Implicit Contracts and fixed Price equilibria OR。 Costas Azariadis; Joseph E. Stiglitz The quarterly Journal of Economics, Vol. 98, Supplement (1983), pp. 1-22 Stable url: http://inksistor.org/sici?sic0033-5533%0281983%02998%3c1%3aicafpe%3e2.0.co%03b2-l The Quarterly Journal of Economics is currently published by The MIT Press Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.htmlJstOr'sTermsandConditionsofUseprovidesinpartthatunlessyouhaveobtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the jsTOR archive only for your personal, non-commercial use Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at Each copy of any part of a JSTOR transmission must contain the same copyright notice that ap on the screen or printed page of such transmission STOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support @jstor. org Tue may I511:18:582007
Implicit Contracts and Fixed Price Equilibria Costas Azariadis; Joseph E. Stiglitz The Quarterly Journal of Economics, Vol. 98, Supplement. (1983), pp. 1-22. Stable URL: http://links.jstor.org/sici?sici=0033-5533%281983%2998%3C1%3AICAFPE%3E2.0.CO%3B2-L The Quarterly Journal of Economics is currently published by The MIT Press. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/mitpress.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. http://www.jstor.org Tue May 15 11:18:58 2007
THE QUARTERLY JOURNAL OF ECONOMICS Vol. XCVIII 1983 Supplement IMPLICIT CONTRACTS AND FIXED PRICE COSTAs AZARIADIS AND JOSEPH E. STIGLITZ ry essay offers a brief guided tour of the main developments ontracts, from its inception to the present. It is not intend as a survey but, rather as an appraisal of the progress that has been made the dif ficulties that remain, and as an outline of the microeconomic and macroeconomic issues hat seem to invite additional work This issue of the Journal brings together several recent contri butions on implicit contracts and quantity-constrained equilibria Almost ten years ago, the theory of implicit contracts signaled a fresh effort by economists to understand the twin empirical regularities of wage stickiness and involuntary unemployment, amid hopes that the microeconomic foundations of Keynesian macroeconomics, especially those of the fixed price method, would be strengthened in the pro- This introductory essay offers a brief guided tour of the ma developments in the theory of implicit contracts, from its inception to the present. Our purpose however, is somewhat different from that of ordinary tourguides: we do not intend to survey the landscape l but, rather, to appraise the progress that has been made to identify some of the difficulties and to outline the microeconomic and macroeco nomic issues that seem to invite additional work We acknowledge with thanks financial support from the National Science Foundat The landscape is surveyed in Azariadis 1979, Hart [1982, Ito [1982 and Schwartz I1982】 lows of Harvard College. Published by John wile t,1983 CCC0033-533/83/030001-22803.20
THE QUARTERLY JOURNAL OF ECONOMICS Vol. XCVIII Supplement IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA* COSTAS AZARIADIS AND JOSEPHE. STIGLITZ This introductory essay offers a brief guided tour of the main developments in the theory of implicit contracts, from its inception to the present. It is not intended as a survey but, rather, as an appraisal of the progress that has been made, the difficulties that remain, and as an outline of the microeconomic and macroeconomic issues that seem to invite additional work. This issue of the Journal brings together several recent contributions on implicit contracts and quantity-constrained equilibria. Almost ten years ago, the theory of implicit contracts signaled a fresh effort by economists to understand the twin empirical regularities of wage stickiness and involuntary unemployment, amid hopes that the microeconomic foundations of Keynesian macroeconomics, especially those of the fixed price method, would be strengthened in the process. This introductory essay offers a brief guided tour of the main developments in the theory of implicit contracts, from its inception to the present. Our purpose, however, is somewhat different from that of ordinary tourguides: we do not intend to survey the landscape1 but, rather, to appraise the progress that has been made, to identify some of the difficulties, and to outline the microeconomic and macroeconomic issues that seem to invite additional work. * We acknowledge with thanks financial support from the National Science Foundation. 1. The landscape is surveyed in Azariadis [1979], Hart [1982], Ito [1982], and Schwartz [1982]. c_ 1983hy the President and Fellows of Harvard College. Published by John Wile?:& Sons, Inc. The Quarterl~ Journal of Economics, Vol. 98. Supplement, 1983 CCC 0033-55:13/8:31030001-22$0:1.20
