publications

2024

1. Large-Scale Model Comparison with Fast Model Confidence Sets

   Sylvain Barde

   SSRN Working Paper No. 4907732, 2024

   Abs

   The paper proposes a new algorithm for finding the confidence set of a collection of forecasts or prediction models. Existing numerical implementations use an elimination approach, where one starts with the full collection of models and successively eliminates the worst performing until the null of equal predictive ability is no longer rejected at a given confidence level. The intuition behind the proposed implementation lies in reversing the process, i.e. starting with a collection of two models and updating both the model rankings and p-values as models are successively added to the collection. The first benefit of this approach is a reduction of one polynomial order in both the time complexity and memory cost of finding the confidence set of a collection of M models, falling respectively from O(M^3) to O(M^2) and from O(M^2) to O(M). The second key benefit is that it allows for further models to be added at a later point in time, thus enabling collaborative efforts using the model confidence set procedure. We prove that this implementation is equivalent to the elimination approach and demonstrate the improved performance on a multivariate GARCH collection consisting of 4800 models.

2. Moran’s I 2-Stage Lasso: for Models with Spatial Correlation and Endogenous Variables

   Sylvain Barde, Rowan Cherodian, and Guy Tchuente

   arXiv preprint arXiv:2404.02584, 2024

   Abs

   We propose a novel estimation procedure for models with endogenous variables in the presence of spatial correlation based on Eigenvector Spatial Filtering. The procedure, called Moran’s 2-Stage Lasso (Mi-2SL), uses a two-stage Lasso estimator where the Standardised Moran’s I is used to set the Lasso tuning parameter. Unlike existing spatial econometric methods, this has the key benefit of not requiring the researcher to explicitly model the spatial correlation process, which is of interest in cases where they are only interested in removing the resulting bias when estimating the direct effect of covariates. We show the conditions necessary for consistent and asymptotically normal parameter estimation assuming the support (relevant) set of eigenvectors is known. Our Monte Carlo simulation results also show that Mi-2SL performs well against common alternatives in the presence of spatial correlation. Our empirical application replicates Cadena and Kovak (2016) instrumental variables estimates using Mi-2SL and shows that in that case, Mi-2SL can boost the performance of the first stage.

3. Transportation Costs in the Age of Road Freight: Evidence from United States 1955-2010

   Sylvain Barde, and Alexander Klein

   SSRN Working Paper No. 4798888, 2024

   Abs

   This paper examines the temporal and spatial evolution of road freight transportation costs between 1955 and 2010. For that purpose, we have constructed a new data set of minimal road transport costs for freight transport in every decade between 1955 and 2010 for 3105×3105 county-pairs. We use a methodology which combines the time-evolution of the road network’s layout with spatially disaggregated transport-related costs: average driving speeds, fuel consumption, fuel prices, labour costs, and vehicle operating costs. The new data set allows to document main facts about the development of county-pair transport costs since the 1950s and examine their temporal as well as spatial changes.

2025

1. Moran’s I Lasso for models with spatially correlated data

   Sylvain Barde, Rowan Cherodian, and Guy Tchuente

   The Econometrics Journal, 2025

   Abs

   This paper proposes a Lasso-based estimator which uses information embedded in the Moran statistic to develop a selection procedure called Moran’s I Lasso (Mi-Lasso) to solve the Eigenvector Spatial Filtering (ESF) eigenvector selection problem. ESF uses a subset of eigenvectors from a spatial weights matrix to efficiently account for any omitted spatially correlated terms in a classical linear regression framework, thus eliminating the need for the researcher to explicitly specify the spatially correlated parts of the model. We proposed the first ESF procedure accounting for post-selection inference. We derive performance bounds and show the necessary conditions for consistent eigenvector selection. The key advantages of the proposed estimator are that it is intuitive, theoretically grounded, able to provide robust inference and substantially faster than Lasso based on cross-validation or any proposed forward stepwise procedure. Our simulation results and an application on house prices demonstrate Mi-Lasso performs well compared to existing procedures in finite samples.

