Production planning and control (PPC) requires human decision-making in several process steps like production program planning, production data management, and performance measurement. Thereby, human decisions are often biased leading to an aggravation of logistic performance. Exemplary, the lead time syndrome (LTS) shows this connection. While production planners aim to improve due date reliability by updating planned lead times, the result is actually a decreasing due date reliability. In current research in the field of production logistics, the impact of cognitive biases on the decision-making process in production planning and control remains at a silent place. We aim to close this research gap by combining a systematic literature review on behavioral operation management and cognitive biases with a case study from the steel industry to show the influence of cognitive biases on human decision-making in production planning and the impact on logistic performance. The result is the definition of guidelines considering human behavior for the design of decision support systems to improve logistic performance.
- cognitive biases
- human behavior
- production planning and control
- system design
While in the area of psychology, anthropology, and sociology human behavior has been investigated intensely and although behavioral aspects became an integral part of economic research , in the field of logistics and production planning and control (PPC), only little research has been conducted . In order to support decision-making processes in PPC and to optimize logistic performance, various models have been developed in order to reach short lead times, high due date reliability, low inventory levels, and high-capacity utilization as the key logistic performance indicators for production systems . However, the underlying proposition of these models is typically the theory of the “homo economicus.” In other words, to apply these models properly, we assume a fully rational human behavior in the decision-making process determined by the purpose of the decision-maker to maximize the personal advantage . Tversky and Kahneman  challenged this assumption and showed that human decisions are biased, which means a systematic deviation from rational judgment. In the fields of logistics and PPC, people are often confronted with uncertainty and high complexity, and research has shown that under these framework conditions, humans systematically take wrong decisions . One example for a complex situation in which biased decision-making leads to a deteriorating logistic performance is the so-called lead time syndrome (LTS). Here, production planners overreact to decreasing due date reliability. The planners adapt standard lead times too often, which eventually leads to an even worse aggravation of due date reliability . To support this variety of decisions, which have to be made in PPC, the so-called decision support systems (DSSs) are used frequently. DSSs are computer-based information systems with the purpose to improve the decision-making process and its outcome .
In this chapter we aim to improve the understanding of the role of cognitive biases in the field of PPC and propose first design guidelines for decision support systems (DSSs). Therefore, we combine a systematic literature review on behavioral operation management and cognitive biases. Taking inspiration from a case study from the steel industry, we show the possible impact of cognitive biases on human decision-making in PPC and on logistic performance. The remainder of this chapter is structured as follows. In Section 2, we outline the typical decision-making processes and the corresponding DSS in PPC. In Section 3, we use the case of the PPC at a steel manufacturer to present several examples of the possible impacts of cognitive biases on PPC decision-making. In Section 4, we give first recommendations on how to avoid the emergence of cognitive biases within PPC decision-making and derive first guidelines for DSS.
2. Decision-making and decision support systems in PPC
2.1 Decision-making in operation management
Already in the 1980s, the
According to  the DMP contains three phases. In the first “intelligence” phase, the problem which requires a solution by the decision-maker is identified and prioritized. Moreover, the first target achievements are defined, and corresponding data gathering is initialized. In the second “design” phase, a general action plan, which contains several action alternatives and their expected outcomes as well as the first evaluation criteria, is defined. In the third “choice” phase, the decision-maker selects the best action alternative based on the evaluation of each alternative. Based on the early works of , several models have been developed in order to explain the DMP. For instance, several authors suggest an extension of  DMP model [9, 10]. They propose a fourth “implementation” phase in which the decision outcome is turned into practice. In a fifth “learning” phase, lessons learned are formulated and shared in the organization to improve the DMP and the decision outcome in the future.
2.2 Decision-making in production planning and control
Moreover, also, within PPC human decisions undergo the suggested phases of the DMP models. Typical decision-making situations of PPC are shown in Figure 1.
PPC contains two subprocesses which are production planning and production control. These two subprocesses contain several task functions which require several decisions. Production planning focuses on the development of the basic concept to determine when to produce what in which quality. Production control has an overall monitoring function to achieve the production targets by the use of different control techniques.
2.3 Insufficient design guidelines for decision support systems
While there is a lot of research on DSSs in general (e.g., ), as well as on several components and modules of DSS (e.g., ), there is only little research about standardized design guidelines for DSS. For instance, [6, 9] criticize the lack of an integrated framework to support a standardized design of DSS. However,  propose a framework as the basis for standardized guidelines for DSS designers.
