HP-Filter and Wavelets Analysis Modelling in Financial Markets

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: editor@ijaiem.org, editorijaiem@gmail.com Volume 2, Issue 1, January 2013 ISSN 2319 - 4847
HP-Filter and Wavelets Analysis Modelling In Financial Markets

Mustapha DJENNAS1, Mohamed BENBOUZIANE2Meriem DJENNAS3
1
University of Tlemcen, MECAS Laboratory (Algeria), BP 226, Tlemcen (13000). University of Tlemcen, MIFMA Laboratory (Algeria), BP 226, Tlemcen (13000)
3
University of Picardie Jules Verne, Chemin du Thil, 80025 Amiens (France) Corresponding Author: Mustapha DJENNAS
ABSTRACT
In this paper, we use the HP-Filter and Wavelet transformation (WT) to provide a detailed characterization of the exchange rates functioning between two currencies. In addition, we introduce a comparison approach between the cyclical component filters and wavelets transform.The analyses are conducted by introducing the main and the most used variable in the literature, the exchange rate, in order to understand how it could evolve both in time and scale, depending on the difference phases of the system construction. We will apply these concepts in the worldwide largest operating financial market, namely the Euro against U.S. Dollar foreign exchange market.
Keywords: HP-Filter Wavelets Exchange Rate.
1. INTRODUCTION
For many years, there was a very thorough debate on stocks markets and their relationship with financial crises, or even macro-financial vulnerabilities. As the foreign exchange markets is widely related to financial trade in both theoretical and empirical literature, and despite the existence of many divergent points of view, there was still a large consensus about the importance of studying foreign exchange markets as a necessary condition of understanding financial trade. In this paper, well try to investigate the importance of some statistical and heuristic tools for modeling the variation in a foreign exchange market, namely the Hodrick-Prescott (HP) Filter, the Stationary Wavelets Transformation (SWT), the Wavelets crossing transform(XWT) and the wavelets coherencetransform (WCT). The present paper has many important objectives embodied in three fundamental axes. First, our objective is to demonstrate at what degree the statistical or the heuristic approach could useful in modeling a financial time series data. The analysis is extended to understand the comparison between the two approaches. We propose the use of a cyclical signal and a denoised signal of the variable under study.Second, well try to give an answer to the following two questions: why inside an exchange market, the analysis may differ from a study to another? Third, we try to introduce a wavelet coherence model based on wavelets analysis to evaluate the two proposed approaches by applying a simulation in the Euro-U.S Dollar exchange market.
2. FOREIGN EXCHANGE RATE

In finance, it is usual to use exchange rate between two currencies as the rate at which one currency will be exchanged for another. Exchange rate is considered as a very important variable in analyzing foreign trade and financial markets operations. We have selected exchange rate as a studied variable for several reasons since the changes in the exchange rate affect widely the economic health of a country. For example, the falling exchange rate of the USD against Euro meant that the cost of importing goods from Euro area increased. This has implications for the business in terms of profitability and pricing in Euro area. In contrary, for a Euro business buying goods from abroad, the weakening euro rates would have the opposite effect with less profitability. For these reasons, understanding exchange rates fluctuations and processes may be very meaningful to traders. Moreover, exchange rate can be used as a very critical determinant in forecasting financial crises. Considered as the two most common currencies in the world, Euro-USD exchange rate is used in this study to understand how we can smooth its time series data by using filters and wavelets in order to
Volume 2, Issue 1, January 2013
Page 66

understand more deeply the relation between the two currencies. Figure 1 present the EU-USD exchange rate evolution across the time in a daily time series data starting from third January 2000 to 31th December 2012 with 3392 observations:
1.8 EU-USD ex change rate 1.7
1.6 1.5 1.4
1.3 1.2
1.1 1 0.9 0.8 2000
2001
2002
2003
2004
2005
2006 2007 Time
2008
2009
2010
2011
2012
Figure 1. EU-USD exchange rate original data signal (S)
3. HODRICK-PRESCOTT FILTER (HP-FILTER)

