Skip Menu

Return to Skip Menu

Main Content


System Dynamics Lab (SDL)


Analytical Methods for Dynamic Modelers. MIT Press. Forthcoming (2015).

Selected papers: 

Health and safety
Science policy 
Project dynamics 
Organizational capabilities and learning 
Modeling and estimation methods

          Last update: Dec 23, 2014

Health and safety: 

  • Hosseini N., Rahmandad H., Wittenborn A., Modeling the Hypothalamus-Pituitary-Adrenal Axis: a Review and Extension, (working paper).

Abstract: Multiple models of the hypothalamus-pituitary-adrenal (HPA) axis have been developed to characterize the oscillations seen in the hormone concentrations and to examine HPA axis dysfunction. We reviewed the existing models, replicated, and compared them by finding their correspondence to a dataset consisting of ACTH and cortisol concentrations of 18 individuals. We found that existing models use different feedback mechanisms, vary in the level of details and complexities, and sometimes offer inconsistent conclusions, while none provides a great fit to validation dataset. We therefore re-calibrated the best performing model using partial calibration and individual fixed effects. Our estimated parameters reduced the mean absolute percent error significantly and offers a validated reference model for diverse applications. Our analysis also suggests that circadian and ultradian cycles are not created endogenously by the HPA axis feedbacks.


  • Jalali M.S., Rahmandad H., Bullock S.L., Lee-Kwan S.H., Ammerman A., Dynamics of Obesity Prevention Interventions inside Organizations, (working paper).

Background: A large number of obesity prevention interventions, from upstream (policy and environmental) to downstream (individual level), have been put forward to curb the obesity trend; however, not all those interventions have been successful. Overall effectiveness of obesity prevention interventions relies not only on the average efficacy of a generic intervention, but also on the successful Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) of that intervention. In this study, we aim to understand how effectiveness of organizational level obesity prevention interventions depends on dynamics of AIM.

Methods: We focus on an obesity prevention intervention, implemented in food carry-outs in low-income urban areas of Baltimore city, which aims to improve dietary behavior for adults through better food access to healthier foods and point-of-purchase prompts. Building on data from interviews and the literature we develop a dynamic model of the key processes of AIM.

Results: We first develop a contextualized map of causal relationships integral to the dynamics of AIM, and then quantify those mechanisms using a system dynamics simulation model. With simulation analysis, we show how as a result of several reinforcing loops that span stakeholder motivation, communications, and implementation quality and costs, small changes in the process of AIM can make a big difference in impact.

Conclusions: We present how the dynamics surrounding communication, motivation, and depreciation of interventions can create tipping dynamics in AIM. Specifically, small changes in allocation of resources to an intervention could have a disproportionate long-term impact if those additional resources can turn stakeholders into allies of the intervention, reducing the depreciation rates and enhancing sustainability. We provide researchers with a set of recommendations to increase the sustainability of the interventions.


  • Jalali M.S., Sharafi Z., Rahmandad R., Ammerman A., Parental social influence in childhood obesity interventions: a systematic review, (working paper).

Abstract: The objective of this study is to understand the pathways through which social influence at the family level moderates childhood obesity interventions. We conducted a systematic review of obesity interventions in which parents’ weight-related behaviors are directly targeted by the intervention due to their potential social and environmental influence on the nutrition and physical activity behaviors of their children. Results for existing mechanisms that moderate parents’ influence on children’s behavior are discussed and a causal pathway diagram is developed to map out social influence mechanisms that affect childhood obesity. We provide health professionals and researchers with a set of recommendations to leverage family-based social influence mechanisms for increasing the efficacy of the obesity intervention programs.


Abstract: Quantifying human weight and height dynamics due to growth, aging, and energy balance can inform clinical practice and policy analysis. This paper presents the first mechanism-based model spanning full individual life and capturing changes in body weight, composition and height. Integrating previous empirical and modeling findings and validated against several additional empirical studies, the model replicates key trends in human growth including A) Changes in energy requirements from birth to old ages. B) Short and long-term dynamics of body weight and composition. C) Stunted growth with chronic malnutrition and potential for catch up growth. From obesity policy analysis to treating malnutrition and tracking growth trajectories, the model can address diverse policy questions. For example I find that even without further rise in obesity, the gap between healthy and actual Body Mass Indexes (BMIs) has embedded, for different population groups, a surplus of 14%–24% in energy intake which will be a source of significant inertia in obesity trends. In another analysis, energy deficit percentage needed to reduce BMI by one unit is found to be relatively constant across ages. Accompanying documented and freely available simulation model facilitates diverse applications customized to different sub-populations.


