approach to the
to sever the links between enterprise manag- problem of natural monopoly is to con-
ers and politicians. . . . There was no other
way to achieve restructuring and efficient
duct an ex ante bidding competition
operation of firms. and award the right to serve the market
to the group that tenders the best bid
The two strategic actors in this re- (Demsetz 1968; George Stigler 1968;
form program were the official bureauc- Richard Posner 1972). Very much in the
racy, which was viewed as “the enemy spirit of Boycko et al., the future will
to be fought at all costs” and the stake- take care of itself once the assets have
holders—managers, employees, and lo- been privatized in this way.
cal governments. The Boycko et al. That sanguine view does not with-
team “consistently and generously rec- stand scrutiny if serious ex post imple-
ognized stakeholders’ claims, and thus mentation problems are in prospect.
ensured their eventual support of priva- Under the governance approach, the
tization” (Boycko, Shleifer, and Vishny award of a monopoly franchise needs to
1995, pp. 13–14). be assessed comparatively. This entails
610 Journal of Economic Literature, Vol. XXXVIII (September 2000)
looking ahead and uncovering ex post verdict. Thus, although privatization was
contractual hazards, thereafter working evidently a success for small firms, it
out the ramifications for alternative was deeply problematic and attended by
modes of governance (Williamson 1976, massive corruption in others. But for
pp. 79–91). Because franchise bidding undue reliance on ex ante property rights
works much better for some natural mo- reasoning, some of these problems
nopoly industries than others (William- could have been anticipated by looking
son 1976, pp. 102–103), the use of fran- ahead and examining the hazards of ex
chise bidding will be reserved for those post implementation. Greater apprecia-
industries where comparative net bene- tion for the shortfalls of the institutional
fits can be projected—but not other- environment in Russia would have led to
wise. Privatization, it turns out, is not an more cautious pronouncements (Anders
all-purpose solution (Victor Goldberg Aslund 1995). Whether added respect
1976; George Priest 1993). for the rules of the game (to include an
Although privatizing an entire econ- appreciation for the limited efficacy of
omy is a much more ambitious undertak- Russian law enforcement) would have
ing than privatizing a natural monopoly resulted in rule improvements in Russia
industry, the key lessons nevertheless could be disputed. Arguably, however,
carry over. Specifically, privatizing needs the effort to reform Russia would have
to go beyond the ex ante award stage to proceeded in a more modest, slow,
include an examination of possible ex molecular, deliberative way.
post implementation problems and, in None of this is to suggest that the
consideration of the differential hazards, NIE could have done it all. The Boycko
to proceed selectively. et al. team made heroic efforts. My
Recall, moreover, that the NIE oper- claim is much more modest: the NIE is
ates at two levels. Upon moving from informative and should be included as
the level of governance to that of the part of the reform calculus.
institutional environment, the rules of
the game come under review. The Levy 5. Concluding Remarks
and Spiller (1994, 1996) study of priva-
tizing telecommunications in five coun- The new institutional economics is a
tries reveals that the decision to priva- boiling cauldron of ideas. Not only are
tize and the nature of privatization turn there many institutional research pro-
critically on the condition and quality of grams in progress, but there are com-
judicial independence, the division of peting ideas within most of them. With
powers between the executive and leg- reference to history, for example, we see
islative branches, the competence of North (1990) and Avner Greif (1999)
the regulatory bureaucracy, and con- pursuing complementary but separate
tractual safeguards. Whether and how agendas. The institutions of embedded-
to privatize telecommunications should ness (Level 1) are an important but un-
therefore be made conditional on these derdeveloped part of the story. Within
features. transaction cost economics we distin-
As Bernard Black, Reinier Kraakman, guish between governance and measure-
and Anna Tarassova detail in their pa- ment branches. The attributes of mixed
per on “Russian Privatization and Cor- ownership modes (alliances, joint ven-
porate Governance: What Went Wrong” tures, franchising, and the like) as well
(1999), the “triumphant completion” of as the mechanisms for supporting cred-
privatization in Russia was a premature ible contracting between autonomous
Williamson: The New Institutional Economics 611