QUARTERLY JOURNAL OF ECONOMICS We begin the tour in Section ii with a description of the empirical regularities to be explained, and review in Sections III and iv the major insights of implicit contract theory-from the older, public information literature, as well as from more recent work on asym metric information. The newer literature makes heavy use of som concepts common to many self-selection problems; we discuss these lrb epts in Section V Section VI covers macroeconomic aspects of implicit contracts-in particular, their relation to the fixed price literature. The concluding section is concerned with a survey of un- resolved issues II Over a typical business cycle average wages fluctuate less vig orously than does labor's marginal revenue product or, for that matter, the total volume of employment(see Hall 1980 ). The great De pression is a sad illustration: from 1929 to 1933 U.S. employment fell precipitously, while real wages managed to creep upward At a less aggregative level, it is standard collective bargaining procedure to predetermine money wage rates for two or three years in advance, even though wage rigidity does not promote employment in recessions. 2 The sluggishness of money wage rates, notably in periods of relatively stable inflation, and the strong contribution of layoffs to yclical unemployment in north america have long been two of the best-documented stylized facts in economics. Wage and price rigidity are also among the key assumptions of Keynesian macroeconomics, both in the Hicksian IS/LM framework(see Hicks [1937 )and in the very interesting concept of quantity-constrained equilibrium origi nally developed by Patinkin [1956, Clower [1965], Hansen [1974] Solow-Stiglitz[1968, Younes[1970, and Barro-Grossman [1971 and formalized by European economists in the 1970s. 4 Keynes's own explanation of wage rigidity [1936, p. 13-15] was a sophisticated form of money illusion; workers resist cuts in money rates because they do not know how widespread these cuts will to be, each worker fearing a fall in his own wage relative to s Relative wage arguments suggest that "fairness "in the wage 2. However, as the recer nce in the United States indicates, if too large the contract. Cousineau and Lacroix [1981] ar an interesting set of ning agreements and Malinvaud 1977] relopmeats appear in Benassy [1975 Dreze[1975], Younes( 1975)
2 QUARTERLY JOURNAL OF ECONOMICS We begin the tour in Section I1 with a description of the empirical regularities to be explained, and review in Sections I11 and IV the major insights of implicit contract theory-from the older, publicinformation literature, as well as from more recent work on asymmetric information. The newer literature makes heavy use of some concepts common to many self-selection problems; we discuss these concepts in Section V. Section VI covers macroeconomic aspects of implicit contracts-in particular, their relation to the fixed price literature. The concluding section is concerned with a survey of unresolved issues. Over a typical business cycle, average wages fluctuate less vigorously than does labor's marginal revenue product or, for that matter, the total volume of employment (see Hall [1980]). The Great Depression is a sad illustration: from 1929 to 1933 U.S. employment fell precipitously, while real wages managed to creep upward. At a less aggregative level, it is standard collective bargaining procedure to predetermine money wage rates for two or three years in advance, even though wage rigidity does not promote employment in recession^.^ The sluggishness of money wage rates, notably in periods of relatively stable inflation, and the strong contribution of layoffs to cyclical unemployment in North America have long been two of the best-documented stylized facts in economic^.^ Wage and price rigidity are also among the key assumptions of Keynesian macroeconomics, both in the Hicksian ISILM framework (see Hicks [1937]) and in the very interesting concept of quantity-constrained equilibrium originally developed by Patinkin [1956], Clower [1965], Hansen [1974], Solow-Stiglitz [1968], Youn6s [1970], and Barro-Grossman [1971], and formalized by European economists in the 1970~.~ Keynes's own explanation of wage rigidity [1936, p. 13-15] was a sophisticated form of money illusion; workers resist cuts in money wage rates because they do not know how widespread these cuts will prove to be, each worker fearing a fall in his own wage relative to others. Relative wage arguments suggest that "fairness" in the wage 2. However, as the recent experience in the United States indicates, if too large a level of unemployment is caused by wage rigidity, both sides may agree to renegotiate the terms of the contract. Cousineau and Lacroix [I9811 analyze an interesting set of data collected from Canadian collective bargainin 3. An example is the work of Feldstein [I976 4. The main developme.~ts appear in Nnassy YounGs [1975], and Malinvaud [1977]
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA structure is a factor to be reckoned with in labor supply decisions,5 but do not develop an operational definition of "fairness. 6 This is perhaps one reason why the relative wage argument did not gain ground in economics Students of human capital8 have provided another theory of layoffs, namely, that the accumulation of job-specific skills requires the sinking of certain expenses for hiring and training. This is an in- vestment the employer makes in anticipation that the worker will remain attached to his job, and one he amortizes over time by paying a wage rate lower than the trained employee's marginal contribution to the firm. If the firm should need to reduce employment in periods of slack demand, it will naturally choose to lay off first the least trained members of its labor force, those who represent the smallest undepreciated investment in training. This story is a satisfactory explanation of the incidence of layoffs g not of their existence; it tells us why layoffs fall on the least skilled workers but leaves open the question why they occur in the first place which is our concern here Furthermore the technical heterogeneity of labor that is crucial for this argument is itself an unnecessary complication in traditional