Band 7, Heft 3

In this paper we propose a unified framework for testing serial independence against time irreversibility. This framework extends the time reversibility tests of Ramsey and Rothman(1996, Journal of Money, Credit and Banking) and Chen, Chou, and Kuan(2000,Journal of Econometrics) in several important directions. It consists of a pair of original-series-based individual and portmanteau tests and the corresponding model diagnostic checks. The former can be used to detect asymmetries in time series, and the latter can be applied to check if such asymmetries are properly explained by an econometric model. A Monte Carlo simulation shows that the proposed tests perform suitably in finite samples. An empirical example further shows that the proposed method is a useful complement to the conventional independence tests for checking the adequacy of different GARCH models.

This study investigates the synchronization of business cycles in different sectors of the economy. Business cycles in different industries or sectors have a tendency to synchronize with one another in what appears to be a national business cycle, yet trade between sectors may not be strong enough for one sector to "drive" business cycle fluctuations in another sector. How are these sectors synchronized? This study suggests that the national business cycle results from a "mode-locking" phenomenon between different sectors, a nonlinear process through which weak coupling between oscillating systems (sectors) tends to synchronize the fluctuations between the oscillating systems (sectors). Simulations, statistical analysis, and spectral analysis are used to attempt to verify this hypothesis. Investigation reveals a moderate amount of econometric support for the sectoral mode-locking hypothesis of business cycle formation.

It has been argued by several authors that the inflationary dynamics in Brazil follow a unit root process, thus displaying some inertia. Indeed, Cati, et al. (Journal of Applied Econometrics, 1999) have found that the inflationary dynamics in Brazil are nearly fully inertial. We estimate the fractional differencing parameter using an ARFIMA specification for the inflation rate in that country and our results suggest that the inflationary dynamics are better modeled by a long memory process than by a unit root mechanism, thus implying that there is no inertia in inflation, contrary to what has been found by other researchers. We also found that the estimates of the fractional parameter are invariant to first differencing.

In this paper we investigate the profitability of non-linear trading rules based on nearest neighbour (NN) predictors. Applying this investment strategy to the New York Stock Exchange for the 1997-2002 period, our results suggest that, taking into account transaction costs, the NN-based trading rule is superior to both a risk-adjusted buy-and-hold strategy and a linear ARIMA-based strategy in terms of returns for all of the years studied, except for 2000 and 2001. In addition, the NN-based trading rule produces higher net returns than those from a simple buy-and-hold strategy, except for 1997. Regarding other profitability measures, the NN-based trading rule yields higher Sharpe ratios than the ARIMA-based strategy for all of the years in the sample except for 2001. As for 2001, in 36 out of the 101 cases considered, the ARIMA-based strategy gives higher Sharpe ratios than those from the NN-trading rule, in 18 cases the opposite is true, and in the remaining 36 cases both strategies yield the same ratios.

The attractive possibility that financial indices may be chaotic has been the subject of much study. In this paper we address two specific questions: "Masked by stochasticity, do financial data exhibit deterministic nonlinearity?", and "If so, so what?". We examine daily returns from three financial indicators: the Dow Jones Industrial Average, the London gold fixings, and the USD-JPY exchange rate. For each data set we apply surrogate data methods and nonlinearity tests to quantify determinism over a wide range of time scales (from 100 to 20,000 days). We find that all three time series are distinct from linear noise or conditional heteroskedastic models and that there therefore exists detectable deterministic nonlinearity that can potentially be exploited for prediction.