X-Efficiency and Allocative Efficiency: What Have We Learned?

by Roger Frantz
X-Efficiency and Allocative Efficiency: What Have We Learned?
Roger Frantz
The American Economic Review
Start Page: 
End Page: 
Select license: 
Select License

In 1966 Harvey Leibenstein postulated the existence of a nonallocative (inlefficiency. Supporters and critics appeared shortly thereafter. Some of his critics, using tortured logic and the concept of leisure in some unusual ways, asserted that such a thing cannot exist. Other critics, writing within a rent-seeking framework, said that it could exist but was inconsequential. In the first part of this paper I discuss the general issue of X-efficiency theory and the criti- cisms against it. In the last part I want to extend the tortured logic of some of the critics and assess some implications of their criticisms for traditional microeconomic models. The X-efficiency concept and the evidence consistent with the theory pointed toward a potentially large source of welfare gains. What have the criticisms pointed to- ward?

I. What Was All the Fuss About?

Arnold Harberger's (1954) estimates of the welfare impact of market power in the United States were very small, on the order of 0.001 of GNP. This led Robert Mundell (1962) to fear that such small estimates would lead the public to believe that economics itself is not important. Leibenstein came to the rescue when he postulated a nonallocative effect of market power; mar- ket power will not only drive a wedge be- tween price and marginal cost, but will also raise the firm's costs above the technologi- cally minimum levels. The precise nature of this inefficiency was unclear to Leibenstein; hence, he named it X. Furthermore, the data- that Leibenstein analyzed suggested that X-inefficiency was often much larger

*Professor of Economics, San Diego State Univer- sity, San Diego, CA 92182.

than allocative inefficiency. The existence of X-inefficiencies meant that improvements in resource allocation greater than had previ- ously been believed were now possible. It also implies the possible existence of a free-lunch. Leibenstein's research generated a considerable amount of initial excite- ment among economists.

Leibenstein's work also generated a con- siderable amount of theoretical and empiri- cal research. All of the empirical research was done by others; almost all of it reported results consistent with the theory. However, in postulating a nonallocative inefficiency Leibenstein also created a controversy. Af- ter all, until that time economists thought mainly about allocative (inlefficiency and as- sumed that firms were always minimizing their costs. In raising doubts about cost minimization Leibenstein spoke increasingly about the possibility of suboptimal decisions and less than fully rational behavior, and what this would mean for economic models. Critics of X-efficiency theory appeared ev- erywhere.

11. Evaluating X-Efficiency Theory With Tests and Assumptions

The best way to test for X-inefficiency is to conduct field studies as close as possible to the micro-micro level. Data-envelope analysis (DEA) has been used for this pur- pose to look at the relative X-efficiency levels of firms (Sumit Majumdar, 19911, sub- groups within firms (Wade Cook et al. 19901, and individuals within firms (Leibenstein and Shlomo Maital, 1992). DEA studies re- port inefficiencies which "look like" X-inef- ficiency.

The studies in support of the theory have often been at the industry level. These stud- ies have reported differences among firms or industries related to either their cost or production functions and in a manner con- sistent with X-efficiency theory. These dif- ferences have been attributed to factors such as market structure and ownership form.

Less rigorous evidence is found in reports of corporate "turnarounds." In many of these stories, employees report both more work effort and more job satisfaction. The preturnaround condition was thus X-inef- ficient.

If these reported X-efficiency gains in productivity can be generalized to the entire economy, then a welfare loss of 10 percent for the economy as a whole does not seem far-fetched. These are large potential gains. Nonetheless, most of the critics of the the- ory and the evidence have attempted to trivialize the matter and have assumed it away.

The critics have not employed empirical tests as their main tool. As I have tried to show (Frantz, 1988), the critics have relied upon assumptions and deductive logic. One of the criticisms is that the assumptions that Leibenstein listed varied among his articles. A second type of criticism was that his X- efficiency theory is almost totally compatible with "standard neoclassical theory" and hence intellectually redundant. Initially, col- leagues complained that X-efficiency theory was not consistent with standard theory, and for that reason they were suspicious about it. Leibenstein (1969) set out to quell their fears by showing how X-efficiency the- ory was consistent with standard theory. His critics then used this as proof that X-efficiency theory is consistent with standard theory and, hence, intellectually redundant.

Leibenstein long ago acknowledged regretting that he ever wrote that article, but not because it is a "smoking gun." He sim- ply ceased trying to make his ideas compati- ble with standard theory; and he came to see clearly that standard theory could make any idea seem compatible with it. This "aha" experience became the basis for his compar- ison between standard theory and "bulls-eye paintingn-the use, ex post, of terms such as leisure, uncertainty, or transactions costs to "explain" all observations (Leibenstein, 1982). So long as a story could be told using any of these words as to why X did or could not exist, the critic felt as if a convincing argument had been put forth.