macroeconomic models that are built on the simpler assumptions of homogeneous inputs and zero transaction The innovation in the early literature on implicit contracts Baily 1974; Gordon, 1974; Azariadis, 1975]was to view the employment relation not simply as a sequential spot exchange of labor services for money, but as a more complicated long-term attachment; labor ser- vices are traded for an insurance contract that protects workers from random, publicly observed fluctuations in their marginal revenue product. The idea, shown in Figure I, is that workers can purchase insurance only from their employers, not from third parties. Risk-averse workers deal with risk-neutral entrepreneurs whose firms consist of three departments: a production department that purchases labor services and credits each worker with his marginal See Akerlof (1980) for a recent attempt at a theory of wage rigidity based on 8. The standard refere e is Becker[ 1964]; our argument is due to Oi [1962]. odel of layoff incidence, see Azariadis [1976
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 3 structure is a factor to be reckoned with in labor supply decisions,5 but do not develop an operational definition of "fairness."6 This is perhaps one reason why the relative wage argument did not gain ground in economics.7 Students of human capital8 have provided another theory of layoffs, namely, that the accumulation of job-specific skills requires the sinking of certain expenses for hiring and training. This is an investment the employer makes in anticipation that the worker will remain attached to his job, and one he amortizes over time by paying a wage rate lower than the trained employee's marginal contribution to the firm. If the firm should need to reduce employment in periods of slack demand, it will naturally choose to lay off first the least trained members of its labor force, those who represent the smallest undepreciated investment in training. This story is a satisfactory explanation of the incidence of layoffsg not of their existence; it tells us why layoffs fall on the least skilled workers but leaves open the question why they occur in the first place, which is our concern here. Furthermore, the technical heterogeneity of labor that is crucial for this argument is itself an unnecessary complication in traditional macroeconomic models that are built on the simpler assumptions of homogeneous inputs and zero transaction costs. The innovation in the early literature on implicit contracts [Baily, 1974; Gordon, 1974; Azariadis, 19751 was to view the employment relation not simply as a sequential spot exchange of labor services for money, but as a more complicated long-term attachment; labor services are traded for an insurance contract that protects workers from random, publicly observed fluctuations in their marginal revenue product. The idea, shown in Figure I, is that workers can purchase insurance only from their employers, not from third parties. Risk-averse workers deal with risk-neutral entrepreneurs whose firms consist of three departments: a production department that purchases labor services and credits each worker with his marginal 5. This is apparent in Okun's posthumous book [1981], pp. 93-97. 6. Such a definition was later developed in welfare economics; see Varian [1974], Schmeidler and Yaari [1971]. 7. See Akerlof 119801 for a recent attempt at a theory of wage rigidity based on "norms." 8. The standard reference is Becker [1964]; our argument is due to Oi [1962]. 9. For a contractual model of layoff incidence, see Azariadis i19761
QUARTERLY JOURNAL OF ECONOMICS WORKER PRODUCTIONMRPL ACCOUNTI NG FIGURE I revenue product (MRPL); an insurance department that sells actu arially fair policies, and depending on the state of nature credits the worker with a net insurance indemnity (Nid)or debits him with a net insurance premium; and an accounting department that pays each employed worker a wage w with the property that w= MRPL + NII in every state of nature. Favorable states of nature are associated with high values of MRPL; in these the net indemnity is negative, and wage falls short of the MRPL, Adverse states of nature correspond to low values of MRPL, to positive net insurance indemnities, and to wages in excess of MRPL. An implicit contract is then a complete description, made before the state of nature becomes known, of the labor services to be rendered unto the firm in each state of nature, and of the corre sponding payments to be delivered to the worker. The contract is implementable if we assume the state of nature is directly observed by all sides An immediate consequence of this framework is that wages are disengaged from the marginal revenue product of labor. In fact, if we fix institutionally the amour nt of labor performed by employed workers, then each worker's consumption is proportional to the wage rate; an actuarially fair insurance policy should make this consump- tion independent of the MRPL by stabilizing the purchasing power of wages over states of nature. Ergo, the real wage rate is rigid In traditional macroeconomic models, of course, wage rigidity by itself is sufficient to cause unemployment: if wages do not adjust or some reason, than neither does the demand for labor. The argu
