GM vs. TCE: Another “Block Upon Block”?

17 April 2009 at 2:13 pm Michael Sykuta 14 comments

| Mike Sykuta |

Ronald Coase has spent the past two decades or more lamenting the lack of progress in economic theory. He bemoans the fact that economics, unlike its physical-science counterparts, fails to dispose of (or pursue new versions of) theories when facts show that prevailing theories are inaccurate or incomplete.

Among his many arguments, Coase has pointed to Williamson’s Transaction Cost Economics (TCE) as one that seems impervious to the facts. Part of Coase’s discontent with the TCE story rests on his observation that many firms sustain relationships characterized by high degrees of asset specificity using contractual means. While Ben Klein and others pointed to General Motors-Fisher Body as evidence to support the TCE story, Coase pointed to relations with auto frame manufacturer A.O. Smith at the same time that were not subsumed by vertical integration. This eventually led to the infamous GM-Fisher Body debate that seems for want of a real conclusion (see some of Peter’s previous comments on this here, here and here).

Well once again, General Motors seemingly plays the foil against TCE. Several weeks ago, GM announced plans to purchase Delphi Group’s global steering manufacturing operations. Delphi operated the steering unit solely for GM’s use. Delphi, in bankruptcy since October 2005, has been able to use GM’s dependence on Delphi’s operations to secure roughly $450 million in liquidity capital from GM to maintain its operations. Sounds like the classic hold-up problem, doesn’t it? But wait!

Although Williamson has long argued that other factors (such as complexity, frequency, and uncertainty) affect the form of organizational governance, the lynch pin of TCE is asset specificity. Asset specificity has come to be broadly conceived to include any circumstance in which the secondary-use value of the asset leaves a quasi-rent on the table, regardless of the source of that quasi-rent. It may be due to technical specifications, geographic location, firm-specific skill sets, dedicated investments, temporal constraints or even something like relative thinness in the local market. Regardless the source, the defining characteristic of TCE research is the role asset specificity plays in determining the optimal organizational form.

GM’s repurchase of the steering manufacturing unit (GM formerly owned Delphi and most of its current facilities) was not the result of a change in the specific nature of these assets relative to GM. As these assets have always been exclusively dedicated to GM’s use — and were GM’s sole source for these components — not only did the degree of asset specificity not change, but it has always been very high. The fact that GM did not already own (or maintain ownership of) the assets should itself raise questions about the role of asset specificity in determining the optimal organizational form. But it is obvious that it wasn’t a change in specificity that drove GM’s recent decision to acquire these assets.

One might argue that the recent state is simply a disequilibrium and that GM is adapting its governance to a more economically sound arrangement. Sadly, reports don’t bear that out either. Even as GM announced it would purchase the assets from Delphi, the company intimated that it intends to turn around and sell the assets right back out to a new vendor — presumably one in better financial health. Thus, the specificity of the assets to GM seems to have little to no bearing on the optimal organizational form.

Supporters of TCE are likely to argue — and rightly so — that there is a wealth of empirical evidence that is consistent with Williamson’s hypothesis concerning asset specificity. Indeed, William himself declared, “transaction cost economics is an empirical success story.” However, consistency and causality are very different things. Empirical research on TCE has forever been plagued by poorly defined concepts and even more poorly measured proxies for asset specificity (see Section 4 of this paper for more discussion). While proxies for asset specificity do often show a correlation with the prescribed organizational forms, it is clear that there are other factors at work that make asset specificity more or less important—and these augmenting (amplifying or dampening) factors have been overlooked far too long, whether as independent factors or as interactive effects with asset specificity.

In the present GM case, it seems clear that the degree of asset specificity itself is not sufficient to explain the change in organizational form. Rather, Delphi’s financial situation — and even that only after more than three years in bankruptcy — finally created enough misalignment of incentives for GM to change its sourcing arrangements. Perhaps Williamson’s assumption of “guile” and opportunistic behavior is the weak link in the theory — wthout such opportunistic behavior, the specificity itself is a non-problem. Too bad that the existing literature generally fails to address that dimension. Like asset specificity itself, too much is assumed and too little is tested.

Entry filed under: Corporate Governance, Former Guest Bloggers, New Institutional Economics, Theory of the Firm.