Usually, when we use a filter in the economic cycles theory, we aim to the study the cyclical component of a time series data. In the case of foreign exchange market, the cyclical component can be obtained by detrending (extract the cyclical component) the exchange rate. The objective is to detrend the output series and estimate its cyclical component. There is different methods for detrending. In an econometric language, they are called filters. The most used ones are: The Butterworth filter (1930), the Band-Pass (BP) filter proposed by Baxter and King (1995), the Hodrick-Prescott filter (HP) (1997), and Christiano-Fitzgerald filter (2003). Hodrick-Prescott (HP) filter was proposed with an appropriate smoothness parameter . According to Ravn and Uhlig (2002), the Hodrick-Prescott filter produces cyclical components comparable to those obtained by the Band-Pass filter. The Band-Pass (BP) filter, evaluated by Stock and Watson (1999) and the Christiano-Fitzgerald filter are often substituted by the HP filter since they have the same properties. Therefore, while minor differences among the results obtained by the four filters are not difficult to detect, the main characteristics are remarkably similar. For this reason, we used the HP-Filter in order to extract the cyclical component of the exchange rate series data due to its popularity. The basic idea is to decompose the economic series of interest into the sum of a slowly-evolving secular trend and a transitory deviation from it which is classified as cycle:
(1)
However, as these constituent parts (trend and cycle) are not readily observed, any decomposition must necessarily be built on a conceptual artifact. Thus, any trending method, must start out by somehow arbitrarily defining what shall be counted as trend and as cycle, before these elements can be estimated from the data.The HP filter extracts the trend by solving the following standard-penalty program:
(2)
Where the smoothing parameter controls the smoothness of the adjusted trend series, approximates the actual series, , while as the trend becomes linear.
, as
, the trend
While Hodrick and Prescott (1997) suggest values for , Marcet and Ravn (2003) recast the formula (2) as a constrained minimization program to determine the value of endogenously. For annual data, should be between 6 and 7, (Ravn and Uhlig, 2002; Maravall, 2004). Note that the HP formula (2) can be written more succinctly as:
(3)
Page 67

Where , and ) is the standard differencing operator and B is the standard backshift (lag) operator, such that . Also define the forward shifting operators:
Equation (3) indicates that the HP filter attempts to maximize the fit of the trend to the series (i.e. minimize the cycle component in (1)) while minimizing the changes in the trends slope. Figures 2a & 2b represent the EU-USD foreign exchange rate decomposition to a cyclical part (figre2a) and a trend part (figure 2b).
1.6
0.08
Trend c om ponent of EU-USD exc hange rate
1.5 1.4 1.3 1.2 1.1 1 0.9 0.8 2000
0.06
cyclical component of EU-USD exchange rate
0.04
0.02
-0.02
-0.04
-0.06 2000
2001
2002
2003
2004
2005
2006 2007 Time t
2008
2009
2010
2011
2012
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Figure 2a. HP cyclical component of EUUSD exchange rate
Figure 2b. HP trend component of EU-USD exchange rate
4. STATIONARY WAVELET TRANSFORM (SWT)

