Bayesian Updating of Input-Output Tables

Oleg Lugovoy, Andrey Polbin and Vladimir Potashnikov

No 332664, Conference papers from Purdue University, Center for Global Trade Analysis, Global Trade Analysis Project

Abstract: The paper continues efforts on developing Bayesian method of updating IO tables, presented by the authors on the 16th Annual Conference on Global Economic Analysis, and extends the methodology and results in several ways. In the current paper, we test our methodology on the “long” survey based IRIOS tables. We compare two point estimates of the Bayesian method of “unknown” IO table: posterior mode and posterior mean with estimates, which come from alternative methods popular in the literature. Than we discuss how to construct an appropriate creditable set for IO coefficients. We also upgrade and extend estimates of SUT tables for Russia. The work consists of three parts. The first part of the paper devoted to conceptual framework for updating, disaggregating and balancing IO tables. Compared to previous paper we improve our sampling methodology by using conjugate vector of Hessian matrix of prior distribution to avoid high autocorrelation of MCMC chains. In addition we use eigen values for proposal density for MCMC algorithm. This technique provides good convergence properties of Markov chains. In the second part of the current paper, we test our methodology on the “long” survey based IRIOS tables (van der Linden and Oosterhaven, 1995). We treat the last table for each country as unknown and estimate it with the Bayesian method using all previously available matrixes for constructing prior distribution. We consider two point estimates of “unknown” IO table: posterior mode and posterior mean. To find posterior mode we use nonlinear optimization techniques, to explore posterior distribution we use proposed MCMC method. Posterior mode robustly outperforms competitive methods, popular in the literature, according to different closeness statistics. Posterior mean perform slightly worse than posterior mode. We conclude that point estimate of Bayesian method at least is compatible with the other methods on real data examples. The main contribution of our method i...

Keywords: Research; Methods/Statistical; Methods (search for similar items in EconPapers)
Date: 2015
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://ageconsearch.umn.edu/record/332664/files/7347.pdf (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:ags:pugtwp:332664

Access Statistics for this paper

More papers in Conference papers from Purdue University, Center for Global Trade Analysis, Global Trade Analysis Project Contact information at EDIRC.
Bibliographic data for series maintained by AgEcon Search ().