nep-cmp New Economics Papers
on Computational Economics
Issue of 2022‒04‒18
twenty papers chosen by
Stan Miles
Thompson Rivers University

  1. Machine Learning for Stock Prediction Based on Fundamental Analysis By Yuxuan Huang; Luiz Fernando Capretz; Danny Ho
  2. Nonsmooth Implicit Differentiation for Machine Learning and Optimization By Bolte, Jérôme; Le, Tam; Pauwels, Edouard; Silveti-Falls, Antonio
  3. Making use of supercomputers in financial machine learning. By Philippe Cotte; Pierre Lagier; Vincent Margot; Christophe Geissler
  4. Closure operators: Complexity and applications to classification and decision-making By Hamed Hamze Bajgiran; Federico Echenique
  5. Towards accountability in machine learning applications: A system-testing approach By Wan, Wayne Xinwei; Lindenthal, Thies
  6. Inspection-L: Practical GNN-Based Money Laundering Detection System for Bitcoin By Wai Weng Lo; Siamak Layeghy; Marius Portmann
  7. Estimating risks of option books using neural-SDE market models By Samuel N. Cohen; Christoph Reisinger; Sheng Wang
  8. GAM(L)A: An econometric model for interpretable Machine Learning By Emmanuel Flachaire; Gilles Hacheme; Sullivan Hu\'e; S\'ebastien Laurent
  9. Performance of long short-term memory artificial neural networks in nowcasting during the COVID-19 crisis By Daniel Hopp
  10. Using Past Violence and Current News to Predict Changes in Violence By Mueller, H.; Rauh, C.
  11. Collusion and Artificial Intelligence: A Computational Experiment with Sequential Pricing Algorithms under Stochastic Costs By Gonzalo Ballestero
  12. Examining spatial disparities in electric vehicle charging station placements using machine learning By Roy, Avipsa; Law, Mankin
  13. Distributionally robust risk evaluation with causality constraint and structural information By Bingyan Han
  14. Missing top incomes and tax-benefit microsimulation: evidence from correcting household survey data using tax records data By Marko Ledic; Ivica Rubil; Ivica Urban
  15. Calibration of Operating Reserve Demand Curves using Monte Carlo Simulations By Cartuyvels, Jacques; Papavasiliou, Anthony
  16. Hybrid order picking: A simulation model of a joint manual and autonomous order picking system. By Winkelhaus, Sven; Zhang, Minqi; Grosse, E. H.; Glock, C. H.
  17. Family Tax Splitting : A Microsimulation of Its Potential Labour Supply and Intra-household Welfare Effects in Germany By Miriam Beblo; Denis Beninger; Francois Laisney
  18. Artificial intelligence and firm-level productivity By Czarnitzki, Dirk; Fernández, Gastón P.; Rammer, Christian
  19. A Real-Business-Cycle Model with Financial Liberalization: Lessons for Bulgaria (1999-2020) By Aleksandar Vasilev
  20. Statistical inference for intrinsic wavelet estimators of SPD covariance matrices in a log-Euclidean manifold By Krebs, Johannes; Rademacher, Daniel; von Sachs, Rainer

  1. By: Yuxuan Huang; Luiz Fernando Capretz; Danny Ho
    Abstract: Application of machine learning for stock prediction is attracting a lot of attention in recent years. A large amount of research has been conducted in this area and multiple existing results have shown that machine learning methods could be successfully used toward stock predicting using stocks historical data. Most of these existing approaches have focused on short term prediction using stocks historical price and technical indicators. In this paper, we prepared 22 years worth of stock quarterly financial data and investigated three machine learning algorithms: Feed-forward Neural Network (FNN), Random Forest (RF) and Adaptive Neural Fuzzy Inference System (ANFIS) for stock prediction based on fundamental analysis. In addition, we applied RF based feature selection and bootstrap aggregation in order to improve model performance and aggregate predictions from different models. Our results show that RF model achieves the best prediction results, and feature selection is able to improve test performance of FNN and ANFIS. Moreover, the aggregated model outperforms all baseline models as well as the benchmark DJIA index by an acceptable margin for the test period. Our findings demonstrate that machine learning models could be used to aid fundamental analysts with decision-making regarding stock investment.