Information Encountering Economic Institutions of Strategy

14 Comments Add your own

1. Michael F. Martin | 17 April 2009 at 3:27 pm

Can somebody explain to me how the concept of “asset specificity” is different from what business-people commonly call a “competitive advatage”?
2. Michael F. Martin | 17 April 2009 at 3:31 pm

If an asset has a higher value in a particular transactional context — i.e., when particular parties, property, or timing are involved — then doesn’t that just give price power to the buyer or seller? That’s just a barrier to entry for competitors (i.e., competitive advantage), right?
3. sykutam | 17 April 2009 at 3:36 pm

Michael,

Competitive advantage may refer to a firm’s ability to exploit a particular asset (asset based, resource base, etc.), perhaps more efficiently or effectively than its competitors.

Asset specificity is a more generic concept. Essentially, it’s the idea that firms may invest in an asset associated with a particular transaction relationship such that the asset/investment’s value is specific to or dependent on that particular transaction relationship (e.g., it’s next best use is of (much?) lower value).

A classic example might be a coal mine and an electric plant being co-located. Once the investment is made, the benefits (or value) of the investment–in this case, the reduced cost of transporting coal long distanced–are dependent on the relation between the parties. One party or the other may have incentive to opportunistically appropriate that value by “holding up” the other party in renegotiation.

Asset specificity may give rise to a competitive advantage (e.g., this electric plant may be able to produce at lower cost than its competitors), but the concepts are not synonymous.
4. Michael F. Martin | 17 April 2009 at 3:56 pm

This is very helpful. But let me press you for a few more details.

I think your definition of competitive advantage is pretty close to what I have in mind — namely, the ability to produce at a lower cost, sell at a higher price, or both.

What I’m wondering specifically is whether asset specificity is defined as either the same set or a subset of these “competitive advantages.”

What I understand you to be saying is that asset specificity and competitive advantage do not overlap because:

(1) asset specificity can be defined with respect to a single transaction;
(2) asset specificity may result in the opposite of a competitive advantage (i.e., a situation in which the buyer or seller is forced to produce at higher cost, sell at lower price, or both)

In the example you give there is asset specificity, but not necessarily competitive advantage because one or the other of the parties (either the mine or the power plant) can choose to appropriate all of the gains from trade with its counter-party, in effect turning that counter-party’s competitive advantage into the opposite of a competitive advantage.

So asset specificity is a necessary, but not a sufficient condition to competitive advantage?

And what’s missing is the variable of opportunism? I.e., will the owner of the asset-specific firm be able to appropriate the rents or will it lose out to its counter-party for some reason?
5. Peter Klein | 17 April 2009 at 9:16 pm

Michael, I’d start by distinguishing between cause (strategy, environment, governance structure, asset characteristic) and effect (performance outcome). Asset specificity is the former, competitive advantage is the latter. So the question is whether asset specificity is necessary or sufficient for competitive advantage.

I’d say neither. Not necessary, because there are many sources of competitive advantage unrelated to asset specificity. A patent, for example, may be source of competitive advantage (whether conceived as a unique resource or as an entry barrier), but need not be relationship-specific (i.e., the patent would be equally valuable in any firm’s hands, or equally valuable if the patent-holder transacted with suppliers or customers other than its current ones). Likewise, a firm may have relationship-specific investments but not perform better than a rival that lacks those investments.

The thing to keep in mind is that asset specificity is invoked not to explain performance, but to explain organizational form. Asset specificity is used in TCE to explain whether a firm will transact on the spot market, use long term contracts, vertically integrate, or employ some hybrid form of governance. (Whether the explanation works is, of course, the issue Mike raises in his post.) As for performance, TCE predicts that firms that correctly align their asset characteristics with their governance structure — i.e., those that vertically integrate under conditions of high asset specificity, but not otherwise — will perform better than firms that are “misaligned.” But that’s not the same thing as saying that firms with relationship-specific investments perform better than firms with general-purpose assets.

I have a survey paper that discusses these issues in some detail:

Click to access KI140-17-433-464–KLEIN-x.pdf
6. Michael F. Martin | 17 April 2009 at 10:25 pm

The cause/effect distinction works better for me. But there are still problems.

