انگیزش از طریق تعیین هدف

We study a principal agent model where agents derive a sense of pride from accomplishing production goals. As in classical models, the principal offers a pay-per-performance wage to the agent, determining the agent’s extrinsic incentives. However, in our model, the principal uses goal setting policies as a tool to manage agents’ intrinsic motivation. To capture the idea that different agents respond differently to different goals we introduce the concept of personal standards which determine what becomes challenging and rewarding to them, and hence the intensity of their intrinsic motivation to achieve goals. We show that, at the optimal contract, the agents’ production, as well as the goals set by the principal, increase with the agents’ personal standards. Moreover, we show that an intrinsically motivated agent gets higher surplus than an agent with no intrinsic motivation in the form of informational rents but an agent with a mid-ranged standard (and hence productivity) could end up being the one most satisfied. Therefore, our model can be helpful to explain some empirical findings in the literature of job satisfaction such as the so called “paradox of happiness”.

In 1968, the American Pulpwood Association became concerned about how to increase its loggers’ productivity as mechanization alone was not increasing the productivity of its logging crews. Two Industrial Organization psychologists – Edwin A. Locke and Gary P. Latham – assured the firm’s managers that they had found a way to increase productivity at no financial expense to anyone. The policy seemed too easy; it merely involved setting specific production goals for the loggers. The novelty was that these goals were wage irrelevant, in contrast with classical wage relevant goals such as bonuses. The psychologists argued that introducing a goal that was difficult but attainable, would increase the challenge of the job while making it clear to the workers what was expected from them. Although the managers were quite skeptical at the beginning, the results were surprising: the performance of logging crews increased 18% and the firm’s profits rose as well.1 This example was followed by many studies in the psychology literature on what is known as “goal setting” (e.g., Yukl and Latham, 1978, Shane et al., 2003 and Anderson et al., 2010).2 The theory states that wage irrelevant goals are an important determinant of employees’ motivation to work and hence affect their productivity (see Locke (1996) and Locke and Latham (2002) for a literature review). The effectiveness of goal-setting is very consistent across studies. According to Locke and Latham (2002) the probability that performance increase after a goal has been set is above 90%. Although in principle we can think about standard explanations of the goal setting evidence, for instance, a goal may be an implicit benchmark for being retained or for future promotions, it is important to remark that there are numerous laboratory experiments showing that individuals who have been assigned specific wage irrelevant goals solve more arithmetic problems or assemble more tinker toys than do people without goals (see Locke, 1996). Therefore, this evidence indicates that there is an important component of individuals motivation through goal setting policies that cannot be explained with classical “carrot and stick” (i.e., wage and threat of fire) arguments.3 Our purpose in this paper is to take this motivation theory addressed in psychology and management and make it precise in standard economic theory. In particular, we propose a model where workers do have a sense of self-achievement and may care about pay-off irrelevant goals. This sense of achievement is different for workers with different personal standards, which is private information to them. In our model, a worker with a high personal standard can only be motivated to accomplish a sufficiently demanding goal. Before describing the key elements of the model, we start by summarizing the main findings in the goal setting literature. The most important and robust finding is that the more difficult the goal is, the greater the achievement will be. This result applies as long as the individual is committed to the goal (i.e., he cares about it).4 The reason why goals affect workers’ achievement is that goals affect the challenge of the job and hence the satisfaction workers obtain from the work itself. As Judge (2000) says: The most effective way an organization can promote job satisfaction of its employees is to enhance the mental challenge in their jobs, and the most consequential way most individuals can improve their own satisfaction is to seek out mentally challenging work (Judge (2000, p. 107). Therefore, goals are an important determinant of workers’ satisfaction because they help develop a sense of achievement. According to the goal setting literature, goals serve as a reference point of self-satisfaction, with harder goals leading to better achievements. Since goals are reference points, it is also plausible that a higher goal lowers the workers’ satisfaction. In fact, supporting this reasoning, Mento, Klein, and Locke (1992) have found that those who produce the most, those with difficult goals, are the least satisfied.5 The question then is why do people accept these goals? According to Locke and Latham, 1990 and Locke and Latham, 2002, the driving force behind this result is that those people with high goals demand more from themselves, thus they are dissatisfied with less. Therefore, their personal standards are higher.6 As Locke and Latham (1990) indicate: …a person with higher standards has to accomplish more to feel that he or she has performed adequately or successfully than the person with lower standards (Locke and Latham (1990, p. 242). Bandura (1988) also argues that individuals use standards for judging the adequacy of their performance. Similarly, Erez and Zidon (1984) and Locke, Latham, and Erez (1988) provide experimental evidence that individuals only accept goals if they are higher than their personal standard. According to this evidence, the accomplishment of an easy goal can only be rewarding for those individuals with low personal standards. As we will see, this personal standards’ idea is going to play a key role in our analysis.7 The previous findings are difficult to support with traditional economic models, such as the classical principal agent model, in which only the goals that are directly linked to the agents’ wage (e.g., bonuses) affect their incentives to work. Our purpose here is to fill this gap by introducing goal setting into an economic model of managerial incentives. Therefore, we look at the following questions: Can a manager increase the workers’ productivity by using goals that are linked to the job’s challenge? How should the manager define the workers’ goals? What are the determinants of job satisfaction? To answer these questions we propose a principal agent model where the agent’s motivation to work is twofold. First, as in standard models, the agent works in response to extrinsic incentives, which in our model are a pay-per-performance wage. Second, the agent has an intrinsic motivation to work because he derives an internal sense of achievement from accomplishing goals.8 Coming back to our introductory example, we can easily imagine harvesting timber to be a monotonous and boring task. However, as we have seen, by setting demanding production goals, the managers were able to increase