Objectives. We present a system dynamics model that quantifies the energy imbalance gap responsible for the US adult obesity epidemic among gender and racial subpopulations. 

Methods. We divided the adult population into gender–race/ethnicity subpopulations and body mass index (BMI) classes. We defined transition rates between classes as a function of metabolic dynamics of individuals within each class. We estimated energy intake in each BMI class within the past 4 decades as a multiplication of the equilibrium energy intake of individuals in that class. Through calibration, we estimated the energy gap multiplier for each gender–race–BMI group by matching simulated BMI distributions for each subpopulation against national data with maximum likelihood estimation. 

Results. No subpopulation showed a negative or zero energy gap, suggesting that the obesity epidemic continues to worsen, albeit at a slower rate. In the past decade the epidemic has slowed for non-Hispanic Whites, is starting to slow for non-Hispanic Blacks, but continues to accelerate among Mexican Americans. 

Conclusions. The differential energy balance gap across subpopulations and over time suggests that interventions should be tailored to subpopulations’ needs. 


Abstract: Obesity is associated with a prolonged imbalance between energy intake and expenditure, both of which are regulated by multiple feedback processes within and across individuals. These processes constitute 3 hierarchical control systems-homeostatic, hedonic, and cognitive-with extensive interaction among them. Understanding complex eating behavior requires consideration of all 3 systems and their interactions. Existing models of these processes are widely scattered, with relatively few attempts to integrate across mechanisms. We briefly review available empirical evidence and dynamic models, discussing challenges and potential for better integration. We conclude that developing richer models of dynamic interplay among systems should be a priority in the future study of obesity and that systems science modeling offers the potential to aid in this goal.


Objectives: To simulate physician-driven dynamics of delivery mode decisions (scheduled cesarean delivery [CD] vs. vaginal delivery [VD] vs. unplanned CD after labor), and to evaluate a behavioral theory of how experiential learning leads to emerging bias toward more CD and practice variation across obstetricians. 

Data Sources/Study Setting: Hospital discharge data on deliveries performed by 300 randomly selected obstetricians in Florida who finished obstetrics residency and started practice after 1991. 

Study Design: We develop a system dynamics simulation model of obstetricians' delivery mode decision based on the literature of experiential learning. We calibrate the model and investigate the extent to which the model replicates the data. 

Principal Findings: Our learning-based simulation model replicates the empirical data, showing that physicians are more likely to schedule CD as they practice longer. Variation in CD rates is related to the way that physicians learn from outcomes of past decisions and accumulate experience. 

Conclusions: The repetitive nature of medical decision making, learning from past practice, and accumulating experience can account for increases in CD decisions and practice variation across physicians. Policies aimed at improving medical decision making should account for providers' feedback-based learning mechanisms.


Abstract: Researchers use system dynamics models to capture the mean behavior of groups of indistinguishable population elements (e.g. people) aggregated in stock variables. Yet many modeling problems require capturing the heterogeneity across elements with respect to some attribute(s) (e.g. body weight). This paper presents a new method to connect the micro-level dynamics associated with elements in a population with the macro-level population distribution along an attribute of interest without the need to explicitly model every element. We apply the proposed method to model the distribution of body mass index and its changes over time in a sample population of American women obtained from the U.S. National Health and Nutrition Examination Survey. Comparing the results with those obtained from an individual-based model that captures the same phenomena shows that our proposed method delivers accurate results with less computation than the individual-based model. 