The patent example unfortunately raises more questions than it answers. Patents are transferable, of course. But many people (investors, managers, and lawyers) underestimate the importance of keeping the legal rights and the know-how together. In fact, outside of a few fields (drugs, surgical devices) naked patent rights are worth less than the transactions costs of defending lawsuits (the fact that these are not small is giving the secondary market bad character). In these fields, patents are probably best valued as an option on an inventor’s future work — good inventors can leave for competitors; but blocking patents can keep prior employers at the table.

But I get the point that if we focus narrowly on the naked patent rights, then asset-specificity is neither a necessary nor a sufficient condition to competitive advantage. In the case of drug patents there is competitive advantage but not asset-specificity. In the case of computer patents you might have asset-specificity, but no competitive advantage.

The problem is that if the definition of asset-specificity is expanded to include human-capital-specific investments, then we’re back where we started because, at bottom, both the drug company and the software company got their patents from the work of a particular inventor or group of inventors — the asset-specificity and competitive advantage again coalesce.

Another way of saying this, I think, is that competitive advantage is always ultimately derived from a specific organization of human capital. That’s something I’ve come around to believing myself, and it seemed at first like that might be where asset-specificity theory was going. But I can see now that the concern is more with organizational form in the first place. I guess I take a very consequentialist view on organizational form, and can’t really conceive of why (in theory at least) anybody would want a firm with structure that wasn’t conducive to competitive advantage.
7. Peter Klein | 18 April 2009 at 1:11 am

Michael, I think you’re confusing asset specificity with asset uniqueness in some kind of VRIO sense. A specific asset (as the term is used in TCE) is an asset that is more valuable to firm X when firm X trades with firm Y than when firm X trades with firm Z. TCE’s focus in this case would be on the contractual relationship between firms X and Y, not the profitability or value of firm X.

Asset specificity refers to assets that are *relationship specific*. An asset — physical or human — can be a source of competitive advantage in the VRIO sense without being relationship specific and vice versa. E.g., suppose Steve Jobs is a unique asset that is the source of Apple’s competitive advantage. TCE asks whether Steve’s human capital is worth more when he works with Apple than it would be when he works for another firm. It deals with the relationship between Steve and Apple, and offers specific predictions about the nature of Steve’s employment contract. An employee can have skills that are specific to one employer, suggesting a long-term employment relationship, without those skills being a source of competitive advantage for the employer. Likewise, an employee could have skills that are rare and valuable, but no more valuable to the current employer than these skills would be to another employer. In that case the employee’s human capital would be a source of competitive advantage, but the employee would not necessarily have a long-term labor contract with the employer.

I do think there are interesting issues about the relationship between asset specificity and competitive advantage that have not been teased out in the literature. But they are conceptually distinct.
8. paolo mariti | 18 April 2009 at 6:00 am

Asset specificity and competitive advantage are “conceptually” distinct. No doubt. Yet when Michael F . Martin writes: “I can see now that the concern is more with organizational form in the first place. [ I ]….. can’t really conceive of why (in theory at least) anybody would want a firm with structure that wasn’t conducive to competitive advantage”, he raises a key point for research. Why do firm get organized? Why do they change orgnizational forms and boundaries over time? Firms do not get organized both internally and externally and over time for the sake of it, rather to vie on their own changing markets (and cope with changing technologies). Within and between firms organization of the firm should not be studied separately from phenomena occurring on the markets where the firm’s goods and services are offered.Organization is precisely the firm’s response to promptings and signals from “the world out there” .
9. Peter Klein | 18 April 2009 at 8:33 am

Paolo, I certainly agree! But the same is true of all the firm’s strategic choices, right? There’s nothing unique about organizational structure in this regard. The question is how large are the marginal effects of organizational form on performance, compared to the effects of product selection and characteristics, pricing, entry and exit, strategic interaction, legal and political strategy, and the thousands of other choices facing the firm.