4 QUARTERLY JOURNAL OF ECONOMICS 1 PRODUCTION MRPL _ ACCOUNT1 NG NII lNSURANCE DEPT. DEPT. DE PT. t f t FIRM FIGUREI revenue product (MRPL); an insurance department that sells actuarially fair policies, and depending on the state of nature, credits the worker with a net insurance indemnity (NII) or debits him with a net insurance premium; and an accounting department that pays each employed worker a wage w with the property that w = MRPL + NII in every state of nature. Favorable states of nature are associated with high values of MRPL; in these the net indemnity is negative, and wage falls short of the MRPL. Adverse states of nature correspond to low values of MRPL, to positive net insurance indemnities, and to wages in excess of MRPL. An implicit contract is then a complete description, made before the state of nature becomes known, of the labor services to be rendered unto the firm in each state of nature, and of the corresponding payments to be delivered to the worker. The contract is implementable if we assume the state of nature is directly observed by all sides. An immediate consequence of this framework is that wages are disengaged from the marginal revenue product of labor. In fact, if we fix institutionally the amount of labor performed by employed workers, then each worker's consumption is proportional to the wage rate; an actuarially fair insurance policy should make this consumption independent of the MRPL by stabilizing the purchasing power of wages over states of nature. Ergo, the real wage rate is rigid. In traditional macroeconomic models, of course, wage rigidity by itself is sufficient to cause unemployment: if wages do not adjust for some reason, than neither does the demand for labor. The argu-
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 5 ment does not carry over to implicit contracts because of the very separation between wages and the marginal revenue product of labor. A complete theory of unemployment must explain why layoffs are preferred to work sharing in adverse states of nature and why laid-off workers are worse off than their employed colleagues. This is not a simple task. Suppose for instance that employers are risk-neutral and that workers' preferences over consumption and leisure can be represented by a strictly quasi-concave, additively separable utility function. Then optimum contracts will result in complete work-sharing [Mortensen, 1978; and if such work-sharing s less profitable than layoffs for technological reasons(e.g, workers produce most efficiently when they put in a full-day' s effort), an op timum contract under perfect information will still equate th workers' marginal utility of consumption in states of employment and unemployment. Individuals may thus voluntarily ei ployed they would rather be laid off than work The resolution of this quandary has been the objective of much recent research on the theory of implicit contracts. The papers of this symposium represent a good step forward, but as we shall see later many questions remain unresolved. To explain unemployment, we need to complicate the analysis in some important way. Some of the complications arise from familiar problems in explicit(as opposed to implicit) insurance contracts, but a few of the problems are peculiar to implicit contracts One distortion that was noted early in the implicit contract lit erature concerns the role of the dole. In very adverse states of nature the flow of insurance indemnities to workers can become a substantial drain on profit; one way to staunch losses is to place the burden of insurance on an outside party, the dole(see Figure I). The practice of layoffs is simply the administrative counterpart of this insur ance-shifting maneuver; workers consent in advance that some of them may be separated from their jobs in order to become eligible for unemployment insurance (UI)payments from an outside public agency. Furthermore, no worker will contract his labor, unless the expected value (utility) of the total package taken over all possible states of nature exceeds the value of being on the dole in every state This means, in turn, that employed workers receive a wage in excess of Ui payments, and are therefore to be envied by their laid- off col- leagues-a situation that many economists would call"involuntary unemployment” This particular insurance contract between a third party(the government)and the other two parties(workers, firmsis not neces-
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 5 ment does not carry over to implicit contracts because of the very separation between wages and the marginal revenue product of labor. A complete theory of unemployment must explain why layoffs are preferred to work sharing in adverse states of nature, and why laid-off workers are worse off than their employed colleagues. This is not a simple task. Suppose, for instance, that employers are risk-neutral and that workers' preferences over consumption and leisure can be represented by a strictly quasi-concave, additively separable utility function. Then optimum contracts will result in complete work-sharing [Mortensen, 19781;and if such work-sharing is less profitable than layoffs for technological reasons (e.g., workers produce most efficiently when they put in a full-day's effort), an optimum contract under perfect information will still equate the workers' marginal utility of consumption in states of employment and unemployment. Individuals may thus become involuntarily employed: they would rather be laid off than work. The resolution of this quandary has been the objective of much recent research on the theory of implicit contracts. The papers of this symposium represent a good step forward, but as we shall see later, many questions remain unresolved. To explain unemployment, we need