Mr. David Amaglobeli, Mr. Valerio Crispolti, Ms. Era Dabla-Norris, Pooja Karnane, and Florian Misch
This paper describes a new, comprehensive database of tax policy measures in 23 advanced and emerging market economies over the last four decades. We extract this information from more than 900 OECD Economic Surveys and 37,000 tax-related news from the International Bureau of Fiscal Documentation using text-mining techniques. The innovation of this dataset lies in its granularity: changes in the rates and bases of personal and corporate income taxes, value added and sale taxes, social security contributions, excise, and property taxes are systematically documented. In addition, the database provides information on the announcement and implementation dates, whether the measures represent major changes, are part of a broader tax package, and phased in over several years. The paper also presents a range of stylized facts suggesting that information from this database is useful to deepen the analysis of tax policy changes for research and policy purposes.
This paper analyzes the experiences of emerging market economies (EMEs) that have liberalized capital flows over the past 15 years with respect to macroeconomic performance and risks to financial stability. The results of the panel data regressions indicate that greater openness to capital flows is associated with higher growth, gross capital flows, and equity returns and with lower inflation and bank capital adequacy ratios. The effects vary depending on thresholds. As a potential application of these findings, the paper explores the possible effects of liberalization on China by applying the coefficients of explanatory variables to the corresponding variables of China in 2012–16.
Huigang Chen, Mr. Alin T Mirestean, and Mr. Charalambos G Tsangarides
This paper extends the Bayesian Model Averaging framework to panel data models where the lagged dependent variable as well as endogenous variables appear as regressors. We propose a Limited Information Bayesian Model Averaging (LIBMA) methodology and then test it using simulated data. Simulation results suggest that asymptotically our methodology performs well both in Bayesian model averaging and selection. In particular, LIBMA recovers the data generating process well, with high posterior inclusion probabilities for all the relevant regressors, and parameter estimates very close to their true values. These findings suggest that our methodology is well suited for inference in short dynamic panel data models with endogenous regressors in the context of model uncertainty. We illustrate the use of LIBMA in an application to the estimation of a dynamic gravity model for bilateral trade.
This paper analyzes the relationship between oil price shocks and bank profitability. Using data on 145 banks in 11 oil-exporting MENA countries for 1994-2008, we test hypotheses of direct and indirect effects of oil price shocks on bank profitability. Our results indicate that oil price shocks have indirect effect on bank profitability, channeled through country-specific macroeconomic and institutional variables, while the direct effect is insignificant. Investment banks appear to be the most affected ones compared to Islamic and commercial banks. Our findings highlight systemic implications of oil price shocks on bank performance and underscore their importance for macroprudential regulation purposes in MENA countries.
Default prior choices fixing Zellner's g are predominant in the Bayesian Model Averaging literature, but tend to concentrate posterior mass on a tiny set of models. The paper demonstrates this supermodel effect and proposes to address it by a hyper-g prior, whose data-dependent shrinkage adapts posterior model distributions to data quality. Analytically, existing work on the hyper-g-prior is complemented by posterior expressions essential to fully Bayesian analysis and to sound numerical implementation. A simulation experiment illustrates the implications for posterior inference. Furthermore, an application to determinants of economic growth identifies several covariates whose robustness differs considerably from previous results.
Mr. Alin T Mirestean, Mr. Charalambos G Tsangarides, and Huigang Chen
Bayesian Model Averaging (BMA) provides a coherent mechanism to address the problem of model uncertainty. In this paper we extend the BMA framework to panel data models where the lagged dependent variable as well as endogenous variables appear as regressors. We propose a Limited Information Bayesian Model Averaging (LIBMA) methodology and then test it using simulated data. Simulation results suggest that asymptotically our methodology performs well both in Bayesian model selection and averaging. In particular, LIBMA recovers the data generating process very well, with high posterior inclusion probabilities for all the relevant regressors, and parameter estimates very close to the true values. These findings suggest that our methodology is well suited for inference in dynamic panel data models with short time periods in the presence of endogenous regressors under model uncertainty.
This paper develops the theoretical background for the Limited Information Bayesian Model Averaging (LIBMA). The proposed approach accounts for model uncertainty by averaging over all possible combinations of predictors when making inferences about the variables of interest, and it simultaneously addresses the biases associated with endogenous and omitted variables by incorporating a panel data systems Generalized Method of Moments estimator. Practical applications of the developed methodology are discussed, including testing for the robustness of explanatory variables in the analyses of the determinants of economic growth and poverty.
A stochastic volatility model where volatility was driven solely by a latent variable called news was estimated for three stock indices. A Markov chain Monte Carlo algorithm was used for estimating Bayesian parameters and filtering volatilities. Volatility persistence being close to one was consistent with both volatility clustering and mean reversion. Filtering showed highly volatile markets, reflecting frequent pertinent news. Diagnostics showed no model failure, although specification improvements were always possible. The model corroborated stylized findings in volatility modeling and has potential value for market participants in asset pricing and risk management, as well as for policymakers in the design of macroeconomic policies conducive to less volatile financial markets.
This paper reviews recent advances in the specification and estimation of Bayesian Vector Autoregressive models (BVARs). After describing the Bayesian principle of estimation, we first present the methodology originally developed by Litterman (1986) and Doan et al. (1984) and review alternative priors. We then discuss extensions of the basic model and address issues in forecasting and structural analysis. An application to the estimation of a system of time-varying reaction functions for four European central banks under the European Monetary System (EMS) illustrates how some of the results previously presented may be applied in practice.
Erik W. Larson, Mr. Sunil Sharma, and Mr. Lars J. Olson
This paper analyzes the stochastic inventory control problem when the demand distribution is not known. In contrast to previous Bayesian inventory models, this paper adopts a non-parametric Bayesian approach in which the firm’s prior information is characterized by a Dirichlet process prior. This provides considerable freedom in the specification of prior information about demand and it permits the accommodation of fixed order costs. As information on the demand distribution accumulates, optimal history-dependent (s,S) rules are shown to converge to an (s,S) rule that is optimal when the underlying demand distribution is known.