Undoubtedly seen as a subject in progression, the application of wavelet theory in economics and finance is still in its beginning since wavelets models are not yet well explored in economic and finance literature. Nevertheless, there is a growing interest in applying wavelet theory to deeply understand economic reality. The following works are considered among the most important in relation with economic cycles theory: Raihan, Wen and Zeng (2005); Crowley and Lee (2005); Crowley, Maraun and Mayes (2006); Gallegati and Gallegati (2007); Yogo (2008); Aguiar-Conraria and Soares (2009). Some of the most fields in which SWT can be used are signal de-noising and pattern recognition. As argued by Bradley (2003), one of the most problems in using the discrete wavelets transformation (DWT) is the resulting shift-variance from the down-sampling process. Therefore, it is possible to skip the down-sampling process by running a stationary wavelets transformation (SWT). SWT is similar to the DWT except that in SWT, the signal is never sub-sampled and instead, the signal is up-sampled at each level of decomposition. Therefore, SWT is shift-invariant and this is a very important condition in studying volatile time series (Torrence and Compo, 1998). In addition, the main difference in the de-noising process in SWT in comparison with DWT is that in DWT only the approximation coefficients are decomposed, while in the SWT, both the detail and approximation coefficients are decomposed. Since SWT can overcome two major shortcomings of the DWT, it is seen as an appropriate wavelets model to get a more complex and flexible analysis. So, why we should denoise a signal? Denoising (also referred to as wavelet shrinkage) is to remove noise as much as possible while preserving useful information as much as possible. The basic noisy signal model as proposed by Guoxiang and Ruizhen (2001) take the following form:
(4)
Where s(x) is the observed signal, f(x) is the original signal, (x) is Gaussian white noise with zero mean and variance 2. The objective of denoising is to suppress the noise part of the signal s and to recover f. The principle idea behind SWT de-noising is that one can define a noise threshold such that variations in the data below the threshold are to be regarded as noise, whereas variations greater than the threshold are regarded as signal. De-noising process is very benefic in the context of models with regime shifts and other forms of discontinuities or points of non-differentiability (Torrence and Compo, 1998). In others words, as the noise in a signal is mostly contained in the details of wavelet coefficients, that is, the high frequency range of the signal (Keinert, 2004), if we set the small coefficients to zero, much of the noise will disappear, and of course, inevitably some minor features of the signal will be removed as well or at least, distorted by the process. The denoising procedure can be done in three steps:
Page 68

1. Select a wavelet and a level n, apply wavelet/wavelet packet decomposition to the noisy signal to produce wavelet coefficients. 2. For each level from 1 to n, choose a threshold value and apply thresholding to the detail coefficients. 3. Perform wavelet/wavelet packet reconstruction to obtain a denoised signal. The most widely-used thresholding methods are hard-thresholding:
(5)
And soft-thresholding (Donoho and Johnstone, 1998; Donoho, 1995):

(6)
Where can be calculated by Stein's Unbiased Risk Estimate method:

(7)
Here, n is the length of the signal. In this study, we used the soft-thresholding approach, because it has been reported that the soft-thresholding is more effective than the hard-thresholding (Gnanadurai and Sadasivam, 2006; Talukder and Harada, 2007). Figures 3a and 3b show the decomposition process of the EU-USD foreign exchange rate into residuals (noise) and denoised signal Ds.
0.2
1.6
0.15
1.5
0.1
1.4
0.05
1.3
1.2
0
1.1
-0.05
1
-0.1
0.9
-0.15 2000
2001
2002
2003
2004
2005
2006 2007 Time
2008
2009
2010
2011
2012
0.8 2000
2001
2002
2003
2004
2005
2006 2007 Time
2008
2009
2010
2011
2012
Figure 3a. SWT residuals of EU-USD exchange rate (S Ds)
Figure 3b. SWT De-noised signal Dsof EUUSD exchange rate
5.
WAVELETS CROSSING AND WAVELETS COHERENCE
The cross wavelets transform (XWT) and wavelets coherence (WCT) are two other wavelets models allowing analyzing the temporal evolution of the frequency content of a given signal or timing series. The application of XWT and WCT to two time series and the cross examination of the two decompositions can reveal localized similarities in time and scale. Areas in the time-frequency plane where two time series exhibit common power or consistent phase behavior indicate a relationship between the signals (Cohen and Walden, 2010a). In our case, these two models are very appropriate to compare exchange rate decomposition by HP-Filter and SWT. In the following two section, the cross wavelets and wavelets coherence models are presented according to the works of Torrence and Compo (1998), Torrence and Webster (1998), and Grinsted et al. (2004). Cross Wavelet Transform XWT Wavelets crossing and wavelets coherence are an extension of the Fourier Coherency Transform. The latter was often used to identify common frequency brands between two time series. Therefore, it is possible to develop a wavelet coherency which could identify both frequency bands and time intervals when the time series were related (Liu, 1994). Unfortunately, in Fourier analysis, it is necessary to smooth the cross spectrum before calculating coherency which is otherwise identically equal to 1. As a result, the used smoothing process in cross-wavelet spectrum was unclear and inadequate to define an appropriate wavelet coherency (Liu, 1994).
Page 69