to complicate the analysis in some important way. Some of the complications arise from familiar problems in explicit (as opposed to implicit) insurance contracts, but a few of the problems are peculiar to implicit contracts. One distortion that was noted early in the implicit contract literature concerns the role of the dole. In very adverse states of nature, the flow of insurance indemnities to workers can become a substantial drain on profit; one way to staunch losses is to place the burden of insurance on an outside party, the dole (see Figure I). The practice of layoffs is simply the administrative counterpart of this insurance-shifting maneuver; workers consent in advance that some of them may be separated from their jobs in order to become eligible for unemployment insurance (UI) payments from an outside public agency. Furthermore, no worker will contract his labor, unless the expected value (utility) of the total package taken over all possible states of nature exceeds the value of being on the dole in every state. This means, in turn, that employed workers receive a wage in excess of UI payments, and are therefore to be envied by their laid-off colleagues-a situation that many economists would call "involuntary unemployment." This particular insurance contract between a third party (the government) and the other two parties (workers, firms) is not neces-
QUARTERLY JOURNAL OF ECONOMICS sarily efficient. It may, however, be the only feasible way of providing third- party insurance; theoretically it is preferable that the govern idemnity to the firm when its profit is loy to the worker when his income is low. The government, however cannot always ascertain with precision the actual income or the op- portunity sets of individuals; what it insures, therefore, is not an ex ogenous event but an endogenous variable that is more readily ob servable, and that, under reasonable circumstances, is correlated with the exogenous event. This creates an important moral hazard prob lem 10 to which we shall return in Sections Iv and v nother source of problems for implicit contracts--which applies as well to the insurance literature but has even more force here- the enforceability of contracts. Implicit contracts are just that- mplicit--and one must ask what happens when either side deviates from the contract. Because the contracts are implicit, contracting parties may not have any legal recourse against breach Contracts lust thus either be self-enforcing or be enforced through tions To put the issue in plainer terms, let us focus on the worker: If his wage on average equals his marginal revenue product, what is to stop him from quitting in the good states, when his marginal revenue product is greater than his wage? The worker would thereby receive the benefits of the insurance offered by the firm (when the wage re ceived exceeds the value of his marginal revenue product), and would refuse to pay the insurance premiums. What is to stop him from reneging on his"“ implicit”’ contract? One early answer focused on the role of reputation: workers on contract might choose to reject outside offers at higher wages if, by doing so, they established a reputation for"reliability?"that would enable them subsequently to attract the preferential contracts handed out to“ reliable” workers The precise manner in which one acquires a particular reputation is rather hard to analyze Fortunately, we do not have to, for reputa tion is essential to the enforceability of implicit labor contracts only within the artificial confines of single-period contracts. Bengt Holmstrom demonstrates the point admirably in his paper"Equi librium Long- Term Labor Contracts"this ournal Holmstrom al lows workers to sign multiperiod contracts that they can abrogate at no cost after one period if they find a higher-paying job in the spot 10. A standard early reference on moral hazard is Arrow (1971 for a more recent Grossman (1977 was among the first to point out this problem
6 QCARTERLY JOURNAL OF ECONOMICS sarily efficient. It may, however, be the only feasible way of providing third-party insurance; theoretically it is preferable that the government pay a lump sum indemnity to the firm when its profit is low, or to the worker when his income is low. The government, however, cannot always ascertain with precision the actual income or the opportunity sets of individuals; what it insures, therefore, is not an exogenous event but an endogenous variable that is more readily observable, and that, under reasonable circumstances, is correlated with the exogenous event. This creates an important moral hazard problemlo to which we shall return in Sections IV and V. Another source of problems for implicit contracts-which applies as well to the insurance literature but has even more force here-is the enforceability of contracts. Implicit contracts are just thatimplicit-and one must ask what happens when either side deviates from the contract. Because the contracts are implicit, contracting parties may not have any legal recourse against breach. Contracts must thus either be self-enforcing or be enforced through reputations. To put the issue in plainer terms, let us focus on the worker: If his wage on average equals his marginal revenue product, what is to stop him from quitting in the good states, when his marginal revenue product is greater than his wage? The worker would thereby receive the benefits of the insurance offered by the firm (when the wage received exceeds the value of his marginal revenue product), and