Determining the cointegrating rank in nonstationary fractional systems by the exact local Whittle approach

We propose to extend the cointegration rank determination procedure of Robinson and Yajima [2002. Determination of cointegrating rank in fractional systems. Journal of Econometrics 106, 217–242] to accommodate both (asymptotically) stationary and nonstationary fractionally integrated processes as the common stochastic trends and cointegrating errors by applying the exact local Whittle analysis of Shimotsu and Phillips [2005. Exact local Whittle estimation of fractional integration. Annals of Statistics 33, 1890–1933]. The proposed method estimates the cointegrating rank by examining the rank of the spectral density matrix of the d$d$th differenced process around the origin, where the fractional integration order, d$d$, is estimated by the exact local Whittle estimator. Similar to other semiparametric methods, the approach advocated here only requires information about the behavior of the spectral density matrix around the origin, but it relies on a choice of (multiple) bandwidth(s) and threshold parameters. It does not require estimating the cointegrating vector(s) and is easier to implement than regression-based approaches, but it only provides a consistent estimate of the cointegration rank, and formal tests of the cointegration rank or levels of confidence are not available except for the special case of no cointegration. We apply the proposed methodology to the analysis of exchange rate dynamics among a system of seven exchange rates. Contrary to both fractional and integer-based parametric approaches, which indicate at most one cointegrating relation, our results suggest three or possibly four cointegrating relations in the data.     

Fractional integration in daily stock market indices at Jordan's Amman stock exchange

Using daily data on five sectoral indices from 2006 to 2014, this paper aims to investigate the possibility of fractional integration in sectoral returns (and their volatility measures) at Jordan's Amman stock exchange (ASE). Empirical analysis, using the log-periodogram (LP) and local whittle (LW) based semi-parametric fractional differencing techniques suggest that all sectoral returns at ASE exhibit short memory. However, in the case of volatility measures, we found evidence of long memory. Following the recent literature that argues that structural breaks in a time series could also explain the presence of long memory, we tested the volatility measures for the presence of structural breaks. We found that long memory in some volatility measures could be attributed to the presence of structural breaks. Furthermore, using impulse response functions (IRF) based on ARFIMA, we found that shocks to sectoral returns at ASE exhibit short run persistence, whereas shocks to volatility measures display long run persistence.     

Do the black market and the official exchange rates converge in the long run?

Establishing cointegration and long-run convergence between the official and the black market exchange rates is a way of testing foreign exchange market efficiency. Earlier research employed the Engle-Granger or Johansen method to test for cointegration between the two rates. Since either method requires each rate to be non-stationary, exchange rates that did not possess this property had to be excluded from analysis. However, with the introduction of the bounds testing approach to cointegration, no exchange rate needs to be excluded since this relatively new technique does not require pre-unit root testing. This paper employs monthly data from 27 developing countries and the bounds testing approach to cointegration by Pesaran et al. (2001) to show that in 22 out of 27 countries, not only are the two rates cointegrated but in all countries the black market exchange rate causes the official rate to adjust and converge toward the black market rate in the long run.     

The distance between rival nonstationary fractional processes

Asymptotic inference on nonstationary fractional time series models, including cointegrated ones, is proceeding along two routes, determined by alternative definitions of nonstationary processes. We derive bounds for the mean squared error of the difference between (possibly tapered) discrete Fourier transforms under the two regimes. We apply the results to deduce limit theory for estimates of memory parameters, including ones for cointegrating errors, with mention also of implications for estimates of cointegrating coefficients.     

