Politics and Public Policy

Public sector management: biases and agency problems

Leaders too often assume that their managerial strategies can be based on a rational analysis of data, trends and opportunities. Nonetheless, cognitive biases and agency problems have the power of shaping leaders’ decisions and employees’ actions alike. If an organizational strategy aspires to be truly rational, it should therefore account for the irrationality of individuals’ and groups’ behaviours, as dynamics playing out in the background are often the biggest driver of what gets decided and, ultimately, of what gets done. Understanding the behavioural aspects which underpin strategic management choices is particularly relevant in the public sector, because of the ultimate social value which public organizations must pursue. Indeed, the delivery of successful outcomes for society is essential, but it might be impeded by adverse agency problems which involve respectively civil servants and politicians, and public sector leaders and the general public. Still, the better the human contingencies are understood and anticipated, the best the strategy will be, for it will be able to overcome the power and the political games which are part of organizational life. 

Agency problems differ from cognitive biases for they are not entirely irrational and unintentional, but they instead originate from behaviours elicited by the incentives which individuals face. In the public sector, for instance, this might lead politicians to act in their own interest or in the interest of their restricted constituency, hence disregarding the needs of the broader general public, or civil servants to not collaborate with elected representatives. Whenever there is an agency problem, the decisions which individuals take are to the benefit of someone else with respect to whom they should be considering according to their formal mandate. Agency problems can arise intentionally or unintentionally, caused by the fact that individuals tend to be concerned about themselves and their inner circles in their judgement and actions. Indeed, achieving an individual objective is generally more persuasive than going the extra mile for others, and the implicit social contract of one’s unit is more compelling than the broader social contract of the organization. Let us now consider three of the most common behavioural issues which are recurrent in the public sector.  

Sunk-cost fallacy on investments 

The first issue to be considered is the sunk-cost fallacy, which is classifiable as partly a cognitive bias and partly an agency problem. In strategic terms, this fallacy refers to continuous investment of resources as a result of a foregone high past investment on a certain service or project. In this case, future courses of actions and business opportunities might be constrained because decision makers get stuck on areas of intervention which are burning money, but have a limited social benefit, only in consideration of an historical costs which is not recoverable. The sunk-cost fallacy is not only an accounting problem, but rather a reputational one, for decision makers instinctively aim at protecting their reputation by not admitting to themselves that the intended outcome will never be achieved. Being caught in a status quo bias, leaders end up pursuing what they unconsciously think will benefit them instead of the social objectives they are intended to. Indeed, whenever an unsuccessful service or business is not discontinued only because the administration has already committed to it by investing too much money, the community incurs in an extremely high opportunity cost.  

Gaming on performance targets 

A second problem which is particularly widespread in the public sector is that of gaming on performance targets. As managers who mandate objectives cannot directly observe the effort and the practices of civil servants and frontline workers, it might be that the establishment of performance indicators actually has an adverse effect on their actions in a number of ways. Although performance measurement systems are designed to positively influence behaviour, when goal displacement occurs it adversely affects performance. Hence, service delivery might be manipulated to ensure the performance targets are achieved to the detriment of actual intended outcomes. Typical gaming problems include: 

  • Output distortion: attempts to achieve targets come at the expense of significant (but unmeasured or unmeasurable) aspects of performance, as once the performance targets are announced individuals focus only on what will be measured. 
  • Ratchet effect: as there is an understanding that future targets are based on the past levels of performance, individuals have an incentive not to exceed targets (even if they could do so) in order to be mandated lower objectives in the following period.  
  • Threshold effect: although targets increase the pressure for underachievers to perform better, in the long term they also induce overachievers to allow their performance to deteriorate to the standard if they are not rewarded for exceeding the threshold. 

In order to avoid goal displacement the organization should carefully define indicators which are able to produce balanced incentives that channel performance towards the socially desired outcomes. If the possibility of gaming is anticipated, either the performance management system should be redesigned from scratch, or other appropriate indicators to counteract the adverse effects should be added. For instance, a good example of goal misplacement taken from the healthcare sector is that induced by the measurement of hospitalization times, for it might have the adverse effect of giving hospitals an incentive to discharge patients too early in the pursue of increased time efficiency. To mitigate this risk of output distortion, the indicator of hospitalization time can be compensated by an indicator measuring the number of re-admissions to hospital care within one month from the discharge, picking up whether insufficient care was initially given to the patient.  

Misaligned risk aversion profiles  

Because of misaligned risk profiles, individuals often have a different attitude to risk than what would be optimal for the scope and ambition of their overall organization. For instance, to avoid any potential blame for failure, public managers might agree only to the plans upon which they are certain they can deliver good results. Although this might be keeping their reputation safe, most often it results in a lack of entrepreneurial spirit and in the stillness of public agencies and organizations. If managers feel their reputation is on the line, they might opt for safe choices which are not necessarily the optimal ones. For instance, they might indict unrequired tenders even for very small contract which would not need them, hence exponentially increasing the bureaucracy surrounding service delivery only to avoid being pointed out as not sufficiently transparent. This behaviour of deliberate underperformance in the delivery of outcomes is also referred to as ‘sandbagging’, the rationale being that, if one does nothing, no one will be able to attribute political blame for their possible errors. On the other hand, acting in the name of public interest comes with a high cost, which is that of possibly making mistakes.  


Arkes, R. and Blumer, C. (1985), The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1): 124-140. 

Banks, J., & Weingast, B. (1992). The political control of bureaucracies under asymmetric information.American Journal of Political Science, 36(2), 509-524.  

Bevan, G. and Hood, C. (2006), What’s measured is what matters: Targets and gaming in the English public health care system. Public Administration, 84: 517-538. 

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s