2024

1. Bayesian estimation of large-scale simulation models with Gaussian process regression surrogates

   Sylvain Barde

   Computational Statistics & Data Analysis, 2024

   Abs

   Large scale, computationally expensive simulation models pose a particular challenge when it comes to estimating their parameters from empirical data. Most simulation models do not possess closed-form expressions for their likelihood function, requiring the use of simulation-based inference, such as simulated method of moments, indirect inference, likelihood-free inference or approximate Bayesian computation. However, given the high computational requirements of large-scale models, it is often difficult to run these estimation methods, as they require more simulated runs that can feasibly be carried out. The aim is to address the problem by providing a full Bayesian estimation framework where the true but intractable likelihood function of the simulation model is replaced by one generated by a surrogate model trained on the limited simulated data. This is provided by a Linear Model of Coregionalization, where each latent variable is a sparse variational Gaussian process, chosen for its desirable convergence and consistency properties. The effectiveness of the approach is tested using both a simulated Bayesian computing analysis on a known data generating process, and an empirical application in which the free parameters of a computationally demanding agent-based model are estimated on US macroeconomic data.

2021

1. Place-based policy in southern Italy: Evidence from a dose–response approach

   Alessandro Cusimano, Fabio Mazzola, and Sylvain Barde

   Regional Studies, 2021

   Abs

   This paper evaluates the effectiveness at a territorial level of a place-based policy for southern Italy, that is, territorial integrated projects (TIPs). We combine classical counterfactual designs and the construction of a dose–response function to assess the impact of the infrastructural interventions on the municipalities involved in a target region (Sicily). The results are robust enough to show policy effectiveness on both the number of workers and the number of plants. In the latter case, we also identify a significant and increasing dose–response function highlighting the positive relationship between funding intensity and the growth of plants.

2020

1. Macroeconomic simulation comparison with a multivariate extension of the Markov information criterion

   Sylvain Barde

   Journal of Economic Dynamics and Control, 2020

   Abs

   The paper aims to address the issue of comparing agent-based models (ABMs) with more traditional VAR and DSGE models by developing a multivariate extension of the Markov Information Criterion (MIC) of Barde (2017). The univariate MIC measures the informational distance between a simulation model and some empirical data by mapping the simulated data to a Markov transition matrix, and is proven to provide an unbiased measurement for all models reducible to a Markov process. As a result, the MIC can accurately measure distance using only simulated data, for a wide class of data generating processes. The paper first presents the multivariate extension of the MIC and its validation on VAR and DGSE models before carrying the first direct comparison between a macroeconomic ABM and a DGSE model, namely the benchmark ABM of Caiani et al. (2016) and Smets and Wouters (2007).

2017

1. A practical, accurate, information criterion for Nth order Markov processes

   Sylvain Barde

   Computational Economics, 2017

   Abs

   The recent increase in the breath of computational methodologies has been matched with a corresponding increase in the difficulty of comparing the relative explanatory power of models from different methodological lineages. In order to help address this problem a Markovian information criterion (MIC) is developed that is analogous to the Akaike information criterion (AIC) in its theoretical derivation and yet can be applied to any model able to generate simulated or predicted data, regardless of its methodology. Both the AIC and proposed MIC rely on the Kullback–Leibler (KL) distance between model predictions and real data as a measure of prediction accuracy. Instead of using the maximum likelihood approach like the AIC, the proposed MIC relies instead on the literal interpretation of the KL distance as the inefficiency of compressing real data using modelled probabilities, and therefore uses the output of a universal compression algorithm to obtain an estimate of the KL distance. Several Monte Carlo tests are carried out in order to (a) confirm the performance of the algorithm and (b) evaluate the ability of the MIC to identify the true data-generating process from a set of alternative models.

2016

1. Direct comparison of agent-based models of herding in financial markets

   Sylvain Barde

   Journal of Economic Dynamics and Control, 2016

   Abs

   The present paper tests a new model comparison methodology by comparing multiple calibrations of three agent-based models of financial markets on the daily returns of 24 stock market indices and exchange rate series. The models chosen for this empirical application are the herding model of Gilli and Winker (2003), its asymmetric version by Alfarano et al. (2005) and the more recent model by Franke and Westerhoff (2011), which all share a common lineage to the herding model introduced by Kirman (1993). In addition, standard ARCH processes are included for each financial series to provide a benchmark for the explanatory power of the models. The methodology provides a consistent and statistically significant ranking of the three models. More importantly, it also reveals that the best performing model, Franke and Westerhoff, is generally not distinguishable from an ARCH-type process, suggesting their explanatory power on the data is similar.

