An algorithm for constructing high dimensional distributions from distributions of lower dimension

We propose a new sequential procedure for estimating multivariate distributions in cases when conventional maximum likelihood has too many parameters and is therefore inaccurate or non-operational. The procedure constructs a multivariate distribution and its pseudo-likelihood sequentially, in each step using lower-dimensional distributions with a small number of parameters. In an application, the procedure provides excellent fit when the dimension is moderate, and remains operational when the conventional method fails.     

An empirical behavioral model of household’s deposit dollarization

We use the behavioral concept to endogenously model the evolution of the link between households’ deposit dollarization and exchange rate developments in Russia. We estimate the model empirically and show that the reaction of households to exchange rate appreciation weakens when exchange rate developments become more volatile. The proposed model outperforms the contemporary nonlinear time series models in forecasting the changes in dollarization during the Bank of Russia’s transition to a flexible exchange rate regime.