would refuse to pay the insurance premiums. What is to stop him from reneging on his "implicit" contract? One early answer focused on the role of reputation: workers on contract might choose to reject outside offers at higher wages if, by doing so, they established a reputation for "reliability" that would enable them subsequently to attract the preferential contracts handed out to "reliable" workers. The precise manner in which one acquires a particular reputation is rather hard to analyze. Fortunately, we do not have to, for reputation is essential to the enforceability of implicit labor contracts only within the artificial confines of single-period contracts. Bengt Holmstrom demonstrates the point admirably in his paper "Equilibrium Long-Term Labor Contracts" [this Journal].Holmstrom allows workers to sign multiperiod contracts that they can abrogate at no cost after one period if they find a higher-paying job in the spot 10. A standard early reference on moral hazard is Arrow jl97lJ:tor a more recent treatment see Arnott and Stiglitz 11982). 11. H. Grossman [I9771 was among the first to point out this problem
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 7 market. Nevertheless, equilibrium contracts will be structured so that workers choose not to annul them: wages in the first period, when workers cannot leave are lower than(the expected value of) MRPL in the second period wages rise to equal either a state-invariant rate or the spot rate, whichey New workers on contract thus pay the firm a"bond"that assures they will behave reliably in the future. As the bond is amortized over time, veteran employees receive higher wages than rookies do--at least as high, in fact, as spot wages. Holmstrom's multiperiod equi librium thus yields reliable behavior on the part of workers(his firms being reliable by definition), wage differentials by seniority class, and a weakening of strict wage rigidity to downward rigidity Holmstrom's argument parallels the standard argument as to how the firm recovers the costs of specific training of workers. 12If workers have limited access to the capital market, increased mobility implies that their consumption stream over time is not so smooth as it otherwise would be and there is a welfare loss as a result in addi tion, however, workers may need to leave the firm for a variety of good reasons(their health is bad, their mother- in-law moves to a nearby state, etc.) Unfortunately, there is no easy way of distinguishing these le- gitimate motives for quitting from the opportunistic motives (i.e simply reneging on the contract). Hence any contract that requires workers to post bonds imposes some risk on them-the risk that they forfeit the bond even if they desire to change jobs for noneconomic reasons. As a result, there will seldom be"complete"bonding. Finally there is always another risk associated with any theory of contract enforcement through bonding: that the employer will fire the worke (or, equivalently, make work conditions so unattractive that the worker will be induced to quit and forfeit the bond). 13 To avoid these difficulties, either a far more complicated bonding scheme must be established, or we must rely on a theory of reputation for firms Let us return to a simpler world in which firms are thoroughly trustworthy and workers never quit for family reasons. Having at least assured ourselves that we can redesign the time path of wage pay ments to extract reliable behavior from workers, we go back to the single-period enforceable contract structure of Figure I to reflect on
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 7 market. Nevertheless, equilibrium contracts will be structured so that workers choose not to annul them: wages in the first period, when workers cannot leave, are lower than (the expected value of) MRPL; in the second period wages rise to equal either a state-invariant rate or the spot rate, whichever is greater. New workers on contract thus pay the firm a "bond" that assures they will behave reliably in the future. As the bond is amortized over time, veteran employees receive higher wages than rookies do-at least as high, in fact, as spot wages. Holmstrom's multiperiod equilibrium thus yields reliable behavior on the part of workers (his firms being reliable by definition), wage differentials by seniority class, and a weakening of strict wage rigidity to downward rigidity. Holmstrom's argument parallels the standard argument as to how the firm recovers the costs of specific training of workers.12 If workers hhve limited access to the capital market, increased mobility implies that their consumption stream over time is not so smooth as it otherwise would be, and there is a welfare loss as a result. In addition, however, workers may need to leave the firm for a variety of good reasons (their health is bad, their mother-in-law moves to a nearby state, etc.). Unfortunately, there is no easy way of distinguishing these legitimate motives for quitting from the opportunistic motives (i.e., simply reneging on the contract). Hence, any contract that requires workers to post bonds imposes some risk on them-the risk that they forfeit the bond even if they desire to change jobs for noneconomic reasons. As a result, there will seldom be "complete" bonding. Finally, there is always another risk associated with any theory of contract enforcement through bonding: that the employer will fire the worker (or, equivalently, make work conditions so unattractive that the worker will be induced to quit and forfeit the bond).13 To avoid these difficulties, either a far more complicated bonding scheme must be established, or we must rely on a theory of reputation for firms. Let us return to a simpler world in which firms are thoroughly trustworthy and workers never quit for family reasons. Having at least reassured ourselves that we can redesign the time path of wage payments to extract reliable behavior from workers, we go back to the single-period enforceable contract structure of Figure I to reflect on 12. A more extensive treatment appears in Arnott and Stiglitz 11981J. 13. See Shapiro and Stiglitz [I9821 for a detailed discussion of this problem