Figure 2 shows the framework suggested by . The framework contains four levels. (1) The first
3. Cognitive biases in production planning and control: the case of a German steel producer
3.1 Foundations of research on cognitive biases
Tversky and Kahneman  were the first who questioned the assumption of rational human behavior and introduced the term of cognitive biases. They state that humans taking decisions systematically go wrong, especially in complex and uncertain environments. In further experiments,  deepens this research of the underlying factors and describes the cognitive processes of intuition and reasoning.
Stanovich and West  named these cognitive processes System I (intuition) and System II (reasoning). While System I acts automatically, fast, emotively, and effortlessly and is hardly controllable, System II operates relatively slowly, reflected, and effortful . System I creates spontaneous impressions and persuasions, which form the basis for further decisions and actions of System II. Based on this two-system view,  claims that impressions are generated in System I and judgments are made in System II.
This fundamental research made clear that there are plenty of different cognitive biases that may affect human decision-making. Ref.  categorized these biases into six main categories:
Memory biases describe biases influencing the storage and the ability to remember information.
Statistical biases are the human tendency to over- or underestimate certain statistical parameter.
Confidence biases act to increase a person’s confidence in their prowess as a decision-maker.
Adjustment biases describe the human tendency to stick to the first available information or to a reference point when making decisions.
Presentation biases influence humans in their decision-making by the way how information is being displayed.
Situation biases describe the way how a person responds to the general decision situation.
3.2 Case study: decreasing due date reliability at a German steel producer
3.2.1 Initial situation
We take inspiration from a case study of the steel industry presented by . The analyzed PPC process takes place within a R&D department of a German steel case company. To compete in the global steel market, a short time to market is crucial. Especially in the R&D department, production and analysis processes are hardly to plan, and it is one of the major challenges of production planners to fulfill the customer requested delivery date. Samples of new alloys have to pass sequences of different tests before they can be launched in the market. In the analyzed R&D process, the first orders for different steel samples are placed through external and internal customers. After estimating the planned lead time for several manufacturing and analysis processes, the orders get a due date. For the scheduling of the production orders, a custom-developed DSS is used. In total, a production system with 20 machines and 35 employees in 1 shift was analyzed over a period of 3 years (from 2011 to 2014, 1.023 orders were analyzed). On-site visits, expert interviews, and observation documents were the used research methods. To evaluate the key performance indicator (KPI) development in terms of due date reliability, inventory, and lead times, feedback data from 13 months based on 240 shop floor calendar days were analyzed.
3.2.2 Observed behavior of key performance indicators
The due date reliability was one of the key performance indicators, and 95% was set as a long-term target for the planners. We observed the so-called lead time syndrome active in this context. When planners recognized decreasing delivery reliability, they started to update the initially planned lead times by releasing waiting orders earlier and adding some additional safety lead times in that cases in which the initially lead time was too short to meet the target due date. Thus, more orders are in the production system which causes an increasing WIP level and growing lead times. As a result, the delivery reliability was even lower than before the update of the lead times. The planners feel pressured to improve the current situation, and the circle of updating lead times reinforces, resulting in an even stronger due date aggravation.
Figure 3 shows the process of the observed planner’s behavior.
3.2.3 Observed behavior of planners: cognitive biases underlying deteriorating due date reliability
We observed several active cognitive biases in various decision-making situations in several PPC tasks. Nevertheless, it is important to understand that this classification of biases is not as concrete in practice as described in theory. Some of the cognitive biases overlap and often occur in several different situations.
126.96.36.199 Memory biases
Memory biases summarize a group of cognitive biases which are related to the storage and availability of information. The
We observe that planners tend to adjust planned lead times based on their intuition instead of entirely considering all influencing variables.
This occurs mainly in the phase of the
We observe the same development for the
188.8.131.52 Statistical biases
Statistical biases describe the tendency to over- or underestimate certain statistical parameters. Ref.  investigates that humans tend to overestimate the probability of two events occurring together if this has already happened once in the past. This effect is described by the
We observe statistical biases in the adaption of lead times within the
We observe these biases also in the
184.108.40.206 Confidence biases
Confidence biases describe the set of biases concerning the person’s confidence in their prowess as a decision-maker. The
Analyzing the case study, we find three examples of the
Confirmation biases were also observed in the phases of
220.127.116.11 Adjustment biases
Adjustment biases describe the human tendency to stick to the first available information or to a reference point when making decision. The
This became obvious in the phase of the
The same effect becomes obvious for the capacity planning. The planners justify their machine capacity planning with target figures of machine utilization rates of the previous years. These figures were not updated to the current situation.