Abstract: Basal metabolic rate (BMR) represents the largest component of total energy expenditure and is a major contributor to energy balance. Therefore, accurately estimating BMR is critical for developing rigorous obesity prevention and control strategies. Over the past several decades, numerous BMR formulas have been developed targeted to different population groups. A comprehensive literature search revealed 248 BMR estimation equations developed using diverse ranges of age, gender, race, fat-free mass, fat mass, height, waist-to-hip ratio, body mass index and weight. A subset of 47 studies included enough detail to allow for development of meta-regression equations. Utilizing these studies, meta-equations were developed targeted to 20 specific population groups. This review provides a comprehensive summary of available BMR equations and an estimate of their accuracy. An accompanying online BMR prediction tool (available at was developed to automatically estimate BMR based on the most appropriate equation after user-entry of individual age, race, gender and weight.

Keywords: Basal metabolic rate; resting metabolic rate; prediction; meta-analysis; review; meta-regression 


Abstract: Although systems science has emerged as a set of innovative approaches to study complex phenomena, many topically focused researchers including clinicians and scientists working in public health are somewhat befuddled by this methodology that at times appears to be radically different from analytic methods, such as statistical modeling, to which the researchers are accustomed. There also appears to be conflicts between complex systems approaches and traditional statistical methodologies, both in terms of their underlying strategies and the languages they use. We argue that the conflicts are resolvable, and the sooner the better for the field. In this article, we show how statistical and systems science approaches can be reconciled, and how together they can advance solutions to complex problems. We do this by comparing the methods within a theoretical framework based on the work of population biologist Richard Levins. We present different types of models as representing different tradeoffs among the four desiderata of generality, realism, fit, and precision.

Keywords: Agent-based model, childhood obesity, complex systems, computational model, Levins framework, social network analysis, statistical model, system dynamics model.


Abstract: Falls are a significant public health risk and a leading cause of non‐fatal and fatal injuries among construction workers worldwide. A more comprehensive understanding of casual factors leading to fall incidents is essential to prevent falls in the construction industry. However, an extensive overview of causal factors is missing from the literature. In this paper, 536 articles on factors contributing to the risk of falls were retrieved. One hundred and twenty‐one (121) studies met the criteria for relevance and quality to be coded, and were synthesized to provide an overview. In lieu of the homogeneity needed across studies to conduct a structured meta‐analysis, a literature synthesis method based on macro‐variables was advanced. This method provides a flexible approach to aggregating previous findings and assessing agreement across those studies. Factors commonly associated with falls included working surfaces and platforms, workers' safety behaviours and attitudes, and construction structure and facilities. Significant differences across qualitative and quantitative studies were found in terms of focus, and areas with limited agreement in previous research were identified. The findings contribute to research on the causes of falls in construction, developing engineering controls, informing policy and intervention design to reduce the risk of falls, and improving research synthesis methods.

Keywords: Accident causes, causal map, literature synthesis methods, safety.


Abstract: We developed an individual-based (IB) model to explore the stochastic attributes of state transitions, the heterogeneity of the individual interactions, and the impact of different network structure choices on the poliovirus transmission process in the context of understanding the dynamics of outbreaks. We used a previously published differential equation-based model to develop the IB model and inputs. To explore the impact of different types of networks, we implemented a total of 26 variations of six different network structures in the IB model. We found that the choice of network structure plays a critical role in the model estimates of cases and the dynamics of outbreaks. This study provides insights about the potential use of an IB model to support policy analyses related to managing the risks of polioviruses and shows the importance of assumptions about network structure.

KeywordsDisease transmission, individual-based model, outbreak response, poliovirus.


Science policy:

Abstract: The academic job market has become increasingly competitive for PhD graduates. In this note, we ask the basic question of ‘Are we producing more PhDs than needed?’ We take a systems approach and offer a ‘birth rate’ perspective: professors graduate PhDs who later become professors themselves, an analogue to how a population grows. We show that the reproduction rate in academia is very high. For example, in engineering, a professor in the US graduates 7.8 new PhDs during his/her whole career on average, and only one of these graduates can replace the professor's position. This implies that in a steady state, only 12.8% of PhD graduates can attain academic positions in the USA. The key insight is that the system in many places is saturated, far beyond capacity to absorb new PhDs in academia at the rates that they are being produced. Based on the analysis, we discuss policy implications.

Keywords: Higher education policy;unemployment;R0;engineering workforce development;research workforce development.  


