On the surface, organizations can seem pretty straightforward from a structural standpoint – once you see a mission statement, an organizational chart and maybe some accounting documents, you might feel like you have a clear picture of how an organization operates. If you look more closely, however, most organizations resemble a complex ecosystem that can’t be easily described or understood via some boxes and lines on a sheet of paper.
In fact, the biological comparison is so apt that fields such as ‘organizational ecology’ have become increasingly popular within academia. In these departments, researchers develop models that largely imitate those used to describe evolutionary processes within biological ecosystems. In organizations, the complexity that drives such models comes largely from two sources. First, the degree of informal and often tacit knowledge held within organizations can lead to habits that make their normal day-to-day operations indecipherable to outside observers. Second, the fact that organizations are made up of people with unique preferences, incentives and agency to act on such motivations means that most business decisions are actually the result of a complex web of interconnected and often individual choices.
While ‘complexity’ as a concept generally has a negative connotation (perhaps except when describing food), there is at least one clear strategic upside to organizational complexity in that, just as it’s overwhelmingly difficult to reverse engineer a sustainable coral reef, it’s quite difficult to replicate most sets of organizational structures and processes. In a competitive sense, organizational complexity implies that, despite managers’ fears that their business models will be immediately copied by new entrants into the market, such systems are actually extremely difficult to emulate, even with visibility into the inner workings of the organization to be copied.
In one well-known historical example, Toyota formed a joint venture with General Motors in the 1980s as an initial foray into building vehicles in the United States 1. At the time, U.S. automakers were very interested in the manufacturing processes of their Japanese counterparts, since they appeared to be making better quality vehicles at lower cost. This partnership, called New United Motor Manufacturing Incorporated (NUMMI), provided both companies with visibility into (and even some training on) the core competencies of the other – namely, the inner workings of U.S. supply chain practices and union relations as well as Japan’s ‘Total Quality Management’ (TQM) and other innovative manufacturing processes.
Counterintuitively, General Motors seemed unable to incorporate the processes it had learned at the NUMMI plant into its overall operations, in part because many of the nuances that arise in complex systems fail to replicate automatically due to differences in underlying conditions and environments. To further describe the implications of this complexity, Harvard Business School professor Jan Rivkin devised a model that suggests that even basic business strategies are complex enough that attempts to imitate them via observing and copying (as in the previous example) constitute a mathematically hard problem and are thus a suboptimal approach to developing a competitive business 2.
While the effects of complexity on interactions between firms are certainly important, it’s worth paying particular attention to why and how the concept of complexity should be considered at various points within an organization as well. Specifically, it is of crucial importance to recognize that much of the backbone of organizational processes consists of a set of game-theoretic strategic interactions, each of which is not nearly as independent or straightforward as it may initially seem.
Game Theory and the Organization
While it is becoming increasingly popular to discuss game theory in the context of organizations and workplace interactions, the notion of ‘playing games’ with employees can still seem a bit manipulative or otherwise off-putting. As such, it’s important to keep in mind that a ‘game’ in a game theory sense refers to any of a wide variety of strategic interactions, or situations where one’s result is not only dependent on the action he takes but also on the actions taken by others. In one simple example of a workplace game, each worker has a choice of whether to work hard or surf the internet, and the payoff to each worker depends not only on whether he works hard but also on whether the other members of his team do so.
In this sense, managers play games with their employees all the time, since organizational payoffs to managers generally depend on both their choices and the actions of their employees, and the same can be said for the payoffs to employees (but in reverse). In fact, managers partake in a game every time they set up an incentive system with their employees! In these situations, the manager’s action is the choice of the activity to incentivize and the structure and amount of the incentive. The employee’s action is the choice of how much effort to put into the activity being incentivized. Granted, incentive games can still feel manipulative to some, but it’s worth noting that they are generally preferable to alternative arrangements where managers take away workers’ agency and rely on direct control instead.
Sales commissions are, at least on the surface, a very simple example of an incentive system. In such a system, the manager decides what type of sale to incentivize and how much to pay the employee for each sale (most often a percentage of the amount of the sale). Upon seeing the rules of the system, the employee decides how much effort to exert and what activities to undertake in order to optimize her payoff, taking into consideration the amount she receives in incentive pay as well as the non-monetary psychological and/or physical cost of her effort. This scenario represents one type of shift from focusing on hours of input to focusing on the output created, and it has a number of benefits both for the organization and for the worker.