A third type of criticism is based on the assumption that individuals are (constrained) fully rational decision makers. As such, individuals always minimize "avoidable" costs of production; "unavoidable" costs are, unavoidable. For workers with a "taste for laziness" most costs of production are "unavoidable." Lazy workers are thus not X-inefficient; they are constrained util- ity-maximizers. Another variant is that higher costs are due to rent-seeking. The argument was that, in any industry where rent-seeking is known or suspected, higher costs must be due to (rational) rent-seeking. In this case as well, higher costs are "un- avoidable."

Implicit in this criticism is that the behav- iors of rational utility-maximizing individu- als can never be called inefficient. If they work for a monopolist, allocative inefficiency (a failure of the market) exists; but an individual can never be inefficient. Hence, the firm cannot be said to be inef- ficient. Hence, X-inefficiency cannot exist.

A fourth and related type of criticism is that Leibenstein is judging human beings against a competitive ideal (nirvana), but this is tantamount to imposing a theoretical standard against which human behavior is being judged. The critics' view is that real human behavior and real human constraints, rather than nirvana, should be the criteria against which efficiency should be judged. Of course, that means that real hu- man behaviors are, by definition, efficient. Once again, X-inefficiency cannot exist.

111. Leisure

Probably the most ingenious criticism has been the on-the-job "leisure" argument. Critics claimed that because employees are rational any reduction in effort that raises costs is due to the employees' preference for leisure over income. Higher costs are thus not a symptom of inefficiency, but the effect of fully rational workers' preference for leisure.

The terms leisure and work effort took on cosmological proportions. "Work effort" was likened to matter, neither capable of being created nor destroyed. So, if workers were not expending energy for producing the firm's commodities, then they must be expending energy on something else; and that something is their leisure output. Hence, lazy people are as efficient as hard- working people, given their lazy effort level. George Stigler's (1976) criticism was much more sophisticated than this, but it attacked the existence of X-inefficiency along these same lines.

Is leisure a legitimate concept for under- standing firm behavior? Of course. For many years Leibenstein did not dwell much on this issue of leisure. He assumed that work breaks are a normal part of the workplace and that they can serve to enhance produc- tivity. His interest was physical and mental laziness ("leisure"), which is auozdable and which leads to a nonallocative form of in- efficiency.

Are all human errors and mistakes a source of X-inefficiency? No. Individuals certainly make, ex ante, unauoidable errors. If one makes a decision using all informa- tion available at the time then any errors are unavoidable and known only ex post. Ex ante, the individuals are fully X-efficient. Ex post, the employee may have decisional regret and wish that he chose the money rather than the leisure, or one set of proce- dures rather than another. However, no X- inefficiency should be inferred; Leibenstein never did.

However, Leibenstein assumed that indi- viduals are at least capable of making, ex ante and ex post, avoidable errors: subop- timal decisions with respect to their income-leisure trade-off and with respect to their day-to-day work behaviors. The reasons for making avoidable errors include laziness, and inappropriate conventions and habitual ways of doing things.

On the other hand, these critics have tried to show that all higher than minimum cost levels are due to rationally calculated leisure which is part of a constrained utility-maximizing strategy and is part and parcel of the firm's output. By definition, the firm would always be on the production frontier, minimizing avoidable costs. By definition, only allocative inefficiency can exist.

IV. Leisure, Subsidies, and X-Inefficiency

Assume that the critics are correct and that leisure is a benefit of monopoly power; leisure is neither a cost of monopoly power nor a source of inefficiency. Furthermore, assume that the monopolist's average cost equals the price, so that all profits have become leisure and X-inefficiency is zero. What are the implications for the effects of a subsidy under these conditions? If one excludes leisure from the analysis, the wel- fare impact of monopoly power is the al- locative inefficiency measured by the Har- berger "triangle." In this case, subsidies are shown to reduce welfare because the costs of the subsidy exceed the benefits: the in- crease in consumer surplus.

Policymakers could eliminate this alloca- tive inefficiency by providing a per-unit out- put subsidy. This includes subsidizing leisure. The subsidy would lower the firm's price, as it also lowers its marginal cost, and would increase the monopolist's output to that of the competitive industry, the socially optimum rate.

Leisure being a benefit, a subsidy that increases output is subsidizing more employees who will also enjoy (more) leisure. If workers enjoy leisure when the monopolist produces its nonsubsidized level of output, then surely they enjoy more leisure when output is increased to the socially optimum level. Therefore, the benefits of a subsidy exceed the increased consumer surplus, be- cause they include added leisure. The bene- fits of the subsidy may thus exceed the cost of the subsidy. A perverse implication fol- lows: that allocative efficiency and welfare can thus be increased by subsidizing X-inef- ficiency or leisure.

V. The Public-Goods Nature of
Rent-Seeking and X-Efficiency

Rent-seeking arguments have tended to allow for the existence of X-inefficiency but to argue that it entails no welfare loss. This is because X-inefficiency and rent-seeking are substitutes for each other; every dollar (rationally) "allocated" to X-inefficiency is not available for rent-seeking. This conclu- sion changes when leisure is included in the analysis.

