Efficient learning of decomposable models with a bounded clique size

The learning of probability distributions from data is a ubiquitous problem in the fields of Statistics and Artificial Intelligence. During the last decades several learning algorithms have been proposed to learn probability distributions based on decomposable models due to their advantageous theoretical properties. Some of these algorithms can be used to search for a maximum likelihood decomposable model with a given maximum clique size, k, which controls the complexity of the model. Unfortunately, the problem of learning a maximum likelihood decomposable model given a maximum clique size is NP-hard for k > 2. In this work, we propose a family of algorithms which approximates this problem with a computational complexity of O(k · n^2 log n) in the worst case, where n is the number of implied random variables. The structures of the decomposable models that solve the maximum likelihood problem are called maximal k-order decomposable graphs. Our proposals, called fractal trees, construct a sequence of maximal i-order decomposable graphs, for i = 2, ..., k, in k − 1 steps. At each step, the algorithms follow a divide-and-conquer strategy based on the particular features of this type of structures. Additionally, we propose a prune-and-graft procedure which transforms a maximal k-order decomposable graph into another one, increasing its likelihood. We have implemented two particular fractal tree algorithms called parallel fractal tree and sequential fractal tree. These algorithms can be considered a natural extension of Chow and Liu’s algorithm, from k = 2 to arbitrary values of k. Both algorithms have been compared against other efficient approaches in artificial and real domains, and they have shown a competitive behavior to deal with the maximum likelihood problem. Due to their low computational complexity they are especially recommended to deal with high dimensional domains.

Inequality for Wage Earners and Self-Employed: Evidence from Panel Data

Published as an article in: Oxford Bulletin of Economics and Statistics, 2009, vol. 71, issue 4, pages 491-518.

Digital Quantum Simulation of Spin Models with Circuit Quantum Electrodynamics

Systems of interacting quantum spins show a rich spectrum of quantum phases and display interesting many-body dynamics. Computing characteristics of even small systems on conventional computers poses significant challenges. A quantum simulator has the potential to outperform standard computers in calculating the evolution of complex quantum systems. Here, we perform a digital quantum simulation of the paradigmatic Heisenberg and Ising interacting spin models using a two transmon-qubit circuit quantum electrodynamics setup. We make use of the exchange interaction naturally present in the simulator to construct a digital decomposition of the model-specific evolution and extract its full dynamics. This approach is universal and efficient, employing only resources that are polynomial in the number of spins, and indicates a path towards the controlled simulation of general spin dynamics in superconducting qubit platforms.

Estrategias de diversificación en los grupos familiares y no familiares: principales diferencias según el tipo de grupo no familiar

[ES] Durante la última década surge un interés por el estudio de la estructura de propiedad como elemento determinante de la diversificación. Sin embargo, existe una carencia de investigaciones que analicen la influencia de la naturaleza del último propietario en el nivel y tipo de diversificación. Por ello, el objeto del presente trabajo es analizar las estrategias de diversificación empleadas por los grandes grupos empresariales españoles cuya empresa matriz cotiza en los mercados de valores, estudiando las diferencias existentes entre grupos familiares y no familiares, y considerando en estos últimos la naturaleza del último propietario. Se parte de una muestra de noventa y nueve grupos empresariales, donde se identifican las compañías que constituyen el grupo empresarial, siendo empleadas como metodologías econométricas los modelos logísticos binomiales y los modelos datos panel. Los resultados muestran como la naturaleza familiar del grupo influye positivamente en la especialización y en el empleo de estrategias de diversificación relacionada, y negativamente en el empleo de estrategias de diversificación no relacionada. Los grupos familiares difieren en mayor medida de aquellos grupos no familiares donde no existe un accionista de referencia que pueda ejercer el control efectivo del grupo y la dispersión de la propiedad es mayor, los denominados grupos sin control efectivo . La investigación permite profundizar en el análisis de las diferencias existentes entre grupos familiares y no familiares, y más concretamente en el ámbito de las estrategias de crecimiento, considerando la naturaleza del último propietario de los grupos no familiares.

Data Revisions in the Estimation of DSGE Models

Revisions of US macroeconomic data are not white-noise. They are persistent, correlated with real-time data, and with high variability (around 80% of volatility observed in US real-time data). Their business cycle effects are examined in an estimated DSGE model extended with both real-time and final data. After implementing a Bayesian estimation approach, the role of both habit formation and price indexation fall significantly in the extended model. The results show how revision shocks of both output and inflation are expansionary because they occur when real-time published data are too low and the Fed reacts by cutting interest rates. Consumption revisions, by contrast, are countercyclical as consumption habits mirror the observed reduction in real-time consumption. In turn, revisions of the three variables explain 9.3% of changes of output in its long-run variance decomposition.

Long-Term Growth and Persistence with Endogenous Depreciation: Theory and Evidence

Previous research has shown a strong positive correlation between short-term persistence and long-term output growth as well as between depreciation rates and long-term output growth. This evidence, therefore, contradicts the standard predictions from traditional neoclassical or AK-type growth models with exogenous depreciation. In this paper, we first confirm these findings for a larger sample of 101 countries. We then study the dynamics of growth and persistence in a model where both the depreciation rate and growth are endogenous and procyclical. We find that the model s predictions become consistent with the empirical evidence on persistence, long-term growth and depreciation rates.

Methods for Measuring Aggregate Costs of Conflict

This paper reviews the methods for measuring the economic cost of conflict. Estimating the economic costs of conflict requires a counterfactual calculation, which makes this a very difficult task. Social researchers have resorted to different estimation methods depending on the particular effect in question. The method used in each case depends on the units being analyzed (firms, sectors, regions or countries), the outcome variable under study (aggregate output, market valuation of firms, market shares, etc.) and data availability (a single cross-section, time series or panel data). This paper reviews existing methods used in the literature to assess the economic impact of conflict: cost accounting, cross-section methods, time series methods, panel data methods, gravity models, event studies, natural experiments and comparative case studies. The paper ends with a discussion of cost estimates and directions for further research.