To avoid this shortcomings situation, the wavelet coherency is used to maintain a smoothing process in both time and scale, with the amount of smoothing dependent on both the choice of wavelet and the scale. The cross wavelet transform (XWT) of two time series Xn and Yn with wavelet transforms defined as: and is
Where * is the complex conjugate of (s), n is the time index and s is the scale. The cross-wavelet spectrum is complex, and hence one can define the cross-wavelet power as . The complex argument can be interpreted as the local relative phase between Xn and Yn in time frequency space. The theoretical cross-wavelet distribution of two time series with theoretical Fourier spectra and is given in Torrence and Compo (1998) as:
(8)
Where the confidence level associated with probability p, X and Y isare the respective standard deviations. If = 1 (real wavelets), Z1(95%) = 2.182, while if = 2 (complex wavelets), Z2(95%) = 3.999. According to figures 4a and 4b, most exchange rate periods have expressed a relative important correlation when we use the HP-Filter or stationary wavelets transform. The weakest similarities are likely common to the period slightly before 1999 (Euro adoption). As in (Torrence and Compo, 1998) and (Grinsted et al, 2004), both XWT and WCT figures have some decisive criteria that may be respected in the results interpretation. In the XWT figures, the 5% significance level against red noise is shown as a thick contour. The black contour designates the 5% significance level estimated by Monte Carlo simulations beta surrogate series. The cone of influence, which indicates the region affected by edge effects, is shown with a thin black line. The relative phase relationship is shown as arrows (with in-phase pointing right, anti-phase pointing left. The color code for power ranges from blue (low power) to red (high power).
Figure 4b. Cross wavelet transform (XWT) Figure 4a. Cross wavelet transform (XWT) For the HP trend component and SWT Ds For the HP cyclical component and SWT residuals Therefore, we can see for example all the time periods where the dollar was relatively weak against the euro and vice versa. Asshowed, the HP-Filter and SWT have both predict fluctuations in the exchange rate, and that the two methods are relatively close. By cons, it can be noted in the two figures that when considering the HP cyclical component in parallel with SWT residuals (figure 4a), we hope have more accurate and refined analysis compared to the HP trend component and the SWT denoised signal (figure 4b). However, more convergence is observed during unstable periods. For example, during the recent economic crisis (late 2008 and 2009), in Figure 4a, we see clearly the degree of divergence between HP cyclical componentmatrix and SWT residuals matrix. Wavelets Coherence Transform (WCT) As the cross wavelet power is used to reveal areas with high common power, the cross wavelets coherence transform is a second useful technique which can be adapted to evaluate coherency in time frequency space. According to Torrence and Webster (1998), the wavelet coherence of two time series is given by the following formula:
Page 70