A Comparison of Johansen's,Bierens’ and the Subspace Algorithm Method for Cointegration Analysis*

The methods listed in the title are compared by means of a simulation study and a real world application. The aspects compared via simulations are the performance of the tests for the cointegrating rank and the quality of the estimated cointegrating space. The subspace algorithm method, formulated in the state space framework and thus applicable for vector autoregressive moving average (VARMA) processes, performs at least comparably to the Johansen method. Both the Johansen procedure and the subspace algorithm cointegration analysis perform significantly better than Bierens’ method. The real‐world application is an investigation of the long‐run properties of the one‐sector neoclassical growth model for Austria. The results do not fully support the implications of the model with respect to cointegration. Furthermore, the results differ greatly between the different methods. The amount of variability depends strongly upon the number of variables considered and huge differences occur for the full system with six variables. Therefore we conclude that the results of such applications with about five or six variables and 100 observations, which are typical in the applied literature, should possibly be interpreted with more caution than is commonly done.     

Asymptotic normality of narrow-band least squares in the stationary fractional cointegration model and volatility forecasting

We consider semiparametric frequency domain analysis of cointegration between long memory processes, i.e. fractional cointegration, allowing derivation of useful long-run relations even among stationary processes. The approach is due to Robinson (1994b. Annals of Statistics 22, 515–539) and uses a degenerating part of the periodogram near the origin to form a narrow-band frequency domain least squares (FDLS) estimator of the cointegrating relation, which is consistent for arbitrary short-run dynamics. We derive the asymptotic distribution theory for the FDLS estimator of the cointegration vector in the stationary long memory case, thus complementing Robinson's consistency result. An application to the relation between the volatility realized in the stock market and the associated implicit volatility derived from option prices is offered.     

Generalized long memory processes,failure of cointegration tests and exchange rate dynamics

This paper presents evidence that the equilibrium relationship in a system of nominal exchange rates is best described as a stationary GARMA process. The implementation of the GARMA methodology helps explain conflicting and puzzling results from the use of linear cointegration and fractional cointegration methods. Furthermore, we use Monte Carlo analysis to document problems with standard cointegration tests when the attraction process is distributed as a long memory GARMA process. Copyright © 2006 John Wiley & Sons, Ltd.     

TEN THINGS WE SHOULD KNOW ABOUT TIME SERIES

Abstract Time series data affect many aspects of our lives. This paper highlights 10 things we should all know about time series, namely, a good working knowledge of econometrics and statistics, an awareness of measurement errors, testing for zero frequency, seasonal and periodic unit roots, analysing fractionally integrated and long memory processes, estimating VARFIMA models, using and interpreting cointegrating models carefully, choosing sensibly among univariate conditional, stochastic and realized volatility models, not confusing thresholds, asymmetry and leverage, not underestimating the complexity of multivariate volatility models, and thinking carefully about forecasting models and expertise.     

Multiple shifts and fractional integration in the US and UK unemployment rates

This paper analyses the long-run behaviour of the US and UK unemployment rates by testing for possibly fractional orders of integration and multiple shifts using a sample of over 100 annual observations. The results show that the orders of integration are higher than zero in both series, which implies long memory. If we assume that the underlying disturbances are white noise, the values are higher than 0.5, i.e., nonstationary. However, if the disturbances are autocorrelated, the orders of integration are in the interval (0, 0.5), implying stationarity and mean-reverting behaviour. Moreover, when multiple shifts are taken into account, unemployment is more persistent in the US than in the UK, implying the need for stronger policy action in the former to bring unemployment back to its original level.     

Crucial exchange rate parity. Evidence for Mexico

Through a structural vector error correction model, one restricted cointegrating relationship for monthly data (1999.01–2012.04) was found between three exchange parities of great relevance for the Mexican economy: US Dollar–Euro, Mexican Peso–US Dollar, and Mexican Peso–Euro. The data's structure revealed endogeneity of the last one, but the first is the one that adjusts the long run (cointegrating) relation. A unitary elasticity of MxP–Euro parity to the other two parities was found, which validates PPP condition in absolute terms. These results are crucial to analyze the possible long run exchange effects on the Mexican real and financial variables because of the possible intensification of the Euro crisis and the currency war.     