the challenge of the job and provide the loggers with a sense of accomplishment that increased their intrinsic motivation to work and hence their performance. In this paper, we capture this effect with a goal payoff function, which measures the intrinsic satisfaction that an agent receives from his production with respect to the goal set by the principal. Thus, an agent gets a positive goal payoff if he produces above and beyond the target set but a negative goal payoff otherwise. Workers, however, differ in their perception of how challenging goals may be. For instance, we may observe that for loggers who demand more from themselves, only those goals that require a greater amount of timber to be harvested will be found challenging. On the other hand, for those loggers who demand little from themselves, lower goals may be just as challenging. We model this goal commitment effect with a reference dependent function in which the reference point is the agent’s own standard. In particular, we consider that those workers with high personal standards are only committed to high goals.9 In our model, agents only differ in their personal standards. Hence, agents with different standards can be motivated differently by the same goal because some of them may consider it to be challenging while others do not. Therefore, the principal will design different contracts (with different goals) for different agent types. We show that at the optimal contract, goals are met by agents and thus they derive a positive intrinsic utility. We also show that the agents’ production as well as the goals set by the principal increase with the agents’ standard. Thus, in our model, goals that are non-binding for the agent, i.e., they are payoff irrelevant for the principal; increase the principal’s profits with respect to the classical principal agent model with no goals. As in classical principal agent models, the principal distorts the low type’s contract in such a way that his production decreases with the standard of higher types. With respect to the utility that agents get in equilibrium, we show two important results. First, in our two type model, the utility of the high type is an inverted U-shaped function of the agent’s standard. Thus, the most satisfied agent is a high type with a mid-ranged standard. Second, in a three type case we show that a mid-ranged agent type could be the one most satisfied. In fact, although the highest type achieves the highest production he can receive a zero utility. The intuition is as follows, if the highest type’s standard is sufficiently high he does not consider the goals assigned to lower types to be challenging, thus his informational rents are zero. These are novel results in principal agent models that provide new insights for economists and psychologists. In particular, the interplay between personal standards and the optimal contract offered by the principal have important implications for a better understanding of job satisfaction and the called “paradox of happiness” where the highest levels of income do not necessarily correspond with the highest individual happiness.10 While in recent years the problem of goal setting has become an extremely popular topic in psychology and management, the idea of goals that are not linked to the workers’ wage may have an economic effect which thus far has received very little attention. Some exceptions deserve to be mentioned. Some papers study the effects of a self-set goal to attenuate the self-control problems of dynamically inconsistent agents. For instance, Hsiaw (2009) studies an optimal stopping problem (or a project termination decision) with hyperbolic discounters in which there is an option value of waiting due to uncertainty. In her model, goals, which act as a reference point up to which agents get an additional positive utility, induce more patient behavior by providing an additional incentive to wait for a higher realization of the project’s value.11 Therefore, the main result is that endogenous goal setting attenuates the impulsiveness of an agent with present-biased time preferences. In our model we use assigned goals in a principal agent model, which makes our research questions and findings completely different. Köszegi and Rabin (2006) study a model of reference dependent preferences, where the reference point is a person’s rational expectations about outcomes. According to this theory, agents are influenced by a “gain-loss sense” that affects the maximum price they are willing to pay. For instance, if a consumer expects to buy a pair of shoes, she experiences a sense of loss if she does not buy them, and this sense of loss increases the maximum price she is willing to pay for the shoes. Daido and Itoh (2007) introduce these preferences in an agency model. They show that under risk aversion, the agent’s higher expectation allows the principal to implement greater effort with lower-powered incentives. Moreover, they obtain two types of self-fulfilling prophecy: the Galatea and the Pygmalion effect. In the former an agent’s self-expectation about his performance determines his actual performance, while in the latter the principal’s expectation about the agent’s performance has an impact on the agent’s performance. Although, as in our model, they study a principal agent model with agents’ reference dependent preferences, the focus of Daido and Itoh (2007) greatly differs from ours. Firstly, the results of a principal agent model with agents’ preferences à la Köszegi and Rabin (2006) can only vary from the standard model if there is common uncertainty about the production function (moral hazard) and not in an adverse selection setting like ours. And more importantly, in our model the agent’s reference point will be, in part, a decision variable of the principal (i.e., the goal) rather than the agent’s rational expectations. This allows us to incorporate goal setting as a part of the principal’s motivation policy. Finally, this paper is related to the models that account for the individuals’ intrinsic motivation to work. For instance, Bènabou and Tirole (2003) study a principal agent model in which the principal has better information than the agent about the agent’s type. The authors show that, although performance incentives lead to an increment of the agent’s effort in the short run, they are negative reinforcements in the long run. The idea is that if the principal pays a bonus to induce low ability agents to work (i.e., the principal increases the agent’s “extrinsic” motivation), then the agent perceives the bonus as a bad signal about his own ability (which reduces his “intrinsic” motivation). Some papers have also studied the optimal incentive contract when agents have intrinsic motivation. For instance, Fischer and Huddart (2008) study a model where the agents’ cost of effort is determined by a social norm; this social norm makes agents work harder in response to an increment in the average effort of their peers. These norms influence the power of financial incentives within an organization. In contrast with this literature, in our model the principal has a more active role since he can directly influence the agent’s intrinsic motivation by setting the reference point of his intrinsic utility. The paper proceeds as follows. Section 2 describes the basic model. In Section 3 we analyze the principal agent relationship by characterizing the optimal contract and studying the two type and the three type cases. Finally, Section 4 concludes.