    Date: 2022–01
  2. By: Bolte, Jérôme; Le, Tam; Pauwels, Edouard; Silveti-Falls, Antonio
    Abstract: In view of training increasingly complex learning architectures, we establish a nonsmooth implicit function theorem with an operational calculus. Our result applies to most practical problems (i.e., definable problems) provided that a nonsmooth form of the classical invertibility condition is fulfilled. This approach allows for formal subdifferentiation: for instance, replacing derivatives by Clarke Jacobians in the usual differentiation formulas is fully justified for a wide class of nonsmooth problems. Moreover this calculus is entirely compatible with algorithmic differentiation (e.g., backpropagation). We provide several applications such as training deep equilibrium networks, training neural nets with conic optimization layers, or hyperparameter-tuning for nonsmooth Lasso-type models. To show the sharpness of our assumptions, we present numerical experiments showcasing the extremely pathological gradient dynamics one can encounter when applying implicit algorithmic differentiation without any hypothesis.
    Date: 2022–03
  3. By: Philippe Cotte (Advestis); Pierre Lagier (Fujitsu Laboratories of Europe Ltd. - Fujitsu Laboratories Ltd.); Vincent Margot (Advestis); Christophe Geissler (Advestis)
    Abstract: This article is the result of a collaboration between Fujitsu and Advestis. This collaboration aims at refactoring and running an algorithm based on systematic exploration producing investment recommendations on a high-performance computer of the Fugaku, to see whether a very high number of cores could allow for a deeper exploration of the data compared to a cloud machine, hopefully resulting in better predictions. We found that an increase in the number of explored rules results in a net increase in the predictive performance of the final ruleset. Also, in the particular case of this study, we found that using more than around 40 cores does not bring a significant computation time gain. However, the origin of this limitation is explained by a threshold-based search heuristic used to prune the search space. We have evidence that for similar data sets with less restrictive thresholds, the number of cores actually used could very well be much higher, allowing parallelization to have a much greater effect.
    Keywords: RIPE,Portfolio Management,rule-based algorithm,Expert Systems,High Performance Computing,Parallel Programming,Multiprocessing,XAI
    Date: 2022–02–28
  4. By: Hamed Hamze Bajgiran; Federico Echenique
    Abstract: We study the complexity of closure operators, with applications to machine learning and decision theory. In machine learning, closure operators emerge naturally in data classification and clustering. In decision theory, they can model equivalence of choice menus, and therefore situations with a preference for flexibility. Our contribution is to formulate a notion of complexity of closure operators, which translate into the complexity of a classifier in ML, or of a utility function in decision theory.
    Date: 2022–02
  5. By: Wan, Wayne Xinwei; Lindenthal, Thies
    Abstract: A rapidly expanding universe of technology-focused startups is trying to change and improve the way real estate markets operate. The undisputed predictive power of machine learning (ML) models often plays a crucial role in the 'disruption' of traditional processes. However, an accountability gap prevails: How do the models arrive at their predictions? Do they do what we hope they do - or are corners cut? Training ML models is a software development process at heart. We suggest to follow a dedicated software testing framework and to verify that the ML model performs as intended. Illustratively, we augment two ML image classifiers with a system testing procedure based on local interpretable model-agnostic explanation (LIME) techniques. Analyzing the classifications sheds light on some of the factors that determine the behavior of the systems.
    Keywords: machine learning,accountability gap,computer vision,real estate,urban studies
    JEL: C52 R30
    Date: 2022
  6. By: Wai Weng Lo; Siamak Layeghy; Marius Portmann
    Abstract: Criminals have become increasingly experienced in using cryptocurrencies, such as Bitcoin, for money laundering. The use of cryptocurrencies can hide criminal identities and transfer hundreds of millions of dollars of dirty funds through their criminal digital wallets. However, this is considered a paradox because cryptocurrencies are gold mines for open-source intelligence, allowing law enforcement agencies to have more power in conducting forensic analyses. This paper proposed Inspection-L, a graph neural network (GNN) framework based on self-supervised Deep Graph Infomax (DGI), with Random Forest (RF), to detect illicit transactions for Anti-Money laundering (AML). To the best of our knowledge, our proposal is the first of applying self-supervised GNNs to the problem of AML in Bitcoin. The proposed method has been evaluated on the Elliptic dataset and shows that our approach outperforms the state-of-the-art in terms of key classification metrics, which demonstrates the potential of self-supervised GNN in cryptocurrency illicit transaction detection.
