nep-cmp New Economics Papers
on Computational Economics
Issue of 2017‒05‒07
eight papers chosen by

  1. Agent-Based Model Calibration using Machine Learning Surrogates By Francesco Lamperti; Andrea Roventini; Amir Sani
  2. The Creative Response and the Endogenous Dynamics of Pecuniary Knowledge Externalities: An Agent Based Simulation Model. By Antonelli, Cristiano; Ferraris, Gianluigi
  3. Multi-objective local environmental simulator (MOLES 1.0): Model specification, algorithm design and policy applications By Ioannis Tikoudis; Walid Oueslati
  4. Profit-oriented scheduling of resource-constrained projects with flexible capacity constraints By Schnabel, André; Kellenbrink, Carolin; Helber, Stefan
  5. Modeling Economic Systems as Locally-Constructive Sequential Games By Tesfatsion, Leigh
  6. A Model of Influence on Trade Policy in a Computable General Equilibrium Model By Franck Viroleau
  7. Forecasting with many predictors using message passing algorithms By Korobilis, Dimitris
  8. Integrated Order Picking and Vehicle Routing with Due Dates By Daniel Schubert; André Scholz; Gerhard Wäscher

  1. By: Francesco Lamperti (Université Panthéon-Sorbonne - Paris 1 (UP1)); Andrea Roventini (Laboratory of Economics and Management); Amir Sani (Université Paris 1 Panthéon-Sorbonne (UP1))
    Abstract: Taking agent-based models (ABM) closer to the data is an open challenge. This paper explicitly tackles parameter space exploration and calibration of ABMs combining supervised machine-learning and intelligent sampling to build a surrogate meta-model. The proposed approach provides a fast and accurate approximation of model behaviour, dramatically reducing computation time. In that, our machine-learning surrogate facilitates large scale explorations of the parameter-space, while providing a powerful filter to gain insights into the complex functioning of agent-based models. The algorithm introduced in this paper merges model simulation and output analysis into a surrogate meta-model, which substantially ease ABM calibration. We successfully apply our approach to the Brock and Hommes (1998) asset pricing model and to the “Island” endogenous growth model (Fagiolo and Dosi, 2003). Performance is evaluated against a relatively large outof-sample set of parameter combinations, while employing different user-defined statistical tests for output analysis. The results demonstrate the capacity of machine learning surrogates to facilitate fast and precise exploration of agent-based models’ behaviour over their often rugged parameter spaces.
    Keywords: Agent-based model; Calibration; Machine learning; Surrogate; Meta-model
    JEL: C15 C52 C63
    Date: 2017–03
  2. By: Antonelli, Cristiano; Ferraris, Gianluigi (University of Turin)
    Abstract: The paper elaborates an agent based simulation model (ABM) to explore the endogenous long-term dynamics of knowledge externalities. ABMs, as a form of artificial cliometrics, allow the analysis of the effects of the reactivity of firms caught in out-of-equilibrium conditions conditional on the levels of endogenous knowledge externalities stemming from the levels of knowledge connectivity of the system. The simulation results confirm the powerful effects of endogenous knowledge externalities. At the micro-level, the reactions of firms caught in out-ofequilibrium conditions yield successful effects in the form of productivity enhancing innovations, only in the presence of high levels of knowledge connectivity and strong pecuniary knowledge externalities. At the meso-level, the introduction of innovations changes the structural characteristics of the system in terms of knowledge connectivity that affect the availability of knowledge externalities. Endogenous centrifugal and centripetal forces continually reshape the structure of the system and its knowledge connectivity. At the macro system level, an out-of-equilibrium process leads to a step-wise increase in productivity combined with non-linear patterns of output growth characterized by significant oscillations typical of the long waves in Schumpeterian business cycles.
    Date: 2017–03
  3. By: Ioannis Tikoudis; Walid Oueslati (OECD)
    Abstract: This paper describes MOLES 1.0, an integrated land-use and transport model developed with Object-Oriented Programming principles in order to combine selected characteristics from Spatial Computable General Equilibrium and microsimulation models.
    Keywords: air pollution, greenhouse gas emissions, land-use model, microsimulation, spatial general equilibrium, transport model
    JEL: C60 C68 D58 D62 H70 R00 R13 R14 R40 R52
    Date: 2017–05–04
  4. By: Schnabel, André; Kellenbrink, Carolin; Helber, Stefan
    Abstract: We consider a novel generalization of the resource-constrained project scheduling problem (RCPSP). Unlike many established approaches for the RCPSP that aim to minimize the makespan of the project for given static capacity constraints, we consider the important real-life aspect that capacity constraints can often be systematically modified by temporarily assigning costly additional production resources or using overtime. We furthermore assume that the revenue of the project decreases as its makespan increases and try to find a schedule with a profit-maximizing makespan. Like the RCPSP, the problem is $\mathcalNP$-hard, but unlike the RCPSP it turns out that an optimal schedule does not have to be among the set of so-called active schedules. Scheduling such a project is a formidable task, both from a practical and a theoretical perspective. We develop, describe, and evaluate alternative solution encodings and schedule decoding mechanisms to solve this problem within a genetic algorithm framework and we compare them to both optimal reference values and the results of a commercial local search solver called LocalSolver.