ARTERLY JOURNAL OF ECONOMICS the nature of layoff unemployment 4 How close is the unemployment we discussed in the previous section to the involuntary unemployment that economists are so concerned about? The fact that laid-off workers would gladly exchange places with their employed colleagues is not in itself sufficient to establish a misallocation of resources. After all, accident victims may very well envy more fortunate individuals without any implication that the insurance industry works poorly Layoffs, by themselves, could be no more than the luck of the draw unless we can demonstrate that they constitute, in some sense, socially inefficient underemployment. This clearly impossible within the Walras-Arrow-Debreu model There are, in fact, two distinct questions that we can pose. One is do limitations on information . transactions costs etc when for mally modeled into the optimal design of the implicit contract, lead to levels of employment that are systematically lower than would occur in a Walrasian equilibrium? The second is, taking these limi tations on information transactions costs etc into account, can design (say, through tax policy)a Pareto improvement in the economy? Like most of the literature, this symposium focuses on the first question: i. e, on conditions under which market equilibrium might be characterized by layoffs, or by hours worked being less th in the corresponding Walrasian equilibrium One fundamental departure from the walrasian paradigm that seems much in the spirit of implicit contracts is to alter the informa tional assumptions; information is no longer“" public”or“ symmetric,” it is"private"or"asymmetric, since only one side of the observes the relevant state of nature. Four of the papers that appear in this issue(by azariadis, Chari, Green and Kahn, and grossman and Hart)study the properties of implicit contracts when the value of labor's marginal revenue product is known only to the entrepre neur Asymmetric information is essential for a thorough under standing of implicit contracts and, as we shall see later in this essay, for their use in macroeconomics as well. what justifies the trading of these contracts in the first place is that third parties simply are not as well informed about someone's income or employment status as is his employer; the employer, in turn, may be less informed about an yaza 4. Developments here were greatly influenced by the paper of Akerlof and Mi o and Phelps [1977 were the first to pose this question; for related work see Hall and Lilien 1979
8 QUARTERLY JOURNAL OF ECONOMICS the nature of layoff unemployment.14 How close is the unemployment we discussed in the previous section to the involuntary unemployment that economists are so concerned about? The fact that laid-off workers would gladly exchange places with their employed colleagues is not in itself sufficient to establish a misallocation of resources. After all, accident victims may very well envy more fortunate individuals without any implication that the insurance industry works poorly. Layoffs, by themselves, could be no more than the luck of the draw unless we can demonstrate that they constitute, in some sense, socially inefficient underemployment. This is clearly impossible within the Walras-Arrow-Debreu model. There are, in fact, two distinct questions that we can pose. One is, do limitations on information, transactions costs, etc., when formally modeled into the optima1 design of the implicit contract, lead to levels of employment that are systematically lower than would occur in a Walrasian equilibrium? The second is, taking these limitations on information, transactions costs, etc., into account, can we design (say, through tax policy) a Pareto improvement in the economy? Like most of the literature, this symposium focuses on the first question: i.e., on conditions under which market equilibrium might be characterized by layoffs, or by hours worked being less than in the corresponding Walrasian equilibrium. One fundamental departure from the Walrasian paradigm that seems much in the spirit of implicit contracts is to alter the informational assumptions: information is no longer "public" or "symmetric," it is "private" or "asymmetric," since only one side of the market observes the relevant state of nature. Four of the papers that appear in this issue (by Azariadis, Chari, Green and Kahn, and Grossman and Hart) study the properties of implicit contracts when the value of labor's marginal revenue product is known only to the entrepreneur.15 Asymmetric information is essential for a thorough understanding of implicit contracts and, as we shall see later in this essay, for their use in macroeconomics as well. What justifies the trading of these contracts in the first place is that third parties simply are not as well informed about someone's income or employment status as is his employer; the employer, in turn, may be less informed about an 14. 1)evelopments here were greatly influenced by the paper of Akerlof and Miyazaki [1980]. 15. Calvo and Phelps [1977] were the first to pose this question; for related work see Hall and 1,ilien [1979]