BTW TCE doesn’t look to organizational form per se, but to the “discriminating match” (Willismaon’s term) between transaction characteristics and organizational form, as a driver of performance. The newer round of empirical TCE studies, such as Nickerson, Silverman, and Freeman (1997) on trucking or various Argyres and Bigelow papers on automobiles, study the effect of “efficient alignment” on profitability or survival. (Sykuta seems to have in mind the older literature, which looks simply at alignment — do transactions characterized by high degrees of asset specificity tend to be vertically integrated — rather than the effects of alignment on performance. But then again, I’ve been trying to straighten him out for years. :-) )
10. Michael F. Martin | 18 April 2009 at 9:32 am

This is slowly coming into focus for me.

Here is how I’m picturing this: Start with a graph of a network with each node representing a firm, and edges representing relationships between the firms. There is also a bundle of assets associated with each node.

What you’re telling me is that an asset in the bundle is “asset-specific” if and only if that asset is “worth more” when it is associated with a particular node than when it is associated with any other node. The specific asset is also “relationship-specific” for the same reason — the relevant relationship being the relationship between the particular node and the specific asset.

But with that picture in mind, it is now obvious that “asset-specificity” does not necessarily result in competitive advantage. It may be the case, for example, that a particular piece of software would be “worth more” if it could run on Google’s servers, and the asset therefore be specific to Google. Yet Google would not necessarily further its competitive advantage by running the software. Indeed, it might destroy its competitive advantage by doing so.

That much seems clear. And this is interesting to me at least from a mathematical perspective because, assuming we have a map of the relationships between firms, then we can say there is a geography to asset specificity with peaks and valleys corresponding to the nodes where each asset is “worth more” or less than at other nodes. I keep putting quotations around “worth more” because I’m not sure how we’d set units of altitude in this picture — worth more than what? But at least we can agree that there is a relative measure of worth that applies.

Now let me add two additional complications: edges between the nodes (representing transactions between the firms) and time (the network evolves as edges appear and disappear with each transaction).

What I am calling “competitive advantage” would be manifest in this model as a node that (given a bundle of assets) maximizes the volume of sale transactions and minimizes the volume of buy transactions within a particular window of time — i.e., a node that extremizes the time-integrated sum of profits for a particular bundle of assets.

Here are a few observations that I think are unobjectionable:

Assuming transactions costs are small relative to how much more an asset is worth relative to a particular node (a big assumption), an asset may be observed to “flow” uphill over time to higher and higher nodes through transactions. (Isn’t this the Coase Theorem?)

If we freeze things at a particular moment in time, then we will observe that there are some nodes that a particular asset is “worth more” at in the sense that profits would not be improved by moving that asset to any other node in the network at the same moment in time. (Isn’t this what is meant by “asset-specifity”?)

Here is my conjecture:

Over periods of time that are very long relative to the period of an average transaction, assets will end up at the nodes with a competitive advantage.

The reason for this is not so difficult to picture using this model. The nodes with a competitive advantage are the same nodes that are a link in the chain of transactions that minimizes the time-integrated sum of finite transactions costs as they are incurred by the network as a whole.

“Transactions costs” here means something more than legal or brokers fees. It would include, for example, the additional time required to complete transactions where nodes are not ready to transact at a particular moment but would have been a moment sooner or a moment later.

Two nodes will “integrate” into a single node and share profits and losses whenever that will minimize the transactions costs in this sense. Anyway this is basically what I had in mind when I started blogging about “synchronized flow.”
11. Michael Sykuta | 18 April 2009 at 10:42 am

Actually, Peter, I specifically added the paragraph about the possibility of this being a realignment as an acknowledgment that the literature has moved in that direction. However, the underlying argument in that literature tends to be based primarily on asset specificity in the “old literature” sense; it simply puts the empirical work in a more dynamic context.
12. Peter Klein | 18 April 2009 at 3:17 pm

What, I was supposed to read the post?
13. TRUTH ON THE MARKET » Weekend Reading | 17 September 2009 at 1:58 pm

[…] Sykuta on GM and Transaction Cost Economics […]
14. Danny L. McDaniel | 5 December 2009 at 4:24 pm

The Fisher Body model is so arcane that it is relegated to ancient American history. General Motors is a shell of its former self with no way to be anywhere close to what it once was, and Fisher Body was killed off in name in 1984. American manufacturing doesn’t operate that way anymore and never will.

Danny L. McDaniel
Lafayette, Indiana