    Date: 2022–03
  7. By: Samuel N. Cohen; Christoph Reisinger; Sheng Wang
    Abstract: In this paper, we examine the capacity of an arbitrage-free neural-SDE market model to produce realistic scenarios for the joint dynamics of multiple European options on a single underlying. We subsequently demonstrate its use as a risk simulation engine for option portfolios. Through backtesting analysis, we show that our models are more computationally efficient and accurate for evaluating the Value-at-Risk (VaR) of option portfolios, with better coverage performance and less procyclicality than standard filtered historical simulation approaches.
    Date: 2022–02
  8. By: Emmanuel Flachaire; Gilles Hacheme; Sullivan Hu\'e; S\'ebastien Laurent
    Abstract: Despite their high predictive performance, random forest and gradient boosting are often considered as black boxes or uninterpretable models which has raised concerns from practitioners and regulators. As an alternative, we propose in this paper to use partial linear models that are inherently interpretable. Specifically, this article introduces GAM-lasso (GAMLA) and GAM-autometrics (GAMA), denoted as GAM(L)A in short. GAM(L)A combines parametric and non-parametric functions to accurately capture linearities and non-linearities prevailing between dependent and explanatory variables, and a variable selection procedure to control for overfitting issues. Estimation relies on a two-step procedure building upon the double residual method. We illustrate the predictive performance and interpretability of GAM(L)A on a regression and a classification problem. The results show that GAM(L)A outperforms parametric models augmented by quadratic, cubic and interaction effects. Moreover, the results also suggest that the performance of GAM(L)A is not significantly different from that of random forest and gradient boosting.
    Date: 2022–03
  9. By: Daniel Hopp
    Abstract: The COVID-19 pandemic has demonstrated the increasing need of policymakers for timely estimates of macroeconomic variables. A prior UNCTAD research paper examined the suitability of long short-term memory artificial neural networks (LSTM) for performing economic nowcasting of this nature. Here, the LSTM's performance during the COVID-19 pandemic is compared and contrasted with that of the dynamic factor model (DFM), a commonly used methodology in the field. Three separate variables, global merchandise export values and volumes and global services exports, were nowcast with actual data vintages and performance evaluated for the second, third, and fourth quarters of 2020 and the first and second quarters of 2021. In terms of both mean absolute error and root mean square error, the LSTM obtained better performance in two-thirds of variable/quarter combinations, as well as displayed more gradual forecast evolutions with more consistent narratives and smaller revisions. Additionally, a methodology to introduce interpretability to LSTMs is introduced and made available in the accompanying nowcast_lstm Python library, which is now also available in R, MATLAB, and Julia.
    Date: 2022–03
  10. By: Mueller, H.; Rauh, C.
    Abstract: This article proposes a new method for predicting escalations and de†escalations of violence using a model which relies on conflict history and text features. The text features are generated from over 3.5 million newspaper articles using a so†called topic†model. We show that the combined model relies to a large extent on conflict dynamics, but that text is able to contribute meaningfully to the prediction of rare outbreaks of violence in previously peaceful countries. Given the very powerful dynamics of the conflict trap these cases are particularly important for prevention efforts.
    Keywords: Conflict, prediction, machine learning, LDA, topic model, battle deaths, ViEWS prediction competition, random forest
    JEL: F21 C53 C55
    Date: 2022–03–22
  11. By: Gonzalo Ballestero (Universidad de San Andrés)
    Abstract: Firms increasingly delegate their strategic decisions to algorithms. A potential concern is that algorithms may undermine competition by leading to pricing outcomes that are collusive, even without having been designed to do so. This paper investigates whether Q-learning algorithms can learn to collude in a setting with sequential price competition and stochastic marginal costs adapted from Maskin and Tirole (1988). By extending a previous model developed in Klein (2021), I find that sequential Q-learning algorithms leads to supracompetitive profits despite they compete under uncertainty and this finding is robust to various extensions. The algorithms can coordinate on focal price equilibria or an Edgeworth cycle provided that uncertainty is not too large. However, as the market environment becomes more uncertain, price wars emerge as the only possible pricing pattern. Even though sequential Q-learning algorithms gain supracompetitive profits, uncertainty tends to make collusive outcomes more dicult to achieve.