18.104.22.168 Presentation biases
Presentation biases summarize a set of cognitive biases which influence humans in their decision-making by the way how information is being displayed. The
The implemented DSS offers plenty of types of analysis (such as the order forecasts, inventory levels, etc.) next to the information which is central for setting planned lead times. The
Further, we identified the influence of the ambiguity effect in the material quantity planning. When planners recognized that the production could run out of material, they just increased the initially ordered quantity. They did not further analyze potential causes like an increasing scarp rate due to an incorrectly set machine, etc. Instead they took the simplest option right in front of them to keep the production running even though the failure cause exponentiated.
22.214.171.124 Situation biases
Situation biases describe the way how a person responds to the general decision situation. The
We identify situation biases in several tasks in PPC. Modern PPC DSSs provide a wide range of information, such as key performance indicators concerning delivery reliability, inventory levels, or throughput times. For many planners, the amount and variety of information are too much to be included in their decision-making. In particular, under time pressure the planners decide to extend lead times just to do anything, even when they do not come to a reasonable conclusion when analyzing the data. At the same time, the planners ignore the fact that their own behavior of extending lead times influences due date reliability in a negative way. Moreover, we find that adjusting lead times is a common method of reacting to decreasing due date reliability. Planners who face the situation of decreasing due date reliability choose planned lead time extension just because their colleagues do so. Also, in the case of a machine breakdown, a similar behavior could be observed. The closer the delivery due date, the more planners decided to switch machines and change the production program sequence just to do anything. This was even true when the effort and time to change machines take in total longer than the repair of the initial machine.
Figure 4 shows a summary of our observed cognitive bias categories within the several PPC tasks. Potentially, there are even more active biases in the several PPC tasks, which we did not observe in our case.
4. Debiasing by the design of decision support systems
DSSs intend to improve the decision outcome by supporting the human decision-making process . Therefore, in the design of DSSs, also human behavioral aspects need to be considered to get an unbiased decision outcome. Based on our identified cognitive biases in PPC decisions, we aim to give first recommendations for system developers of DSS.
The proposed framework of  serves as the basis and is extended by a so-called behavioral layer. In this, already in the design phase, the DSS should foresee adequate
Debiasing is a method to reduce or eliminate the influence of cognitive biases within the decision process. Keren  proposed the following three steps for effective debiasing:
Identification of the existence and nature of the potential bias and the underlying influencing factors
Consideration of ways and techniques to lower the impact of bias
Monitoring and evaluation of the effectiveness of the selected debiasing technique
The proposed steps should be included in the design of a DSS. Based on our findings about the active cognitive biases in PPC, we already fulfilled the first step. In this section we contribute to the second step and aim to propose ways and techniques to lower the impact of biases. The third step then needs to be analyzed and observed over time.
These steps form the generic basis for a debiasing approach which contains further debiasing categories describing the concrete method of debiasing.
Kaufmann et al.  propose five categories for effective debiasing strategies in supplier selection processes:
Put yourself in the shoes of
Draw attention to alternative outcome
General bias awareness
We used these categories as a basis for the development of the first design guidelines for DSS in the field of PPC.
Within the DSS this method can be applied in two ways. (1) First, before making a final decision, potential scenarios can be presented to the decision-maker. This should also include the implications for logistic performance (e.g., inventory levels, lead time, due date reliability). For example, the implications of a change the production program may have for machine and personnel capacity as well as for material requirement planning should be made visible to the planner even if these implications are only of interest for other departments (e.g., the logistics and the purchasing department). (2) Second, historical data on the decision-making and the outcomes in terms of logistic performance can be provided (e.g., the decision about the lead time in the previous month caused this delivery reliability).
Behavioral aspects in operations management have been investigated for several years. We contribute to this research stream by analyzing the meaning of cognitive biases for decision-making in the field of PPC. The aim of this chapter is to determine first design guidelines for DSS considering the behavioral factors influencing human decision-making. The presented case study shows this need for industrial practice. Frameworks that aim to give advice to designers of DSS ignore the importance of the human factor in decision-making. We contributed to this research by extending the proposed design framework of  by adding the human behavioral layer. Moreover, we show first possible design techniques for DSS considering debiasing methods especially for decisions in PPC.