Volume 22, Issue 2

We extend the class of linear quantile autoregression models by allowing for the possibility of Markov-switching regimes in the conditional distribution of the response variable. We also develop a Gibbs sampling approach for posterior inference by using data augmentation and a location-scale mixture representation of the asymmetric Laplace distribution. Bayesian calculations are easily implemented, because all complete conditional densities used in the Gibbs sampler have closed-form expressions. The proposed Gibbs sampler provides the basis for a stepwise re-estimation procedure that ensures non-crossing quantiles. Monte Carlo experiments and an empirical application to the U.S. real interest rate show that both inference and forecasting are improved when the quantile monotonicity restriction is taken into account.

Housing is distinct from other financial assets, since it is a durable consumer good for households. Due to the irreversible nature of housing investment, uncertainty should be an important determinant of housing investment. From a theoretical point of view, though, this impact is ambiguous. This paper extends previous empirical work by employing the techniques of bivariate Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models in a group of forty-eight US states. In particular, we use data on housing permits as a proxy for housing investment and the house price index for the forty-eight contiguous US states and estimate bivariate GARCH models (BEKK) for each state, in order to obtain proxies of housing investment and house price uncertainty. Moreover, we use the Economic Policy Uncertainty index as an alternative measure of uncertainty. This setup allows us to test for the impact of uncertainty on housing investment growth and house price inflation and examine whether the effects differ across the different states. In general, we find that in most states uncertainty tends to increase housing investment growth and to decrease house price inflation. The cross-state differences in results may be due to variation in the degree of speculation in housing markets.

We analyse the impact of exchange rate misalignment on economic growth for a sample of emerging economies from 1970 to 2014 using a panel smooth transition regression vector error correction model. Besides, we provide a granger causality test conducted in a non-linear framework. We find that a rise in misalignment increases significantly the output in the short-run when currencies are close to equilibrium. When they are highly misaligned, the impact on growth is reduced. However, no significant impact of output on misalignment was found in the short-run. We provide evidence that misalignment granger causes the output at any given level of misalignment both in the short and long-run. Weaker granger causality was found between output and misalignment. This raises some important policy implications. Although emerging economies can use undervaluation as a growth strategy, the benefits are smaller when currencies are highly undervalued. There is, therefore, an incentive to keep exchange rates closer to their equilibrium.

In this paper, we modify the multivariate nonlinear causality test to be panel nonlinear causality test and we apply these and other existing related tests to examine the causal relationship between growth in economic policy uncertainty (EPU) and real housing returns in China and India using quarterly data from 2003:01 to 2012:04. Both panel linear and nonlinear Granger causality tests suggest the existence of only linear and nonlinear unidirectional causality relationships from growth in EPU to real housing returns in both China and India, and bivariate linear Granger causality tests suggest the existence of only linear unidirectional causality relationship from growth in EPU to real housing returns only in China. However, nonlinear bivariate Granger causality tests conclude the existence of nonlinear bidirectional causality relationships between growth in EPU and real housing returns in both China and India and cross bivariate linear and nonlinear Granger causality tests discover that there is only a linear causality relationship from Indian growth in EPU to Chinese housing returns. The results confirm the relevance of EPU data to better understand and predict the future behaviour of housing market returns in these countries.

An important assumption underlying standard threshold regression models and their variants in the extant literature is that the threshold variable is perfectly measured. Such an assumption is crucial for consistent estimation of model parameters. This paper provides the first theoretical framework for the estimation and inference of threshold regression models with measurement errors. A new estimation method that reduces the bias of the coefficient estimates and a Hausman-type test to detect the presence of measurement errors are proposed. Monte Carlo evidence is provided and an empirical application is given.

Linearity tests against smooth transition nonlinearity are typically based on the standard least-squares (LS) covariance matrix estimator. We derive an expression for the bias of the LS estimator in the presence of ARCH errors. We show that the bias is downward, and increases dramatically with the persistence of the variance process. As a consequence, conventional tests spuriously indicate nonlinearity. Next, we examine an alternative maximum likelihood approach. Our findings suggest that this approach has substantially better size properties than tests based on least-squares and heteroskedasticity-consistent matrix estimators, and performs comparably to a bootstrap technique.