New economics research paper sheds light on why people and companies keep repeating the same planning mistakes

New economics research paper sheds light on why people and companies keep repeating the same planning mistakes

New economics research paper sheds light on why people and companies keep repeating the same planning mistakes

A group of behavioural economists; Matej Lorko, Dr Lyla Zhang and Professor Maroš Servátka from the Department of Economics recently published a paper explaining why business projects often fail as a result of unrealistically short schedules.

Matej Lorko described his experience that had sparked the idea for this research agenda:

“A couple of years back, I was in charge of managing software development business projects for a large corporation. When planning one of them, a project team member estimated that he would rewrite a certain application module to a different programming language in two weeks. In reality, it took him almost two months. Later that year, we initiated a follow-up project, in which the same person needed to rewrite another module, of a similar size. His estimate was again two weeks. What was the actual delivery time? Unsurprisingly, it was two months,” Lorko said.

Psychologists and economists have established many theories of why project schedules are being misestimated. Mistakes are often attributed to optimism bias, unjustified skill signalling or using short deadlines as commitment devices. In other instances, project planners are believed to not estimate at all. Instead, they just strategically maximise the available timeframe by considering what is still acceptable to customers and project stakeholders.

Projects are usually novel endeavours, featuring a lot of uncertainty and ambiguity. Getting the schedule right for something that has never been done before is a challenging task. Nevertheless, one would expect that after a misestimation, companies will learn their lesson and plan better in the future, at least when it comes to similar projects. That is, however, not something we observe. Instead, a high failure rate is typical also for companies with extensive history of project management processes in place.

Why are project planners unable to learn from their experience and as a result, keep repeating the same estimating mistakes over and over again? The MGSM team tested a conjecture that the observed pattern could be explained by a phenomenon called anchoring. Think of an anchor as a seed, a certain number, coming from the management or project customer and representing an opinion about how much time should a particular project last. Such anchors often appear in a form of an initial wild guess (“How long? Three months, maybe?”), a suggestion (“Do you think two weeks are enough for you to get it done?”), customer expectations (“We would really like to introduce the product to the market before the summer season.”) or, perhaps a tentative deadline (“The CEO expressed the intention to finish the project by the end of the year.”). Are such anchors capable of influencing the actual project duration estimates?

“In order to answer this question, we conducted a controlled experiment. We invited participants and faced them with a simple task of comparing two-digit numbers, so that everyone had a clear idea of what to do. We took away their watches and mobile phones to prevent them from measuring time. First, we asked them to estimate how long it would take them to complete the task and then we measured their actual completion time, which was approximately 11 minutes on average. We financially incentivized their performance as well as estimation accuracy – a crucial element of experimental economics methodology. The sooner they finished the task and the more accurately they estimated its duration, the more money they earned,” Lorko said.

“Now here comes the hack. Before the estimation, we anchored the participants. We asked some of them whether they thought they would finish the task in three minutes and some of them whether they thought they would finish it in twenty minutes,” Professor Maroš Servátka explained.

The participants’ estimates were clearly biased by the anchors. Those anchored on three minutes provided very low estimates, which resulted in underestimation of the actual task duration. On the other hand, those anchored on 20 minutes, overestimated the time.

“Such result did not come as much of a surprise. Behavioural scientists have accumulated substantial evidence showing that anchors have powerful influence on subsequent judgments. Indeed, although the task was simple and straightforward, participants had never really done it before, so they could have interpreted the anchor as a well-meant advice,” added Dr Lyla Zhang.

However, there is more to the experiment. The participants continued with estimating and completing the same task for two additional times. In order to test the effect of experience itself, the research team did not provide the participants with any feedback on their performance at any time. Did the gained experience help participants to overcome the anchoring effect, estimate the duration accurately, and earn more money? Not at all. Although both (differently anchored) groups performed the identical tasks and it took them approximately the same amount of time to complete it, the estimates were vastly different, even in the third repetition. The effect of anchors persisted even against the rising experience. The participants were repeating the same estimating mistakes again and again. The seed that was once planted was never forgotten.

“The lesson in all this is that uninformed suggestions and expectations offered to project planners can be harmful. Does it mean that if a project manager is able to block them, the planners will estimate accurately, or at least become more accurate after gaining some experience? Not quite,” Servátka summarised.

“In the experiment, we also had a third group of participants, to which no anchor was presented. At first glance, it seemed that these participants were remarkably accurate in their estimates. The average estimation error for the third repetition of the task was only four seconds! However, a closer look at the individual estimates revealed a disturbing pattern. Although the average bias was negligible, the individual accuracy was not much better in this group in comparison with the anchored groups. In fact, we found that those who heavily underestimated in the first estimation, generally kept their estimates too low, while those who initially overestimated, kept their estimates too high, even though the actual task duration did not differ between the two groups at all. Effectively we found that if no anchor is given before the estimation, people tend to anchor themselves on their first estimate, and the gained experience with the task does not lead to better estimation accuracy,” Lorko explained.

“In order to investigate the robustness of the anchoring effect, we collected retrospective duration estimates after the participants completed the last repetition of the task. We asked our participants to estimate how long that last task actually took them to complete. In these retrospective estimates, we observed the same pattern as in the estimates produced before the task. Participants anchored (or self-anchored) on lower duration estimated that it actually took them significantly less time than it did in reality. We found the opposite pattern for participants anchored (or self-anchored) on longer duration. This evidence points out that anchors distort not only estimation before the task, but also retrospective duration estimation, which in turn can influence future estimation, creating a persistent anchoring effect,” Zhang added.

Conclusions and managerial implications

  • Project planners become anchored on uninformed suggestions and expectations.
  • As a result, their estimates are biased in the direction of the anchor.
  • If project planners do not receive detailed feedback regarding the actual duration of project tasks once the project is completed, they remain unaware of their misestimation and prone to repeating the same estimating mistakes again.
  • In the absence of feedback, own imprecise estimates can also be a source of persistent inaccuracy.

What can one do to help elicit more accurate estimates from project planners?

  • Provide feedback on the actual duration of projects, especially if the estimates were inaccurate.
  • Carefully consider past experiences. Simple descriptive statistics from similar past projects can complement the more traditional step-by-step planning based on the specification of the current project.
  • A measure as simple as the average duration of completed similar projects can outperform planners’ own estimates in terms of estimation accuracy, irrespectively whether in the presence or absence of anchors.
  • Given the power of anchors, managers could provide the average duration of a past project as a helpful anchor.

Matej Lorko, Maroš Servátka, and Le Zhang, ‘Anchoring in Project Duration Estimation’, Journal of Economic Behavior & Organization, 162, 2019, 49–65.

Matej Lorko:
Maroš Servátka:

Back to the top of this page