Multicollinearity often leads to inconsistent and unreliable parameter estimates in regression analysis. This situation will be more severe in the presence of outliers it will cause fatter tails in the error distributions than the normal distributions. The wellknown procedure that is robust to multicollinearity problem is the ridge regression method. This method however is expected to be affected by the presence of outliers due to some assumptions imposed in the modeling procedure. Thus, the robust version of existing ridge method with some modification in the inverse matrix and the estimated response value is introduced. The performance of the proposed method is discussed and comparisons are made with several existing estimators namely, Ordinary Least Squares (OLS), ridge regression and robust ridge regression based on GM-estimates. The finding of this study is able to produce reliable parameter estimates in the presence of both multicollinearity and outliers in the data.

Multicollinearity and outliers are often leads to inconsistent and unreliable parameter estimates in regression analysis. The well-known procedure that is robust to multicollinearity problem is the ridge regression method. This method however is believed are affected by the presence of outlier. The combination of GM-estimation and ridge parameter that is robust towards both problems is on interest in this study. As such, both techniques are employed to investigate the relationship between stock market price and macroeconomic variables in Malaysia due to curiosity of involving the multicollinearity and outlier problem in the data set. There are four macroeconomic factors selected for this study which are Consumer Price Index (CPI), Gross Domestic Product (GDP), Base Lending Rate (BLR) and Money Supply (M1). The results demonstrate that the proposed procedure is able to produce reliable results towards the presence of multicollinearity and outliers in the real data.

This study applies a second-generation panel unit-root tests to determine the stochastic properties of real exchange rates for 14 Asian countries. Based on three popular alternative definitions of a currency crisis, we identify the several important currency crisis episodes in the region. The purchasing power parity (PPP) hypothesis was overwhelmingly supported after accommodating these heterogeneous noisy and unstable observations. Our panel unit-root test that controls for cross-sectional dependence and is robust to structural breaks confirms that the crisis in all the countries fits well with the second-generation models of currency crisis, that is, the root cause of the currency crises may not lie in economic fundamentals. PPP relation emerges when breaks and cross country dependency has been taken into account for these 14 countries.

The panel data modeling has received a great attention in econometric research recently. This is due to the availability of data sources and the interest to study cross sections of individuals observed over time. However, the problems may arise in modeling the panel in the presence of cross sectional dependence and outliers. Even though there are few methods that take into consideration the presence of cross sectional dependence in the panel, the methods may provide inconsistent parameter estimates and inferences when outliers occur in the panel. As such, an alternative method that is robust to outliers and cross sectional dependence is introduced in this paper. The properties and construction of the confidence interval for the parameter estimates are also considered in this paper. The robustness of the procedure is investigated and comparisons are made to the existing method via simulation studies. Our results have shown that robust approach is able to produce an accurate and reliable parameter estimates under the condition considered.

This paper investigates the stationarity behavior of the ex-post real interest rates (RIRs) for 12 Asian countries. Formal tests conducted indicate that high persistence is an intrinsic characteristic in the majority of the RIRs. We consider local-persistent model to assess the degree of persistence in these series. The findings from this devise reveal that RIRs are persistent, but are characterized by a mean-reverting process. The consistency of the persistence comparable amongst the series is confirmed after accounting for the volatility of the consumption growth. Applying a test proposed by Leybourne et al. (2007a) that allows for long memory dynamics, we reconfirm the characteristic of the series. Building on previous studies, this paper provides favorable support for the long-run Fisher hypothesis and help to solve an intertemporal consumption behavior puzzle across the emerging and advanced countries. Finally, the results for the G-7 countries are presented for comparison. (C) 2017 Elsevier B.V. All rights reserved.