Diagnostic testing for cointegration

We develop a sequence of tests for specifying the cointegrating rank of, possibly fractional, multiple time series. Memory parameters of observables are treated as unknown, as are those of possible cointegrating errors. The individual test statistics have standard null asymptotics and are related to Hausman specification test statistics: when the memory parameter is common to several series, an estimate of this parameter based on the assumption of no cointegration achieves an efficiency improvement over estimates based on individual series, whereas if the series are cointegrated the former estimate is generally inconsistent. However, a computationally simpler but asymptotically equivalent approach, which avoids explicit computation of the “efficient” estimate, is instead pursued here. Two versions of it are initially proposed, followed by one that robustifies to possible inequality between memory parameters of observables. Throughout, a semiparametric approach is pursued, modelling serial dependence only at frequencies near the origin, with the goal of validity under broad circumstances and computational convenience. The main development is in terms of stationary series, but an extension to non-stationary ones is also described. The algorithm for estimating cointegrating rank entails carrying out such tests based on potentially all subsets of two or more of the series, though outcomes of previous tests may render some or all subsequent ones unnecessary. A Monte Carlo study of finite sample performance is included.     

Cointegration in a macroeconomic system

This paper investigates the long-run relationships within a set of six quarterly time-series on the Austrian economy by means of cointegration. After analysing the univariate properties, especially with respect to the appropriate seasonal filter, the maximum-likelihood method proposed by Johansen (1988) is applied to estimate and test the cointegrating relationships. We found three such relations, implying that the system is driven by three independent stochastic time trends. In a next stage we investigate whether the empirically determined cointegrating relationships are compatible with implications derived from the neoclassical growth model with exogenous stochastic technical progress. It is found that the Austrian data strongly reject the propositions that the real interest rate and the log ratios of consumption to output, investment to output, and the real gross wage sum to output are stationary.     

Understanding the Fisher equation

It is argued that univariate long memory estimates based on ex post data tend to underestimate the persistence of ex ante variables (and, hence, that of the ex post variables themselves) because of the presence of unanticipated shocks whose short‐run volatility masks the degree of long‐range dependence in the data. Empirical estimates of long‐range dependence in the Fisher equation are shown to manifest this problem and lead to an apparent imbalance in the memory characteristics of the variables in the Fisher equation. Evidence in support of this typical underestimation is provided by results obtained with inflation forecast survey data and by direct calculation of the finite sample biases. To address the problem of bias, the paper introduces a bivariate exact Whittle (BEW) estimator that explicitly allows for the presence of short memory noise in the data. The new procedure enhances the empirical capacity to separate low‐frequency behaviour from high‐frequency fluctuations, and it produces estimates of long‐range dependence that are much less biased when there is noise contaminated data. Empirical estimates from the BEW method suggest that the three Fisher variables are integrated of the same order, with memory parameter in the range (0.75, 1). Since the integration orders are balanced, the ex ante real rate has the same degree of persistence as expected inflation, thereby furnishing evidence against the existence of a (fractional) cointegrating relation among the Fisher variables and, correspondingly, showing little support for a long‐run form of Fisher hypothesis. Copyright © 2004 John Wiley & Sons, Ltd.     

Singular value decomposition in cointegration analysis: a note regarding the difference stationary series

Many methods and techniques have been developed gradually to compute cointegration vectors. We present here a comparatively simple method for computing the matrix of cointegrating vectors, by applying singular value decomposition. With this method, one can easily accommodate in the cointegrating vectors any deterministic factors, such as a dummy, apart from the constant term and the trend. Besides the errors corresponding to the finally selected cointegrating vector have the property of minimum variance. It is noted that this procedure is not mentioned in the relevant literature. Additionally, a sequential technique is introduced, for determining the order of integration for a given series. With this procedure one can directly detect whether the differencing process produces a stationary series or not, since it seems to be a common belief that differencing a variable (one or more times) we will always get a stationary series [Harris, R.: Using Cointegration Analysis in Econometric Modelling. Prentice Hall, London (1995)]. It will be seen that this is not necessarily the case.     