The critics have assumed that the monopolist will divide its profits between rent-seeking and X-inefficiency. Its rent-seeking expenditures will secure the firm a government subsidy. As in the previous ex- ample, this includes subsidizing leisure X- inefficiency. The benefits of this subsidy in- clude added profits (perhaps captured as added rent-seeking expenditures), added leisure/X-inefficiency, and added consumer surplus. As in the previous example, the benefits of this subsidy may exceed the costs. In addition, rent-seeking and X-(in)efficiency/leisure are not substitutes for each other but have the characteristics of public goods. That is, either is a benefit accruing to someone without there being an explicit cost to anyone else. The possibility thus exists that X-inefficiency does have welfare impli- cations. Again, there is a perverse implica- tion: subsidizing X-inefficiency improves welfare.

VI. From the Ridiculous to the Sublime

With tongue in cheek, I will now extend some of the tortured logic of the critics of X-efficiency theory.

(A) The socially optimum rate of output must take both commodities and leisure into account. Competitive industries produce more commodity output but less leisure out- put. Monopolies on the other hand, pro- duce less commodity output but more leisure output. Since leisure is a benefit of monopoly, individuals are as likely to prefer the monopoly bundle of outputs as they are to prefer the competition bundle. In other words, the bundle of leisure and commodity outputs produced by the monopolist is not inefficient in any meaningful way. To claim otherwise is to apply a type of nirvana crite- rion to the definition of efficiency, in that one's own discounting of leisure is applied to the population at large. The implication is that if X-(inlefficiency cannot exist be- cause it is simply leisure, then allocative (in)efficiency cannot exist for the same rea- son. Mundell may have been correct after all!

VII. Conclusions

Are there welfare impacts of monopoly power not covered in the Harberger trian- gle? Leibenstein offered X-efficiency theory as a nonallocative effect. Empirical evidence has supported this theory. Is more work needed? Yes. Micro-micro studies that in- clude organizational and employee goals would indicate whether any existing X-inef- ficiency is the result of "noneconomic" indi- vidual decision-making procedures or insti- tutional factors. At the same time, such studies would also help properly measure the extent of X-inefficiency. The contribu- tion of X-efficiency research continues to be a better understanding of the effects of mar- ket and organizational forces on the use of resources and, hence, on making improve- ments in the use of resources. These poten- tial improvements may be large.

On the other hand, what have the critics of the theory contributed? First, an odd use of the term "leisure," and one which leads to strange and perverse implications. Sec- ond, they have given us mostly assumptions about why X-inefficiency cannot exist. In fact, X-inefficiency cannot exist only within their own models and given their own methodologies and use of language. The economics profession has the capacity to entertain alternative models and method- ologies.


Cook, Wade, Roll, Yaakov and Kazakov, Alex, "On the Measurement of the Relative Efficiency of a Set of Decision-Making Units," in Klaus Weiermair and Mark Perlman, eds., Studies in Economic Ratio- nality: X-Eficiency Examined and Extolled, Ann Arbor: University of Michigan Press, 1990, pp. 351-68.

Frantz, Roger, X-Eficiency: Theory, Euidence, and Applications, Nonvell, MA: Kluwer, 1988.

Harberger, Arnold, "Monopoly and Resource Allocation," American Economic Reuiew, May 1954 (Papers and Proceedings), 44, 77-87.

Leibenstein, Harvey, "Allocative Efficiency vs. 'X-Efficiency'," American Economic Re- ~jiew, June 1966, 56, 392-415. ,"Organizational or Frictional Equi-

libria, X-Efficiency, and the Rate of Inno- vation," Quarterly Journal of Economics, November 1969, 83, 600-23.

, "On Bull's-Eye Painting Economics," Journal of Post-Keynesian Eco- nomics, Spring 1982, 4, 460-5.

and Maital, Shlomo, "Empirical Esti- mation and Partitioning of X-Inefficiency: A Data-Envelopment Approach," Ameri- can Economic Reuiew, May 1992 (Papers and Proceedings), 82, 428-33.

Mqjumdar, Sumit, "X-Efficiency in Emerging Competitive Markets: The Case of U.S. Telecommunications,~' unpublished manuscript, School of Business Administra- tion, University of Michigan, Ann Arbor, 1991.

Mundell, Robert, [Book Review: L. H. Janssen, Free Trade, Protection and Cus- toms Union], American Economic Reuiew, June 1962, 52, 621-2.

Stigler, George, "The X-istence of X-Efficiency," American Economic Review, March 1976, 66, 213-16.

So long as the wage reflects the marginal product, then, in the long run, workers who demand on-the-job leisure will accept lower wages. Lower productivity is thus balanced by lower wages. The firm does not pay for on-the-job leisure, and hence leisure is an externality. The implica- tion is that management should not care very much about productivity in the long run.
Form does not matter because all forms are simply arrangements of matter (which can neither be created nor destroyed). The implication is that preferences do not matter.
U.S. productivity has not fallen. There has been a shift from producing commodity output to leisure output. The implication here is that there is not any productivity "puzzle" to solve.
  • Recommend Us