First, output-based incentive systems reduce the need for costly managerial monitoring, since it takes far less effort to verify a sales figure than to constantly drop in on an employee to ensure he is doing, well, whatever it takes to effectively make a sale. As a related matter, this focus on output rather than input empowers the employee to take ownership and leverage his expertise. As a result, this shift of agency also lessens the burden on the manager to know how best to address the complexities and challenges inherent in the jobs of all of her reports. The payment for output rather than input, even if the sales commission is in addition to a guaranteed salary, also shifts some of the financial risk away from the organization since the incentive creates a situation where revenue and labor expense move together.
The Role of Complexity in Game Theory
Even the basic example of a sales commission becomes far more complex when placed in the overall context of organizational dynamics. In practice, the incentive game described above doesn’t just happen once but is instead repeated periodically between the same manager and employees. As a result, an employee is likely not only thinking about what action is best today but also about how that response will affect the incentives that are offered in the future.
For example, it’s not uncommon for managers to ‘raise the bar’ over time if employees appear to be easily meeting their incentive targets, effectively resulting in less compensation for the same work as before. If employees anticipate this behavior they will likely respond by holding back their productivity to some degree, since the short-term payoff could be outweighed by the longer-term incentive reduction. Furthermore, such behavior from management can breed distrust and have negative impacts on employee wellbeing. Not surprisingly, sociologists have even documented a form of social pressure amongst employees to not work too hard for fear that such effort will lead to future incentive reductions for all workers, not just the hard worker 3. Given these dynamics, managers would be wise to look for ways to credibly commit to an incentive structure for a reasonably long period of time.
More generally, games within organizations tend to be complex because all of the players involved are human beings, not computers. One implication of human nature is that managers are typically too focused on what they want to happen in response to an incentive and less cognizant of how their employees will respond to the incentive game in practice. With regard to employees, human nature can make responses to incentives nuanced and even unpredictable and the resulting games more complex. This human element of games often leads to outcomes that are suboptimal or at least odd-seeming from a managerial perspective.
Sales quotas illustrate this idea clearly. Quotas are a common form of incentive systems, again likely due to their apparent simplicity – after all, if you have a sales target or other goal that you want to achieve, why not just offer workers a single payment for reaching the goal? First off, you’re probably missing the bigger picture – from the organization’s perspective, the quota amount is probably not a target so much as a minimum acceptable standard, and it’s generally the case that more output is better and less output is worse, despite what the binary outcome of ‘meeting the target’ would imply. Nonetheless, employees are likely to respond to the incentive as it’s presented to them, not as the manager wants them to think about it. In the case of the sales quota, employees will likely decrease productivity once they approach the quota amount, both because they don’t get payment for sales above the quota amount and because they don’t want to encourage a higher quota in the future. In some cases, workers have even been shown to try to hold back sales until the next fiscal period once their quota is met and try to pull forward next period’s sales if it pushes them over the quota line 4. At the opposite extreme, workers are likely to give up (and do the minimum to not get fired) if they determine that they will not be able to meet the quota for the current period.
To some degree, these worker behaviors could be helpful to organizations that have an explicit goal of showing consistent results from one period to the next. Even in these cases, however, the complexity of the sales process leads to some likely undesirable consequences, most notably that workers often lower prices in order to time purchases in a payoff-maximizing manner, which can decrease revenue for the organization. Such discounts also introduce a layer of chaos into the organization in that they are a function of each employee’s position vis-à-vis their target and not based on the overall business position that the organization is in.
This underappreciated complexity of even simple incentive systems means that managers have a responsibility to their employees, and by extension the entire organization, to think carefully about the incentive games they are designing in order to enable workers to flourish rather than frustrate and demotivate them. Once the underlying complexity is properly acknowledged, managers can begin to manage the complexity by putting themselves (hypothetically) in the employee’s, supplier’s or customer’s position to better think through how each counterparty will respond to the game under consideration.
The preceding discussion highlights two principles that are crucial for managing complexity in organizational games. First, acknowledge and respect the complexity rather than being quick to oversimplify, and, second, learn to properly anticipate the likely responses of your counterparties via an understanding of various facets of human psychology. Let’s look at each of these principles in more detail.
Managing Complexity in Organizational Games
In 1975, British economist Charles Goodhart noted that ‘any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.’ About 20 years later, anthropologist Marilyn Strathern codified what became known as Goodhart’s law: ‘when a measure becomes a target, it ceases to be a good measure’ 5. Similarly, economist Robert Lucas argued in what became the Lucas critique that it is ill-advised to use historical observational data as a proxy for what to expect if one attempts to exploit the historical relationship. For example, a manager might observe that a worker is consistently more productive at 8:00 PM than at 11:00 AM, but this doesn’t mean that the manager can elicit a productivity increase by requiring the worker to always work at 8:00 PM.