Abstract: The US government has been increasingly supporting postdoctoral training in biomedical sciences to develop the domestic research workforce. However, current trends suggest that mostly international researchers benefit from the funding, many of whom might leave the USA after training. In this paper, we describe a model used to analyse the flow of national versus international researchers into and out of postdoctoral training. We calibrate our model in the case of the USA and successfully replicate the data. We use the model to conduct simulation-based analyses of effects of different policies on the diversity of postdoctoral researchers. Our model shows that capping the duration of postdoctoral careers, a policy proposed previously, favours international postdoctoral researchers. The analysis suggests that the leverage point to help the growth of domestic research workforce is in the pregraduate education area, and many policies implemented at the postgraduate level have minimal or unintended effects on diversity. 

Keywords: Research workforce development;diversity;biomedical science;postdoctoral researchers;National Institutes of Health.


Abstract: What happens within the university-based research enterprise when a federal funding agency abruptly changes research grant funding levels, up or down? We use simple difference equation models to show that an apparently modest increase or decrease in funding levels can have dramatic effects on researchers, graduate students, postdocs, and the overall research enterprise. The amplified effect is due to grants lasting for an extended period, thereby requiring the majority of funds available in one year to pay for grants awarded in previous years. We demonstrate the effect in various ways, using National Institutes of Health data for two situations: the historical doubling of research funding from 1998 to 2003 and the possible effects of “sequestration” in January 2013. We posit human responses to such sharp movements in funding levels and offer suggestions for amelioration.

Keywords: Research funding, grants, grant duration, sequestration, system dynamics, modeling, simulation


Supplement files


Project dynamics:

  • Parvan K., Rahmandad H., Haghani A., Inter-phase Feedbacks in Construction Projects, (working paper).

Abstract: Understanding diverse project performance trajectories is an increasingly important area for applying dynamic models. While interactions between multiple phases of projects are commonly assumed to be important in project dynamics, the strength of these feedback mechanisms has not been rigorously evaluated. In this study we use data from 15 construction projects to estimate the feedbacks between design and construction phases. The estimated factors reveal that undiscovered design rework diminishes construction quality and production rate significantly and construction completion speeds up the detection of undiscovered design rework. Together these feedbacks can explain as much as 20% of variability in overall project costs. Comparison of model predictions with a separate set of 15 projects shows good predictive power for cost and schedule outcomes and their uncertainty. Methodologically, the modeling and estimation framework offers a template for estimating both case-specific and case-independent parameters of dynamic models using data from multiple cases.

Abstract: In this paper we discuss a dynamic efficiency measurement model for evaluating the performance of highway maintenance policies where the inter-temporal dependencies between consumption of inputs (i.e., maintenance budget) and realization of outputs (i.e., improvement in road condition) are explicitly captured. We build on a micro representation of pavement deterioration and renewal processes and study the impact of the allocation of scarce maintenance budgets over time. We provide a measure of efficiency that contrasts the optimized budget allocations to the actual ones. The developed model is then applied to an empirical dataset of pavement condition and maintenance expenditures over the years 2002 to 2008 corresponding to seventeen miles of interstate highway that lay in one of the counties in the state of Virginia, USA. The policies that were found through optimization showed that road authorities should give higher priorities to preventive maintenance than corrective maintenance. In essence, by applying preventive maintenance, the road authorities can effectively decrease the need for future corrective maintenance while spending less overall.

Abstract: In software markets, pricing and openness interact to determine profitability through the direct effect of openness on feasible prices and costs, and their impact on different reinforcing feedback loops such as network effects. We develop a simulation model of the inner workings of a software firm, product life cycle and market aggregation, and analyze their interactions in a competitive market. The reinforcing loops increase the value of early market lead and put pressure on the competing firms to seek such advantage. The competitive equilibrium under strong reinforcing loops calls for highly open software products with early discounts that significantly compromise the profitability of the players. Proprietary platforms and higher prices are favored when those mechanisms are weak. We discuss the implications of this analysis for the dynamics of different competitive markets as well as the contributions to the system dynamics literature in analyzing Nash equilibrium of differential games. 


Purpose – The importance of physical assets has been increasingly recognized in recent decades. The significant returns on small improvements in overall equipment effectiveness (OEE) justify investment in the management of physical assets, but the wide variation of OEE across firms raises a question: “Why do these differences persist despite a high return on investments to maximize OEE?”. To address this question the dynamic processes that control the evolution of OEE through time need to be better understood. This paper aims to answer this question.