    Keywords: Project scheduling; encodings; heuristics; local-search; genetic algorithm; RCPSP; overtime
    JEL: C61 C44 M11
    Date: 2017–04
  5. By: Tesfatsion, Leigh
    Abstract: Real-world economies are open-ended dynamic systems consisting of heterogeneous interacting participants. Human participants are decision-makers who strategically take into account the past actions and potential future actions of other participants. All participants are forced to be locally constructive, meaning their actions at any given time must be based on their local states; and participant actions at any given time affect future local states. Taken together, these properties imply real-world economies are locally-constructive sequential games. This study discusses a modeling approach, agent-based computational economics (ACE), that permits researchers to study economic systems from this point of view. ACE modeling principles and objectives are first concisely presented. The remainder of the study then highlights challenging issues and edgier explorations that ACE researchers are currently pursuing.
    Date: 2017–04–30
  6. By: Franck Viroleau
    Abstract: This paper aims at making explicit the micro foundations of the government's preference function in an influence-driven political economy model. It also addresses the behavior functions of domestic and foreign firms in their attempts to gain policy favors. These favors are granted by means of subsidies. In our model, the government simultaneously chooses three interdependent policy instruments under the political influence of domestic and foreign firms. Thus, we create a political market characterized by utility-maximizing and profit-maximizing behaviors of its actors, which takes place in a computable general equilibrium model. Endowed with these features, this model fills a gap in the literature. However, our results demonstrate that the model is only valid under a reasonable set of constraints on its parameters. Finally, this paper formally shows the key role of the subsidy elasticity of political cost in limiting the distortions created by the influence of interest groups.
    Keywords: Lobbying, Public Policies, Computable General Equilibrium Model.
    JEL: C68 D72 D78 F13 H32 P16
    Date: 2017
  7. By: Korobilis, Dimitris
    Abstract: Machine learning methods are becoming increasingly popular in economics, due to the increased availability of large datasets. In this paper I evaluate a recently proposed algorithm called Generalized Approximate Message Passing (GAMP) , which has been very popular in signal processing and compressive sensing. I show how this algorithm can be combined with Bayesian hierarchical shrinkage priors typically used in economic forecasting, resulting in computationally efficient schemes for estimating high-dimensional regression models. Using Monte Carlo simulations I establish that in certain scenarios GAMP can achieve estimation accuracy comparable to traditional Markov chain Monte Carlo methods, at a tiny fraction of the computing time. In a forecasting exercise involving a large set of orthogonal macroeconomic predictors, I show that Bayesian shrinkage estimators based on GAMP perform very well compared to a large set of alternatives.
    Keywords: high-dimensional inference; compressive sensing; belief propagation; Bayesian shrinkage; dynamic factor models
    Date: 2017–01
  8. By: Daniel Schubert (Faculty of Economics and Management, Otto-von-Guericke University Magdeburg); André Scholz (Faculty of Economics and Management, Otto-von-Guericke University Magdeburg); Gerhard Wäscher (Faculty of Economics and Management, Otto-von-Guericke University Magdeburg)
    Abstract: Supermarkets typically order their goods from a centrally located distribution center (warehouse). Each order that the warehouse receives is characterized by the requested items, the location of the respective supermarket and a due date by which the items have to be delivered. For processing an order, a human operator (order picker) retrieves the requested items from their storage locations in the warehouse first. The items are then available for shipment and loaded on the vehicle which performs the tour including the respective location of the supermarket. Whether and to which extent a due date is violated (tardiness) depends on the composition of the tours, the corresponding routes and the start dates of the tours (vehicle routing subproblem). The start date of a tour, however, is also affected by the assignment of orders to pickers and the sequence according to which the orders are processed by the pickers (order picking subproblem). Although both subproblems are closely interconnected, they have not been considered simultaneously in the literature so far. In this paper, an iterated local search algorithm is designed for the simultaneous solution of the subproblems. By means of extensive numerical experiments, it is shown that the proposed approach is able to generate high-quality solutions even for large instances. Furthermore, the economic benefits of an integrated solution are investigated. Problem classes are identified, where the sequential solution of the subproblems leads to acceptable results, and it is pointed out in which cases an integrated solution is inevitable.
    Keywords: Vehicle Routing, Order Picking, Parallel Machine Scheduling, Iterated Local Search
    Date: 2017–04

General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.