(9)
Where S is a smoothing operator in both time and scale. Here, the coherency parameter is always included between 0 and 1, (0 ). Hence, wavelet coherence is often seen as a localized correlation coefficient in time frequency space. It is an accurate representation of the normalized covariance between the two time series. Therefore, to assess the statistical significance of the estimated wavelet coherency, the Monte Carlo simulation methods are used, and the confidence interval is defined as the probability that the true wavelet power at a certain time and scale lies within a certain interval about the estimated wavelet power (Torrence and Compo, 1998). The theoretical wavelet power 2Pk with the true wavelet power, defined as . The confidence interval for is then:
(10)
Where p is the desired significance (p = 0.05 for the 95% confidence interval), is a chi-square distributed variable with two DOFs (Jenkins and Watts, 1968), and (p/2) represents the value of 2 at p/2. In the WCT figures expressing the coherence between HP-Filter and wavelets, the black thick contour designates the 5% significance level estimated by Monte Carlo simulations using beta surrogate series. The 5% significance level against red noise is shown as a thick contour. All significant sections show anti-phase behavior. The color code for coherency ranges from blue (low coherency close to zero) to red (high coherency close to one). See figure 5.
Figure 5. Wavelet coherence transform (WCT) For the HP cyclical component and residuals As mentioned above, the cross-wavelets transformation gives information on the delay, or synchronization, between oscillations or scales between two time-series. Unfortunately, this information is sometimes incomplete because there is always some redundancy in the time-series (Torrence and Compo, 1998). Consequently, wavelets coherence are used to avoid this situation. As, the cross-wavelet transform will tell us if the correlation is significant or not, the wavelet coherence transformation has the advantage of being normalized by the power spectrum of the two time-series (Torrence and Compo, 1998). Hence, all the regions which represent likely high coherency between two time series are synonym of strong local correlation. In other words, time series with common high coherency areas represent strong possibilities of creating very similar evolution over the time.Looking at figure5, we observe that HP cyclical component and residuals are very synchronous, even for the critical periods of crises. In addition, the regions of high coherence are situated at high frequencies which mean that we can consider that the two approach are very similar.
6. DISCUSSION
After presenting the methodology of using HP-filter and the Wavelets transform approaches in extracting cyclical evolution in time series, it will be more appropriate to ask the following question: if the two approaches are so similar and give the same result, so then what is the objective of such a comparison? To answer this question, we plotin the same graph the HP-Filter cyclical component against SWT residuals time series (figure 6a) and in a second graph the HP-Filter trend component against the SWT denoised signal (figure 6b). As shown in both figures, the plotting of the HP-Filter trend component against the SWT denoised signal is relatively misleading, and it seems that the two series
Page 71

are virtually the same. However, since the moment we are interested to the HP-Filter cyclical component and the SWT residuals, there is a big difference. Certainly both approaches can predict,with a strong robustness,changes in the exchange rate, but even more we can see that the wavelets do not impoverish the information contained in the time series, which is not the case of the HP-Filter. Thus, we can conclude that wavelets are more efficient (in some cases, of course) than the HP-Filter or, in general, the cyclical filters used the economic cycles theory.
0.2 cyclical component of EU-USD exchange rate SWT residuals (S-Ds) of EU-USD exchange rate 0.15
1.6
1.5
HP Trend component of EU-USD exchange rate SWT denoising signal S of EU-USD exchange rate
0.1
1.4
1.3
0.05
1.2
0
1.1
-0.05
1
-0.1
0.9
-0.15 2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
0.8 2000
2001
2002
2003
2004
2005
2006 2007 Time
2008
2009
2010
2011
2012
Figure 6a. HP cyclical component vs. SWT residuals
Figure 6b. HP trend component vs. SWT denoised signal
In a less restrictive optics, both HP-Filter and wavelets can be considered as fundamental tools in analyzing economic phenomena, or even other process in a multidisciplinary perspective. According to figure 7, even if HP-Filter procure more information loss in the time series, this will not be so harmful in the analysis stage, because the difference between the HP cyclical component and the SWT residuals matrix is not very large. Among the whole data set starting from 2000 to 2012, the highest values of the differentials matrix between the cyclical component and the residuals did not exceed the threshold of 0.02 except for two intervals. The first is the one before launching the monetary union in Europe with the adoption of the single currency. The second, starting from the end of 2008 and continue until 2011 with less critical degree, represent the beginning of the worldwide economic crisis and its knock-on effect on the world economy. As we have selected a very decisive currency (USD) which represent the major part of the worldwide transactions in financial and foreign trade markets, and its value of exchange against another outclassed currency (Euro), the value of exchange between these two currencies is a strong index used in predicting world currency crises and financial downturns.
0.15
0.1
d = HPc - SWTr
0.05
-0.05
-0.1 2000
2001
2002
2003
2004
2005
2006 2007 Time
2008
2009
2010
2011
2012
Figure 7. Differentialsmatric between HP cyclical component and SWT residuals
7. CONCLUSION
In this paper, we have tried to make a deeper understanding of the use of filters in the economic cycles theory and we tried to make a comparison between one of the most used filters in the economic literature, the Hodrick-Prescott filter, and the Wavelets transform as an advanced part of the Fourier transform. The used techniques were applied in a simulation case of the Euro-USD foreign exchange rate data. Moreover, our work has included a cross-wavelets and a wavelets coherence process to assess the similarities between the two approaches. The main results have showed that the chosen techniques can perform well in studying a very fluctuating time series data as long as we consider the cyclical component instead of the trend component of the HP-Filter, and also, considering the residuals matrix of the stationary waveletstransform instead of the denoised signal.
Page 72