At their core, the statements above provide an implicit lesson about the need to respect and acknowledge complexity, essentially asserting that we often don’t understand enough about organizational processes in order to say confidently what will happen to output when we purposely modify an input quantity. On top of this, it is also often the case that the mere act of observation or measurement changes employees’ behavior. The psychology world has documented this phenomenon, referred to as the Hawthorne effect, and it makes objective measurement of output and productivity less than straightforward. The difficulty in fully understanding complex systems doesn’t mean that managers should give up, however, but instead that they would be well-served by testing and gathering feedback before committing to changes in various organizational processes and systems. Similarly, employees are better off when managers are thoughtful in this regard rather than imposing potentially unproductive changes.
You might be thinking ‘I hear what you’re saying, but using this measure as a target doesn’t seem like a horrible idea.’ And you could be right in some cases! That said, complex organizations are generally interested in maximizing (or minimizing) multiple quantities of interest as opposed to having a single objective, and it’s important to acknowledge that focusing on those quantities that are easily measurable generally happens at the expense of those that are not, leading to distortions in effort and output across the organization.
To see why this is the case, picture yourself as a manager who has just found a new tool for analysing the number of customer emails answered by your employees. You tell your employees about this tool and say that you’re going to look at the reports every week, and soon you see a sharp increase in the number of customer service emails answered. Your initial thought is that that is a bit odd since nothing really changed, but you continue analysing the reports because encouraging more customer service emails to be answered is good, right? Unfortunately, what you’re not seeing in these reports is that the number of emails increased because employees started responding to some customer inquiries by email rather than by phone in order to make the reported numbers look better. In fact, some departments even put policies in place requiring workers to respond via email even if the customer had requested a phone interaction! This is likely not optimal for the organization but is hardly an unforeseeable consequence of the game that was implemented once the manager placed focus on a particular measure.
Overall, most managerial guidance is geared toward focusing incentives and monitoring on outputs versus inputs for the reasons discussed earlier, i.e. lower monitoring costs, worker empowerment and risk shifting. These reasons are quite compelling, but managers must be careful to consider how human nature will likely drive the response to any given incentive system. In addition to the warnings provided by earlier scenarios, it’s worth noting that high-powered, output-based incentives often create high employee stress levels as well as a temptation to cheat and engage in illicit behaviors, which in turn creates a new kind of monitoring problem. Furthermore, it’s often the case that output-based incentives affect the pool of workers that an organization is able to hire from. For example, women have historically been more likely to stay away from endeavors that involve competition-based incentives, at least in experimental settings 6. Given these tradeoffs, it is worth considering whether there are cases where incentivizing inputs (i.e. participation, effort, etc.) rather than outputs would make sense for an organization and its workers, and it turns out that there are at least two general scenarios where human behavior implies that focusing on inputs could be preferable.
Recall that part of the justification for focusing measurement and incentives on outputs rather than inputs is that those who are closer to the work being done are likely in a better position to understand the specific relationship between various forms of effort and output. But what if this is not the case? In such scenarios, employers could offer output-based incentives that generate poor results and unhappy employees largely because, while the incentives may induce effort, that effort is likely to be unproductive without proper guidance as to how that effort should be directed. If workers are self-aware regarding their lack of expertise, it could even be the case that the output-based incentives don’t even induce effort, since workers would know that their effort is likely to be useless. In these cases, specifically incentivizing inputs could help overcome this knowledge problem and the resulting loss of flexibility could be worthwhile to the organization.
This logic is illustrated by an interesting series of experiments conducted in the New York City public school system 7. In one experiment, researchers took the output-based approach of offering students financial incentives in return for good grades, with the hypothesis that this system would induce effort while enabling students to utilize whatever learning methods and tools worked best for them. In another experiment, researchers paid students directly for the inputs to the ‘educational production function’ – attendance, complete homework and so on. Somewhat counterintuitively, the latter experiment produced bigger increases in student achievement, and one explanation is that many students needed to be taught how to learn and told what activities are most beneficial in order to improve their performance. It is worth noting, however, that the input-based approach seemed to be more effective for lower-performing students, and, similarly, a realistic assessment of the workers in an organization will likely give useful guidance as to whether an input or output based incentive approach is more appropriate.
