Volume 11, Issue 3

This paper extends the Fractionally integrated GARCH (FIGARCH) model by incorporating Normal Inverse Gaussian Distribution (NIG). The proposed model is flexible and allows one to model time-variation, long memory, fat tails as well as asymmetry and skewness in the distribution of financial returns. GARCH and FIGARCH models for daily log exchange rate returns with Normal, Student's t and NIG error distributions as well as GARCH/FIGARCH-in-mean models with t errors are estimated and compared both in terms of sample fit as well as out-of-the-sample predictive ability in several dimensions. The FIGARCH model with symmetric and asymmetric NIG errors outperform alternatives both in-sample fit and 1-day and 5-day ahead predictions of the quartiles of the exchange rate return distributions.

This paper considers the problem of testing for and dating changes (at unknown points) in the order of integration of a time series between different trend-stationary and difference-stationary regimes. While existing procedures in the literature are designed for processes displaying only a single such change in persistence, our proposed methodology is also valid in the presence of multiple changes in persistence. Our procedure is based on sequences of doubly-recursive implementations of the regression-based unit root statistic of Elliott et al. (1996). The asymptotic validity of our procedure is demonstrated analytically. We use Monte Carlo methods to simulate both finite sample and asymptotic critical values for our proposed testing procedure and to simulate the finite sample behaviour of our procedure against a variety of single and multiple persistence change series. The procedure is shown to work well in practice. The impact of deterministic level and trend breaks on our procedure is also discussed. An empirical application of the procedure to interest rate data is considered.

In this paper we analyze the dynamics shown by the neoclassical one-sector growth model with differential savings as in Bohm and Kaas (2000) while assuming CES production function and the labour force dynamic described by the Beverton Holt equation (see Beverton and Holt, 1957). The resulting dynamic system is bidimensional, autonomous and triangular: we investigate its qualitative and quantitative dynamic properties. The study herewith presented aims at confirming that the system can exhibit cycles or even a chaotic dynamic pattern, if shareholders save more than workers, when the elasticity of substitution drops below one (so that capital income declines). The analytical results are supplemented by numerical experiments.

We estimate real US GDP growth as a threshold autoregressive process, and construct confidence intervals for the parameter estimates. However, there are various approaches that can be used in constructing the confidence intervals. We construct confidence intervals for the slope coefficients and the threshold using asymptotic results and bootstrap methods, finding that the results for the different methods have very different economic implications. We perform a Monte Carlo experiment to evaluate the various methods. Surprisingly, the confidence intervals are wide enough to cast doubt on the assertion that the time-series responses of GDP to negative growth rates are different than the responses to positive growth rates.

Measuring country exchange rates relative to a common reference basket results in a set of no-arbitrage prices, unlike trade-weighted indexes, the usual method of comparing country exchange rate histories. The reference basket is analogous to a portfolio, and its choice can be resolved by drawing on required economic interpretations or uses. We use currency reference rates to examine the historical variability of different currencies over designated cyclical bands. The temporal decompositions used are those provided by wavelet analysis, which is light on maintained assumptions about data generating processes. Some countries, notably Japan and New Zealand do exhibit a powerful but irregular medium term cycle, while others are much more stable. Implications are briefly examined for investment, hedging, monetary policy and common currency studies.

The large decline in output volatility experienced by most industrialized countries in the last decades has been thoroughly analyzed using standard time and frequency domain methods. In this paper we investigate the issue of moderation of volatility in G-7 economies and its sources, applying a multi-scaling approach to the industrial production indices of G-7 countries between 1961:1-2006:10. Using the MODWT estimates of wavelet variance we provide a scale-based analysis of variance that allows us to characterize the decline in volatility and to detect the importance of the various explanations of the moderation. The main scale-by-scale results stemming from multi scale analysis of variance are: i) a reduction in volatility which, although displayed by all the G-7 countries, is not uniform across time scales (as the decline is larger at short-term scales than at business cycle scales for France and Italy, and quite uniform across scale for the UK and the US) nor countries (as the decline is significant for a subset of countries only, i.e. France, Italy, the UK and the US); and ii) the moderation has to be attributable to the decline in the variance of both common (in the 1970s) and country-specific (in the 1960s) exogenous disturbances hitting the economy.