Also, the results have showed that in some critical time period within the time series data, the HP-Filter is less effective in comparison with the wavelets transform. This interesting statement is very important from an economic perspective, and more precisely when we analyze a financial time series data. Moreover, by considering a very sensitive financial variable like the foreign exchange rate, we must build the modelling process very carefully as long as the slightest variation in the exchange rate value is considered as a very crucial information. Modeling is a very complicated process, and consequently, policy makers need to have a sounded overview of the phenomenon under study and, at the same time, know how to manage environmental constraints. The challenge is to know how to reconcile the two extremes: model richness and integrity against modeling cost. Wavelets seems to be among the tools that help to create this complicated equilibrium as we have shown in this work.
REFERENCES
[1]. L. Aguiar-Conrariaand M.J. Soares,Business Cycle Synchronisation across the Euro Area: A Wavelet Analysis. University of Minho, NIPE Working Papers8, 2009.[Online]. Available at: http://www3.eeg.uminho.pt/economia/nipe/docs/2009/NIPE_WP_8_2009.pdf.[Accessed: Aug. 15, 2012] [2]. M. Baxter and R.G. King, Measuring business cycles approximate band-pass filters for economic time series. NBER Working Paper No. 5022, National Bureau of Economic Research, 1995.[Online]. Available at:http://www.nber.org/papers/w5022. [Accessed: Sept. 05, 2012] [3]. P. Bradley,Shift-invariance in the discrete wavelet transform. In Proceedings of the Seventh International Conference on Digital Image Computing: Techniques and Applications, pp. 29-38, 2003. [4]. S. Butterworth, On the theory of filter amplifiers. Journal of Experimental Wireless and the Wireless Engineer vol.7, pp. 536-541, 1930. [5]. L.J. Christianoand T.J. Fitzgerald, The band pass filter. International Economic Review, vol.44, pp. 435-465, 2003. [6]. E.A. Cohen and A.T. Walden,A Statistical Analysis of Morse Wavelet Coherence, IEEE Transactions of Signal Processing, vol. 58(3), pp. 980-989, 2010a. [7]. P.M. Crowley and J. Lee,Decomposing the Co-movement of the Business Cycle: A Time-Frequency Analysis of Growth Cycles in the Euro Area. Bank of Finland, Bank of Finland Research Discussion Papers 12, 2005. [Online]. Available at: http://www.suomenpankki.fi/en/julkaisut/tutkimukset/keskustelualoitteet/Documents/0512netti.pdf. [Accessed: Nov. 09, 2012] [8]. P.M. Crowley, D. Maraun and D. Mayes,How Hard Is the Euro Area Core? An Evaluation of Growth Cycles Using Wavelet Analysis. Bank of Finland, Bank of Finland Research Discussion Papers 18, 2006. [Online]. Available at: http://www.suomenpankki.fi/en/julkaisut/tutkimukset/keskustelualoitteet/Documents/0618netti.pdf. [Accessed: Nov. 09, 2012] [9]. D. Donoho and I.M. Johnstone,Adapting to unknown smoothness via wavelet shrinkage. Journal of the American Statistical Association, vol. 90, pp. 1200-1224, 1995. [10]. D.L. Donoho,Denoising via soft thresholding. IEEE Transactions on Information Theory, vol. 41, pp613-627, 1995. [11]. M. Gallegati and M. Gallegati,Wavelet Variance Analysis of Output in G-7 Countries. Studies in Nonlinear Dynamics and Econometrics, vol.11(3): art. 6, 2007.[Online].Available at: http://ideas.repec.org/a/bpj/sndecm/v11y2007i3n6.html [Accessed: Dec. 02, 2012] [12]. D. Gnanadurai and V. Sadasivam,An efficient adaptive thresholding technique for wavelet based image denoising. International Journal of Signal Processing, vol. 2, pp. 114-120, 2006. [13]. A. Grinsted, J.C. Moore and S. Jevrejeva,Application of the cross wavelet transform and wavelet coherence to geophysical time series, Nonlinear Processes in Geophysics in European Geosciences Union, vol. 11, pp. 561566, 2004. [14]. S. Guoxiang and Z. Ruizhen,Three novel models of threshold estimator for wavelet coefficients. WAA 2001, Lecture Notes in Computer Science, vol. 2251, pp. 145-150, 2001. [15]. R.J. Hodrick and E.C. Prescott, Postwar U.S. business cycles: An empirical investigation. Journal of Money, Credit, and Banking,vol.29, pp. 1-16, 1997. [16]. G.M. Jenkins and D.C.Watts,Spectral Analysis and Its Applications. Holden-Day, p525, 1968. [17]. F. Keinert,Wavelets and multiwavelets, studies in advanced mathematics. Chapman & Hall/CRC Press, vol.5, pp. 25-29, 2004. [18]. P.C. Liu,Wavelet spectrum analysis and ocean wind waves. Wavelets in Geophysics, E. Foufoula-Georgiou and P. Kumar, Eds., Academic Press, pp.151-166, 1994. [19]. S.M. Raihan, Y. Wen and B. Zeng,Wavelet: A New Tool for Business Cycle Analysis. Federal Reserve Bank of St. Louis, Working Paper 050A, 2005. [Online]. Available at: http://research.stlouisfed.org/wp/more/2005050/. [Accessed: Nov. 25, 2012]
Page 73