Design/methodology/approach – Building on insights from system dynamics and strategy literature, the paper maps the reinforcing feedback loops governing the maintenance function and its interactions with various elements in a firm. Building on strategy literature it hypothesizes that these loops can explain wide variations in observed persistent variations in OEE among otherwise similar firms. The paper draws on previous literature, extensive case studies and consulting projects to provide such mapping using the qualitative mapping tools from system dynamics.

Findings – The research outlines several reinforcing loops; once active, any of them could lead a firm towards a problematic mode of operation where reactive maintenance, poor morale, and a culture of fire-fighting dominate. Actions taken to fix problems in the short-run often activate vicious cycles, erode the capability of the organization over the long run, and lead to a lower OEE.

Social implications – Knowing the factors affecting the asset management function of a plant increases the plant's safety and limits its environmental hazards.

Originality/value – Some of the common dynamics of organizations' asset management practices are illustrated and modeled. The strategic importance of OEE and its effect on companies' market capitalization is demonstrated.

Keywords: Maintenance, production equipment, cost effectiveness, asset management, performance management.


Abstract: The rework cycle is at the heart of modeling projects, one of the major research and application areas in system dynamics. The current formulations for the rework cycle assume each task is either defective or not. Yet in many projects multiple defects can occur in one task. In this study we introduce a new rework cycle formulation that accounts for multiple defects per task and flexibly captures the testing process. We compare this formulation with two established formulations from the literature. Analysis shows some differences in simulated projects' finish time and delivered quality across different models and we discuss how these differences depend on project characteristics and task structure. The new formulation could be especially useful when data on tasks, defects, and testing have different measures, and when each new defect in a task significantly increases the chances of the task being rejected.


Abstract: Effective highway maintenance depends on several activities, including the understanding of current and the prediction of future pavement conditions and deciding how to best allocate limited resources for maintenance operations. In this paper, a dynamic micro-level simulation model of highway deterioration and renewal processes is presented. This model is calibrated with data from eight road sections in Virginia and is coupled with an optimization module that optimizes maintenance operations. The analysis offers alternative priority setting schemes that improves current maintenance practices at the project and network levels. This approach provides a blueprint for designing optimal highway maintenance practices.


Abstract: In a concurrent development process different releases of a software product overlap. Organizations involved in concurrent software development not only experience the dynamics common to single projects, but also face interactions between different releases of their product: they share resources among different stages of different projects, including customer support, they have a common code base and architecture that carries problems across releases, they use the same capabilities, and their market success in early releases impacts their resources in later ones. Drawing on two case studies we discuss some of the feedback processes central to concurrent software development and build a simple simulation model to analyze the resulting dynamics. This model sheds light on tipping dynamics, the nature of inter-project feedback loops, and alternative resource allocation policies relevant to management of concurrent software development.


Organizational capabilities and learning:

  • Ashouri-Rad A., Rahmandad H. Reconstructing Online Behavior through Effort Minimization, (working paper).

Abstract: Increasingly online interaction data informs our understanding of fundamental patterns of human behavior as well as commercial and social enterprises. However, this data is often limited to traces of users’ interactions with digital objects (e.g. votes, likes, shares) and does not include potentially relevant data on what actually people observe online. One may expect notable enhancements in understanding and prediction if what users see could be estimated. We propose a method to reconstruct online behavior based on data commonly available in practice. The method infers a user’s most likely browsing trajectory assuming that people minimize their effort in online browsing. Applying this method to data from a social news website, we obtain significant improvements in prediction and inference in comparison with multiple alternatives across a collaborative filtering and a regression validation problem.


  • Bandari R., Rahmandad H., Roychowdhury V.P., Extracting Gestalts through Socially Regularized Large-Scale Content Analysis, (working paper).