Psychologists and experts in management have long documented the importance of goal setting in workers’ motivation. In particular, they have found that when workers are committed to challenging but attainable goals, their performance increases even if those goals are not directly linked to wages. In this paper, we have introduced goal setting in a principal agent model of managerial incentives. Agents care about goal setting because achieving those goals creates a sense of achievement and accomplishment that modifies their intrinsic motivation to work. We have shown that, in an optimal contract, more challenging objectives increase agents’ performance and that the goals set by the principal increase with the agent’s standard. Therefore, goals that are payoff irrelevant, since they do not directly affect agents’ extrinsic incentives, increase the principal’s profits. We have also shown that a mid-ranged standard gives the highest satisfaction to an agent and that a mid-ranged agent type could be the most satisfied among all the agent types. Therefore, being very demanding can be detrimental. There are some promising lines for future research. First of all, our goal commitment function is a very simple one; an agent is committed to a goal when it exceeds his personal standard sufficiently for him to consider the goal to be challenging. Psychologists have found that there are other determinants of goal commitment that should be studied in an economic model, such as the agents’ self-efficacy (i.e., ability confidence) and the agents’ participation in the goal setting processes (see Anderson et al., 2010; Bush, 1998). Moreover, if we consider a more realistic model with production uncertainty, an agent can be satisfied from trying hard even if the goal is not achieved. Similarly, the agent could be dissatisfied from not having tried hard enough because he suffers from self-control problems: being tempted to be lazy when deciding effort and suffering regret from not achieving a goal after production take place. Therefore, the principal could use assigned goals to attenuate workers’ self-control problems. This extension would link our model with the existing self-set goals models (Hsiaw, 2009, Kock and Nafziger, 2011a and Kock and Nafziger, 2011b) with present-biased preferences. A very interesting line of future research is to endogenize the personal standard parameter. There are several ways to do this. First, in a model with different abilities we can imagine that the agent’s standard is in part determined by his ability. Second, we can think that the personal standard is determined by the agent’s rational expectation about outcomes. This would provide a very good link between the present model with goal dependent preferences and the reference dependent utility from expectations literature (such as models with preferences à la Köszegi and Rabin (2006)). In fact goal setting can provide an additional explanation of the formation of reference states. For instance, with an experimental study Matthey (2010) finds evidence that apart from an individual’s own past, present and expected future outcomes and the outcomes of relevant others, reference states also depend on environmental factors that do not influence outcomes. Therefore, the kind of payoff irrelevant goals studied in this paper could affect reference states in a similar way, so an experimental study in the area is very appealing. Another topic would be to introduce competition in the model. If we consider that firms compete for workers, we should reconsider our result that very demanding (and hence productive) agents may be the least satisfied. With competition we have two opposite effects. On the one hand, as we have shown in this paper, very demanding workers may get lower satisfaction than lower types. But, on the other hand, firms compete for more demanding agents offering them more attractive offers (wages and goals), which has a positive effect on the satisfaction of very demanding agents. Finally, there is evidence that goal setting policies have more impact on agents’ performance as time goes by Ivancevich (1974) finds that in a manufacturing company a goal setting program significantly improves workers’ performance within 6 months after implementation. Therefore, it would be interesting to extend our model to allow for dynamic considerations. One possibility is to consider personal standards changing with past goals. Thus, after achieving a certain goal the agent’s personal standard increases so he will be more committed to higher goals in the future. This captures the intuitive idea of goal setting as a motivation tool to push the agent’s limits. When standards at any point depend on the goal-setting history, some interesting effects, like addition to goals, may arise.