[20]. M.O. Ravnand H. Uhlig, On adjusting the Hodrick-Prescott Filter for the frequency of observations, Open Access publications from University College London, 2002. Available at: http://discovery.ucl.ac.u. [Accessed: Dec. 12, 2012]. [21]. J.H. Stockand M.W. Watson,Business cycle fluctuations in us macroeconomic time series, In Handbook of Macroeconomics, in: J. B. Taylor & M. Woodford (ed.), Handbook of Macroeconomics, edition 1, volume 1, chapter 1, pp. 3-64 Elsevier, 1999. [22]. K.H. Talukder and K. Harada,Haar wavelet based approach for image compression and quality assessment of compressed image. IAENG International Journal of Applied Mathematics, vol. 36, pp. 41-42, 2007. [23]. C. Torrence and G.P. Compo,A Practical Guide to Wavelet Analysis, Bulletin of the American Meteorological Society, vol. 79, pp. 605-618, 1998. [24]. C. Torrence and P. Webster,Interdecadal Changes in the ESNOMonsoon System, J. Clim. Vol. 12, pp. 26792690, 1999. [25]. M. Yogo, Measuring Business Cycles: A Wavelet Analysis of Economic Time Series. Economics Letters, vol. 100(2), pp208-212, 2008. [Online]. Available at: http://ideas.repec.org/a/eee/ecolet/v100y2008i2p208-212.html. [Accessed: Dec. 12, 2012]
Page 74

HP-Filter and Wavelets Analysis Modelling in Financial Markets

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

HP-Filter and Wavelets Analysis Modelling in Financial Markets

Hochgeladen von

Copyright:

Verfügbare Formate

International Journal of Application or Innovation in Engineering & Management (IJAIEM)