Mean reversion and long memory in African stock market prices

We examine the behavior of stock market prices in several African countries by means of fractionally integrated techniques. In doing so, we can test for mean reversion in these markets. Our results can be summarized as follows: we cannot find evidence of mean reversion in any single market, and evidence of long memory returns (i.e., orders of integration above 1 in the logged stock prices) is obtained in the cases of Egypt and Nigeria, and, in a lesser extent in Tunisia, Morocco and Kenya. Permitting the existence of a structural change, the break dates take place in the earlier 2000s in the majority of the cases, and evidence of mean reversion seems to have taken place in the periods before the breaks in most of the countries. If we focus on the absolute and squared returns, evidence of long memory is obtained in Nigeria and Egypt. Thus, for these two countries, a long memory model incorporating positive fractional degrees of integration in both the level and the volatility process should be considered.     

Maximum Eigenvalue Test for Seasonal Cointegrating Ranks*

The maximum eigenvalue (ME) test for seasonal cointegrating ranks is presented using the approach of Cubadda [Oxford Bulletin of Economics and Statistics (2001), Vol. 63, pp. 497–511], which is computationally more efficient than that of Johansen and Schaumburg [Journal of Econometrics (1999), Vol. 88, pp. 301–339]. The asymptotic distributions of the ME test statistics are obtained for several cases that depend on the nature of deterministic terms. Monte Carlo experiments are conducted to evaluate the relative performances of the proposed ME test and the trace test, and we illustrate these tests using a monthly time series.     

Testing of Fractional Cointegration in Macroeconomic Time Series*

We propose in this article a two‐step testing procedure of fractional cointegration in macroeconomic time series. It is based on Robinson's (Journal of the American Statistical Association, Vol. 89, p. 1420) univariate tests and is similar in spirit to the one proposed by Engle & Granger (Econometrica, Vol. 55, p. 251), testing initially the order of integration of the individual series and then, testing the degree of integration of the residuals from the cointegrating relationship. Finite‐sample critical values of the new tests are computed and Monte Carlo experiments are conducted to examine the size and the power properties of the tests in finite samples. An empirical application, using the same datasets as in Engle & Granger (Econometrica, Vol. 55, p. 251) and Campbell & Shiller (Journal of Political Economy, Vol. 95, p. 1062), is also carried out at the end of the article.     

Price-wage dynamics in a transition economy: The case of Poland

In this paper we analyse the wage-price relationship of an economy in transition characterized by important structural changes. It is known (see Perron, 1989) that structural breaks in stationary time series can induce apparent unit roots. The stationarity analysis of the series employed in the present model is conducted jointly with the assumption that the breakpoint location is unknown. We follow a testing procedure recently proposed by Zivot and Andrews (1992). Cointegration analysis of wages and prices in the presence of structural breaks finds empirical evidence in favour of two cointegrating vectors involving prices and wages. Our analysis focuses on the different structural behaviour of the price-wage dynamic relationship in the short and long term; we also demonstrate the relative importance of import prices as a source of wage-price fluctuations.     

A Note On Long Memory Time Series

This note presents the fractional integrated processes which are the main models used to describe long memory phenomena.¹ Section 1 briefly defines the concept of fractional integration, shows the fundamental properties and provides a short summary of the estimation methods. Section 2 consists of a survey of their extensions in order to model long-term cycles.     

Exchange rate determination: Empirical evidence for the Greek drachma

We examine the monetary approach to the exchange rate from several perspectives, using monthly data on five Greek bilateral exchange rates. Using the Johansen multivariate cointegration technique we demonstrate that the monetary model is validate as a long-run relationship. Moreover, upon imposing the long-run monetary model restrictions we find that the proportionality of the exchange rate to relative monetary aggregates is accepted for the bilateral exchange rate of the Greek drachma with respect to the US dollar, the deutschemark, the pound sterling, the French franc and the Italian lira. Finally, using the Hansen-Johansen (1993) testing procedure, we reject the hypothesis of sample independency of the dimension of the cointegration space in all cases, apart from the German one, but we accept the hypothesis that the estimated coefficients do not display instabilities in recursive estimations.