Even if workers have full information regarding how, at least theoretically, to efficiently transform their effort into output, it may be the case that this input-output relationship has so much uncertainty that risk-averse employees aren’t motivated by output-based incentive systems. As an extreme example, consider a scenario where a researcher is offered a substantial bonus for each form of cancer that she develops a cure for. The researcher is likely fully aware of the steps that she has to take in order to best move toward this goal, but it is far from clear whether any of her effort will ultimately be successful. If putting in effort is akin to buying a lottery ticket, then why bother? (other than the curing cancer part of course). In such a case, it likely makes sense, if incentive systems are going to be used at all, to focus on research inputs rather than outcomes. Of course, this transfers a degree of financial risk from the researcher to the company, since the company could end up making incentive payments even in the absence of profit-producing output, but this is likely appropriate because companies tend to have multiple projects under their purview and thus have better diversification across uncertain outcomes. Note, however, that effectively incentivizing inputs entails focusing on various (hopefully) productive actions and doesn’t simply fall back to the old standard of using hours worked as the main unit of input or productivity.
To complicate incentive games even further, there may even be situations within an organization where the best approach is to not play the game at all. Author Dan Pink discusses this concept at length in his book Drive: The Surprising Truth About What Motivates Us, and a few points are worth mentioning briefly here 8. First, and most shocking, is the observation that extrinsic rewards – i.e. incentives or game payoffs – tend to psychologically crowd out the internal motivation to do good work that most humans (and monkeys, according to Pink) appear to possess. Economists Uri Gneezy and Aldo Rustichini, in a paper appropriately titled Pay Enough – Or Don’t Pay At All, outline a number of scenarios where this principle comes into play, including an observation that small incentive payments are associated with lower performance on an effort-based analytical test 9. Interestingly, however, they also find that the average performance decline is accounted for by a small subset of people who appear to be very negatively affected by the monetary payments, which highlights the importance as a manager of understanding your particular pool of workers before trying to design incentive systems for them.
This concept is obvious to many of us who have tried to turn a pleasant hobby into a source of employment only to find that working for monetary compensation somehow makes the hobby less pleasurable, but it applies fairly broadly in the workplace as well. Of course, it’s not really possible in a workplace environment to not provide some form of monetary compensation, but paying workers a fair salary and avoiding the incentives game altogether can be an effective strategy in scenarios where workers have high levels of ownership over their work and a personal drive for success that is well-aligned with the objectives of the organization. Most academic environments are likely good candidates for this approach, which could explain why we see tenure contracts rather than compensation for, say, each research paper published. (To be fair, the uncertainty in the publication process could also play a role here!)
The second factor to consider regarding whether to incentivize is the significant body of evidence that suggests that specific monetary payments in return for output requiring creative or open-ended thinking can have negative effects on productivity, even above and beyond the effect on intrinsic motivation 10. The underlying explanation is that output-based incentives tend to focus one’s thinking, which is helpful for directed tasks that largely require effort rather than ingenuity, but this focus is not helpful for various types of thought-intensive work. Therefore, it is crucial for managers not only to know the nature of their particular groups of workers (keeping in mind that this population will likely change over time) but also to properly characterize the potentially wide variety of tasks that their employees perform on a day-to-day basis and plan their incentive systems accordingly.
There is definitely a place in organizations for game-theoretic thinking, particularly as it relates to incentive ‘games’ played between managers and their employees. In general, the guidance regarding such incentive systems favors focusing on outputs rather than inputs, but various caveats regarding complexity in such systems should be addressed carefully to avoid unexpected organizational outcomes. In addition, there are a number of cases where organizational dynamics and human nature create a world where either input-based incentives or no incentives at all will likely lead to better outcomes than output-based incentives, but it’s hard to envision a scenario where the traditional focus on hours worked is itself optimal. The effective manager will weigh the considerations mentioned here in order to design a system that works for their particular, likely complex, set of employees and tasks. If this is done well, the systems will not only benefit the organization but will enhance workers’ sense of well-being in addition to their productivity, both because workers like to feel productive and because the incentive games in play are working with human nature rather than against it.
Jodi N. Beggs
Jodi is a behavioral economist who specializes in how human psychology affects organizational and market dynamics. She is the founder of Economists Do It With Models, a company that focuses on producing educational content for use in and out of the classroom. In addition, Jodi works as an economist in the tech sector, writes for various publications, and is a competitive figure skater.
Share this article and let us know what you think. We're here to help and answer any questions you might have. Any suggestions or feedback? Let us know what you think and we will use your input for the future improvements.
The return to journalism, the pursuit of truth and the utmost respect for solid, peer-reviewed science. You're just one click away from receiving the best of The Habtic Standard straight to your inbox. Subscribe to our monthly newsletter now and keep up to date with the latest corporate wellbeing insights from our experts around the globe.