Abstract: Massive Internet-scale data, comprising content and associated user action logs, have raised hopes that data-driven models in the social sciences will proliferate. This promise, however, is predicated on our ability to identify and compute socially-significant and summative macroscopic observables analogs of quantities such as temperature or pressure in the natural sciences. Extant summarization methods tend to focus on a single dimension of a complex and social phenomenon: Topic modeling, for example, analyzes text while missing the social context and requires complex mathematical regularizations. We, instead, hark back to sociological theories of structure, and postulate that at any time, a few underlying functional macrostructures, i.e., gestalts, emerge as the driving force, giving meaning to both content and user actions. We present an automated and unsupervised gestalt computing (GC) framework that turns the scale of the data to its advantage: It first determines the functional footprint of these latent gestalts as collective patterns in user action, and then uses these contexts as social regularization to compute distributions that describe the gestalts. We apply GC to four years of data from a popular social news site. The resulting gestalt map reveals a nuanced panorama, studded with distinct political preferences and themes and vividly captures the transformative effect of a contentious political event. The derived themes are not only statistically significant, but also meaningful to human understanding. GC is scalable, provides a compelling summary of the corpus, and being rooted in sociological concepts, has the potential to apply to different contexts.


  • Rahmandad H., Henderson R., Repenning N.P., Making the Numbers? “Short Termism” & the Puzzle of Only Occasional Disaster, (working paper).

Abstract: Much recent work in strategy and popular discussion suggests that an excessive focus on “managing the numbers”—delivering quarterly earnings at the expense of longer term performance—makes it difficult for firms to make the investments necessary to build competitive advantage. “Short termism” has been blamed for everything from the decline of the US automobile industry to the low penetration of techniques such as TQM and continuous improvement. Yet a vigorous tradition in the accounting literature establishes that firms routinely sacrifice long-term investment to manage earnings and are rewarded for doing so. This paper presents a model that reconciles these apparently contradictory perspectives. We show that if the source of long-term advantage is modeled as a stock of capability that accumulates over time, a firm’s proclivity to manage short-term earnings at the expense of long-term investment can have very different consequences depending on whether the firm’s capability is close to a critical “tipping threshold.” When the firm operates above this threshold, managing earnings smoothes revenue and cash flow with few long-term consequences. Below it, managing earnings can tip the firm into a vicious cycle of accelerating decline. Our results have important implications for understanding managerial incentives and the internal processes that create sustained advantage.


Abstract: The notion of capability is widely invoked to explain differences in organizational performance and research shows that strategically relevant capabilities can be both built and lost. However, while capability development is widely studied, capability erosion has not been integrated into our understanding of performance heterogeneity. To understand erosion, we study two software development organizations that experienced diverging capability trajectories despite similar organizational and technological settings. Building a simulation-based theory, we identify the adaptation trap, a mechanism through which managerial learning can lead to capability erosion: well-intentioned efforts by managers to search locally for the optimal workload balance lead them to systematically overload their organization and, thereby, cause capabilities to erode. The analysis of our model informs when capability erosion is likely and strategically relevant.

Keywords: Capability;Erosion;Resource Based View;Simulation;System Dynamics


Abstract: Developing organizational capabilities and resources is tightly connected with a firm’s performance in competitive markets. Therefore setting investment priorities among production, product development, brand name, internationalization, and many other capabilities and resources should be understood in the context of competitive pressures and growth opportunities a firm faces. Building on resource based view, this study examines firm level capability development as it relates to market level dynamics of competition and growth through simulation experiments. Investing in capabilities that enhance performance in the short-run become superior to investing in long-term initiatives as the former strengthens the reinforcing loop between performance, available effort, and capability development; providing growth opportunities and competitive edge. Moreover, in strategic competition, firms are forced to further ignore long-term capability building in favor of survival. We explore how tradeoffs between short-term and long-term investments depend on different firm and industry characteristics. These results provide a complementary explanation for the persistence of myopic organizational decisions that does not rest on discounting, short-term managerial incentives, decision biases, or learning arguments. The results also point to another mechanism through which market competition may disfavor firms with highest long-term performance potential.

Keywords: Resource based view, capability, dynamics, simulation, competition, firm performance.