    Keywords: Competition Policy, Artificial Intelligence, Algorithmic Collusion
    JEL: D43 K21 L13
    Date: 2022–02
  12. By: Roy, Avipsa; Law, Mankin
    Abstract: Electric vehicles (EV) are an emerging mode of transportation that has the potential to reshape the transportation sector by significantly reducing carbon emissions thereby promoting a cleaner environment and pushing the boundaries of climate progress. Nevertheless, there remain significant hurdles to the widespread adoption of electric vehicles in the United States ranging from the high cost of EVs to the inequitable placement of EV charging stations (EVCS). A deeper understanding of the underlying complex interactions of social, economic, and demographic factors which may lead to such emerging disparities in EVCS placements is, therefore, necessary to mitigate accessibility issues and improve EV usage among people of all ages and abilities. In this study, we develop a machine learning framework to examine spatial disparities in EVCS placements by using a predictive approach. We first identify the essential socioeconomic factors that may contribute to spatial disparities in EVCS access. Second, using these factors along with ground truth data from existing EVCS placements we predict future ECVS density at multiple spatial scales using machine learning algorithms and compare their predictive accuracy to identify the most optimal spatial resolution for our predictions. Finally, we compare the most accurately predicted EVCS placement density with a spatial inequity indicator to quantify how equitably these placements would be for Orange County, California. Our method achieved the highest predictive accuracy (94.9%) of EVCS placement density at a spatial resolution of 3 km using Random Forests. Our results indicate that a total of 74.18% of predicted EVCS placements in Orange County will lie within a low spatial equity zone – indicating populations with the lowest accessibility may require the highest investments in EVCS placements. Within the low spatial equity areas, 14.86% of the area will have a low density of predicted EVCS placements, 50.32% will have a medium density of predicted EVCS placement, and only 9% tend to have high EVCS placements. The findings from this study highlight a generalizable framework to quantify inequities in EVCS placements that will enable policymakers to identify underserved communities and facilitate targeted infrastructure investments for widespread EV usage and adoption for all.
    Date: 2022–02–22
  13. By: Bingyan Han
    Abstract: This work studies distributionally robust evaluation of expected function values over temporal data. A set of alternative measures is characterized by the causal optimal transport. We prove the strong duality and recast the causality constraint as minimization over an infinite-dimensional test function space. We approximate test functions by neural networks and prove the sample complexity with Rademacher complexity. Moreover, when structural information is available to further restrict the ambiguity set, we prove the dual formulation and provide efficient optimization methods. Simulation on stochastic volatility and empirical analysis on stock indices demonstrate that our framework offers an attractive alternative to the classic optimal transport formulation.
    Date: 2022–03
  14. By: Marko Ledic (Faculty of Economics & Business, University of Zagreb); Ivica Rubil (The Institute of Economics, Zagreb); Ivica Urban (Institute of Public Finance, Zagreb)
    Abstract: Using the microsimulation model EUROMOD for Croatia, we compare the results of simulation based on the original survey data (EU-SILC) with those based on the survey data corrected using tax records data and a recent survey correction method. We show that the correction method, although it debiases inequality estimates, may not be able to correct the in-come structure by source if some income sources are severely under-represented. In Croatia, this is the case for income from capital, property, and contractual work. As a solution, we propose to complement the correction method with an ad hoc pre-correction procedure. The corrections bring the aggregate amount, distribution, and structure of survey income closer to those in the tax data. Consequently, the simulated fiscal instruments become more like those in the tax data. Simulation of a hypothetical tax reform shows the results based on the uncorrected data may be misleading in terms of the estimated budgetary impact and the distributional incidence of the reform.
    Keywords: top incomes, survey data, tax records, tax-benefit microsimulation, EUROMOD, EU-SILC
    JEL: D31 H24
    Date: 2022–03
  15. By: Cartuyvels, Jacques (Université catholique de Louvain, LIDAM/CORE, Belgium); Papavasiliou, Anthony (Université catholique de Louvain, LIDAM/CORE, Belgium)
    Abstract: Scarcity pricing has been proposed to enhance investment in flexible assets through the use of an adder on real-time energy and the application of that adder on real-time reserve. We implement a Monte-Carlo simulator for obtaining statistically confident estimates of scarcity pricing adders which is motivated from the implementation of this mechanism in Belgium. The analysis is based on a multi-level, multi-horizon simulation of day-ahead and real-time operations in the Belgian market. The methodology relies on k-means clustering for selecting a set of representative day-ahead forecasts, followed by the generation of synthetic real-time load scenarios for simulating real-time operations.