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 9 employee s nonlabor income and job opportunities than is the worker himself Let us face these complications one at a time. First, how does the market write and evaluate state-contingent contracts when the state itself is observed by employers alone? Suppose that s is the set of all possible states, and consider the possibility that wages and employ- ment are predetermined functions of the state announced by the 8*, which the parties find optimal under symmetric information; the employer observes the true state s, announces some state 0, and the age-employment combination is whatever d* specifies for A. will the employer tell the truth by announcing s =0? If so, we say that o* is implementable; if not, we must pick another contract. We recall from the previous section that the contract o*does not generally equate wages with MRPL, which implies that, if the firm told the truth, it would not maximize profit in state s. To do so for sufficiently adverse states, the firm will sometimes(that is, for some but not all, possible combinations of worker and entrepreneur pref erences)announce a state worse than what actually occurs If an optimum public-information contract is unimplementable under asymmetric information, we know from economic theory6 that it can be appropriately modified to motivate entrepreneurs to tell the truth in each state of nature This is accomplished by making truth the value-maximizing strategy for firms in each state of nature All four symposium papers on asymmetric information exploit this straightforward idea but differ in operational details. The papers by azariadis, and Grossman and Hart begin with optimum symmet- ric-information contracts that are not implementable when infor- mation is asymmetric because entrepreneurs will announce a state lower than actual to cure this lack of incent atibility, em ployment in all states but the highest is reduced below its optimum symmetric-information value. By appropriately restricting the level of employment and restructuring wages, truthful announcement come to be in the entrepreneur's own best interest. of nature, except the lowest, beyond its full-information lev % Ap In the papers by Chari, and green and Kahn, however, ent preneurs'state announcements are biased upward under that op lum full-information contract. To change the incentive structure asymmetric-information contracts increase employment in all states 16.Th e fundamental ideas on allocation mechanisms with the revelation property are developed rerson( 1979] and Harris and Townsend (1981
IMPLICIT CONTRACTS AND FIXED PRICE EQUILIBRIA 9 employee's nonlabor income and job opportunities than is the worker himself. Let us face these complications one at a time. First, how does the market write and evaluate state-contingent contracts when the state itself is observed by employers alone? Suppose that S is the set of all possible states, and consider the possibility that wages and employment are predetermined functions of the state announced by the employer. For the sake of concreteness, we begin with the contract 6*, which the parties find optimal under symmetric information; the employer observes the true state s, announces some state 8, and the wage-employment combination is whatever 6* specifies for 19.Will the employer tell the truth by announcing s = 8? If so, we say that 6* is implementable; if not, we must pick another contract. We recall from the previous section that the contract 6* does not generally equate wages with MRPL, which implies that, if the firm told the truth, it would not maximize profit in state s. To do so for sufficiently adverse states, the firm will sometimes (that is, for some, but not all, possible combinations of worker and entrepreneur preferences) announce a state worse than what actually occurs. If an optimum public-information contract is unimplementable under asymmetric information, we know from economic theory16 that it can be appropriately modified to motivate entrepreneurs to tell the truth in each state of nature. This is accomplished by making truth the value-maximizing strategy for firms in each state of nature. All four symposium papers on asymmetric information exploit this straightforward idea but differ in operational details. The papers by Azariadis, and Grossman and Hart begin with optimum symmetric-information contracts that are not implementable when information is asymmetric because entrepreneurs will announce a state lower than actual; to cure this lack of incentive-compatibility, employment in all states but the highest is reduced below its optimum symmetric-information value. By appropriately restricting the level of employment and restructuring wages, truthful announcements come to be in the entrepreneur's own best interest. In the papers by Chari, and Green and Kahn, however, entrepreneurs' state announcements are biased upward under that optimum full-information contract. To change the incentive structure, asymmetric-information contracts increase employment in all states of nature, except the lowest, beyond its full-information level. 16. The fundamental ideas on allocation mechanisms with the revelation property are developed in Myerson [I9791 and Harris and Townsend [1981]