Abstract: Elwin, Juslin, Olsson, and Enkvist (2007) and Henriksson, Elwin, and Juslin (2010) offered the constructivist coding hypothesis to describe how people code the outcomes of their decisions when availability of feedback is conditional on the decision. They provided empirical evidence only for the .5 base rate condition. This commentary argues that the constructivist coding hypothesis imposes an ever-declining selection rate and overestimates base rate bias for high base rate conditions. We provide support based on a simulation model of learning under selective feedback with different base rates. Then we discuss possible extensions to constructivist coding that can help overcome the problem.


Abstract: Understanding barriers to organizational learning is central to understanding firm performance. We investigate the role of time delays between taking an action and observing the results in impeding learning. These delays, ubiquitous in real-world settings, are relevant to tradeoffs between long term and short term. We build four learning heuristics, with different levels of complexity and rationality, and analyze their performance in a simple resource allocation task. All reliably converge to the optimal solution when there are no/short delays, and when those delays are correctly assessed. However, learning is slowed significantly when decision makers err in assessing the length of the delay. In many cases, the decision maker never finds the optimal solution, wandering in the action space or converging to a suboptimal allocation. Results are robust to the organization's level of rationality. The proposed heuristics can be applied to a range of problems for modeling learning from experience in the presence of delays.


Abstract: We examine how delays between actions and their consequent payoffs affect the process of organizational adaptation. Formal conceptions of organizational learning typically include the assumption that payoffs immediately follow their antecedent actions, making the search for better strategies relatively straightforward. However, previous actions influence current organizational performance through their effects on organizational resources and capabilities. These resources and capabilities cannot be modified instantly, so delays—from actions to changes in resources and capabilities to altered organizational performance—are inevitable. Our computational experiments show that delays increase learning complexity and performance heterogeneity through two mechanisms. First, complexity of state-space and, therefore, of learning grows exponentially with delay length. Second, the time required to experience the benefits of long-term strategies means the intermediate steps of those strategies are initially undervalued, prompting premature abandonment of potentially fruitful regions of the strategy space. We find that these mechanisms often cause organizations to converge to suboptimal, routine-like cycles of actions, based on organizations' continually updated cognitive maps of how actions influence payoffs. Furthermore, the evolution of these cognitive maps exhibits path dependence, leading to heterogeneity across organizations. Implications for overcoming temporal complexity and the impact of initial cognitive maps are discussed.

Keywords: Organizational learning; delay; complexity; simulation; heterogeneity; path dependence


Modeling and estimation methods:

Abstract: Public policies often fail to achieve their intended result because of the complexity of both the environment and the policy-making process. In this article, we review the benefits of using small system dynamics models to address public policy questions. First we discuss the main difficulties inherent in the public policy-making process. Then, we discuss how small system dynamics models can address policy-making difficulties by examining two promising examples: the first in the domain of urban planning and the second in the domain of social welfare. These examples show how small models can yield accessible, insightful lessons for policy making stemming from the endogenous and aggregate perspective of system dynamics modeling and simulation.


Abstract: When is it better to use agent-based (AB) models, and when should differential equation (DE) models be used? Whereas DE models assume homogeneity and perfect mixing within compartments, AB models can capture heterogeneity across individuals and in the network of interactions among them. AB models relax aggregation assumptions, but entail computational and cognitive costs that may limit sensitivity analysis and model scope. Because resources are limited, the costs and benefits of such disaggregation should guide the choice of models for policy analysis. Using contagious disease as an example, we contrast the dynamics of a stochastic AB model with those of the analogous deterministic compartment DE model. We examine the impact of individual heterogeneity and different network topologies, including fully connected, random, Watts-Strogatz small world, scale-free, and lattice networks. Obviously, deterministic models yield a single trajectory for each parameter set, while stochastic models yield a distribution of outcomes. More interestingly, the DE and mean AB dynamics differ for several metrics relevant to public health, including diffusion speed, peak load on health services infrastructure, and total disease burden. The response of the models to policies can also differ even when their base case behavior is similar. In some conditions, however, these differences in means are small compared to variability caused by stochastic events, parameter uncertainty, and model boundary. We discuss implications for the choice among model types, focusing on policy design. The results apply beyond epidemiology: from innovation adoption to financial panics, many important social phenomena involve analogous processes of diffusion and social contagion.

Keywords: Agent-based models; networks; scale free; small world; heterogeneity; epidemiology; simulation; system dynamics; complex adaptive systems; SEIR model.