2015

1. Back to the future: economic self-organisation and maximum entropy prediction

   Sylvain Barde

   Computational Economics, 2015

   Abs

   This paper shows that signal restoration methodology is appropriate for predicting the equilibrium state of certain economic systems. A formal justification for this is provided by proving the existence of finite improvement paths in object allocation problems under weak assumptions on preferences, linking any initial condition to a Nash equilibrium. Because a finite improvement path is made up of a sequence of systematic best-responses, backwards movement from the equilibrium back to the initial condition can be treated like the realisation of a noise process. This underpins the use of signal restoration to predict the equilibrium from the initial condition, and an illustration is provided through an application of maximum entropy signal restoration to the Schelling model of segregatio

2012

1. Of Ants and Voters: Maximum Entropy Prediction of Agent-Based Models with Recruitment

   Sylvain Barde

   Revue de l’OFCE, 2012

   Abs

   Maximum entropy predictions are made for the Kirman ant model as well as the Abrams-Strogatz model of language competition, also known as the voter model. In both cases the maximum entropy methodology provides good predictions of the limiting distribution of states, as was already the case for the Schelling model of segregation. As an additional contribution, the analysis of the models reveals the key role played by relative entropy and the model in controlling the time horizon of the prediction.

2011

1. Non-negativity and agglomeration behaviour of the quasi-linear logarithmic model of NEG

   Sylvain Barde, and John Peirson

   Letters in Spatial and Resource Sciences, 2011

   Abs

   It is shown that negative consumption of agricultural goods can occur in the literature using the quasi-linear logarithmic NEG model initially developed in Pflüger (Reg. Sci. Urban Econ. 34:565–573, 2004). This is because the positive consumption condition stated in the original framework is not implemented. Satisfying this condition requires a constraint on the relative size of two of the core model parameters. Importantly, this is found to modify the agglomeration behaviour of the model which affects the results of the original framework and those of ensuing contributions.

2010

1. Increasing returns and the spatial structure of French wages

   Sylvain Barde

   Spatial Economic Analysis, 2010

   Abs

   New Economic Geography presents increasing returns to agglomeration as a central explanation for concentration of economic activity. The estimation of the size of these effects remains, however, a standing issue in the field. The focus of this study is to investigate the presence of increasing returns to agglomeration in the spatial structure of wages in France, using the methodology developed by Fingleton and initially used in the UK. The central finding is the statistically significant presence of such returns to density for France, as was the case for the UK in the original study. Compared to Fingleton’s original work, it is shown that returns to density play a larger role in explaining French labour productivity, while commuting plays a smaller role than in the UK.

2. Knowledge spillovers, black holes and the equilibrium location of vertically linked industries

   Sylvain Barde

   Journal of Economic Geography, 2010

   Abs

   Using a generalized version of the Venables (1996, International Economic Review, 37: 341–359) model, this article explores the relative locations of two vertically linked sectors with knowledge spillovers. Analytical investigation shows that the dynamic properties of the Venables model are significantly affected by the presence of spillovers. In particular, the own-cost reduction effects at low transport costs can be so strong that runaway agglomeration dynamics appear in a manner consistent with the black hole concept found in the literature. However, due to the decay of information over space, these black hole dynamics are endogenous to the model and disappear when transport costs are high enough. Importantly, the location predictions obtained in simulations of the model are consistent with the empirical finding that industrials sector that benefit from spillovers are typically more agglomerated than sector that do not benefit from such spillovers.

2008

1. Agglomeration incentives in a spatial Cournot model with increasing returns to scale

   Sylvain Barde

   Letters in spatial and resource sciences, 2008

   Abs

   This paper uses a spatial Cournot competition model to examine the effect of increasing returns to scale (IRS) in manufacturing as an agglomerating force between a fixed number of firms competing over regional markets. As one would expect, with a constant returns to scale (CRS) manufacturing sector the dispersed symmetric equilibrium will always be stable. The first important result is the confirmation that even low levels of IRS provide incentives for the manufacturing firms to deviate from this symmetric equilibrium and move towards agglomeration. An interesting aspect of the results, however, is that IRS do not automatically imply an unstable symmetric equilibrium, and the threshold required to provoke instability increases with transport costs.