    Keywords: Operating reserve demand curve ; scarcity pricing ; unit commitment ; k-means
    Date: 2022–02–01
  16. By: Winkelhaus, Sven; Zhang, Minqi; Grosse, E. H.; Glock, C. H.
    Date: 2022–05
  17. By: Miriam Beblo (Centre for European Economic Research (Mannheim, Germany) - Zentrum für Europäische Wirtschaftsforschung (ZEW) - Universität Mannheim [Mannheim]); Denis Beninger (Centre for European Economic Research (Mannheim, Germany) - Zentrum für Europäische Wirtschaftsforschung (ZEW) - Universität Mannheim [Mannheim]); Francois Laisney (BETA - Bureau d'Économie Théorique et Appliquée - INRA - Institut National de la Recherche Agronomique - UNISTRA - Université de Strasbourg - UL - Université de Lorraine - CNRS - Centre National de la Recherche Scientifique, Centre for European Economic Research (Mannheim, Germany) - Zentrum für Europäische Wirtschaftsforschung (ZEW) - Universität Mannheim [Mannheim])
    Abstract: This paper assesses the effects that an introduction of the French family splitting mechanism would have on German families' labour supply and intra-household consumption behaviour. We use simulated real world microdata created by means of a 'deterministic' collective labour supply model. The data are generated by a compound procedure of estimation and calibration based on GSOEP data. In a microsimulation the present tax-benefit system with child benefit/allowance is replaced by a tax scheme with family splitting. The resulting changes in labour supply are surprisingly small, even for women. Welfare effects are also modest, but differ for husbands and wives.
    Keywords: Collective model,Household labour supply,Intra-household allocation,Tax reform,Family splitting
    Date: 2022–03–25
  18. By: Czarnitzki, Dirk; Fernández, Gastón P.; Rammer, Christian
    Abstract: Artificial Intelligence (AI) is often regarded as the next general-purpose technology with a rapid, penetrating, and far-reaching use over a broad number of industrial sectors. A main feature of new general-purpose technology is to enable new ways of production that may increase productivity. So far, however, only very few studies investigated likely productivity effects of AI at the firm-level; presumably because of lacking data. We exploit unique survey data on firms' adoption of AI technology and estimate its productivity effects with a sample of German firms. We employ both a cross-sectional dataset and a panel database. To address the potential endogeneity of AI adoption, we also implement an IV approach. We find positive and significant effects of the use of AI on firm productivity. This finding holds for different measures of AI usage, i.e., an indicator variable of AI adoption, and the intensity with which firms use AI methods in their business processes.
    Keywords: Artificial Intelligence,Productivity,CIS data
    JEL: O14 O31 O33 L25 M15
    Date: 2022
  19. By: Aleksandar Vasilev (Lincoln International Business School, UK.)
    Abstract: Financial openness is introduced into a real-business-cycle setup augmented with a detailed government sector. The model is calibrated to Bulgarian data for the period following the introduction of the currency board arrangement (1999-2020). The quantitative importance of financial openness is investigated for the stabilization of cyclical fluctuations in Bulgaria. The computational experiment performed in this paper reveals that greater financial openness increases the impact of technology shocks on output, investment, consumption, labor hours, and net exports. This amplification effect is due to the following mechanism: openness provides a cheap access to foreign funds. Unfortunately, the new results come at odds with a major empirical observation, i.e. that consumption and net exports are strongly pro-cyclical; the model, however, produces a countercyclical consumption, as well as net exports. Thus, such a setup is not yet ready to be used for policy analysis.
    Keywords: business cycles, progressive capital taxation, Bulgaria
    JEL: E24 E32
    Date: 2022–04
  20. By: Krebs, Johannes; Rademacher, Daniel; von Sachs, Rainer
    Abstract: In this paper we treat statistical inference for an intrinsic wavelet estimator of curves of symmetric positive definite (SPD) matrices in a log-Euclidean manifold. This estimator preserves positive-definiteness and enjoys permutation-equivariance, which is particularly relevant for covariance matrices. Our second-generation wavelet estimator is based on average-interpolation and allows the same powerful properties, including fast algorithms, known from nonparametric curve estimation with wavelets in standard Euclidean set-ups. The core of our work is the proposition of confidence sets for our high-level wavelet estimator in a non-Euclidean geometry. We derive asymptotic normality of this estimator, including explicit expressions of its asymptotic variance. This opens the door for constructing asymptotic confidence regions which we compare with our proposed bootstrap scheme for inference. Detailed numerical simulations confirm the appropriateness of our suggested inference schemes.
    Keywords: Asymptotic normality ; Average interpolation ; Covariance matrices ; Intrinsic polynomials ; log-Euclidean manifold ; SPD matrices ; Matrix-valued curves ; Nonparametric inference ; Second generation wavelets
    Date: 2022–02–14

This nep-cmp issue is ©2022 by Stan Miles. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
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.