See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/316426732

Forecasting of volatility risk for Jordanian banking sector

Article  in  Far East Journal of Mathematical Sciences · April 2017

DOI: 10.17654/MS101071491

CITATIONS READS

2 120

4 authors:

Jamil J. Jaber Ismail Noriszura

University of Jordan Universiti Kebangsaan Malaysia
10 PUBLICATIONS   11 CITATIONS    95 PUBLICATIONS   316 CITATIONS   

SEE PROFILE SEE PROFILE

Sadam Alwadi Mohammad Saleh

University of Jordan University of Jordan
32 PUBLICATIONS   75 CITATIONS    10 PUBLICATIONS   18 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Article ‫ ﺗﺤﻠﻴﻞ ﻣﺆش ا رت ﻣﺨﺘﺎرة ﻟﻈﺎﻫﺮة اﻟﻔﻘﺮ ﻓﻲ ﻣﺤﺎﻓﻈﺎت اﻟﺠﻨﻮب ﻓﻲ اﻷردن‬View project

forecasting View project

All content following this page was uploaded by Sadam Alwadi on 11 July 2017.

The user has requested enhancement of the downloaded file.

 Far East Journal of Mathematical Sciences (FJMS)
© 2017 Pushpa Publishing House, Allahabad, India
http://www.pphmj.com
http://dx.doi.org/10.17654/...
Volume …, Number …, 2017, Pages … ISSN: 0972-0871

FORECASTING OF VOLATILITY RISK FOR

JORDANIAN BANKING SECTOR

Jamil J. Jaber1,2,*, Noriszura Ismail2, S. Al Wadi1 and

Mohammad H. Saleh1
1
Department of Risk Mathematics and Insurance
Faculty of Management and Finance
The University of Jordan
Jordan
e-mail: j.jaber@ju.edu.jo
jameljaber2011@hotmail.com
2
School of Mathematical Sciences
Faculty of Science and Technology
Universiti Kebangsaan Malaysia
Malaysia

Abstract

In recent years, the instability and unpredictability of financial markets

have played an essential part in risk management. Volatility risk is
characterized as the standard deviation of the constantly compound
return per day. This paper shows forecasting of volatility for the
Jordanian banking sector after 2006 crisis. Parameters p, d, and q are
estimated by ARIMA and ARIMA-WT. After that, several accuracy
criteria are used to compare between these models for each bank that
are incorporated at Amman stock exchange (ASE). Finally, points of
volatility are estimated at 95% confidence interval for the future. In
Received: November 20, 2016; Accepted: January 9, 2017
2010 Mathematics Subject Classification: Kindly provide.
Keywords and phrases: volatility, risk, market, ARIMA.
*
Corresponding author
2 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh
the results, we found ARIMA-WT has more accuracy than ARIMA.
ARIMA-WT is more fitted for several reasons. First, RMSE for
ARIMA-WT is less than ARIMA. Second, AIC and BIC are closer
than zero.

1. Introduction

In recent decent, several specific financial types of research have been

focused on the volatility and forecasting volatility of models that play a
crucial role in risk management and financial asset pricing such as stocks,
bonds, and derivatives contracts. The accuracy of forecasting volatility helps
financial market participants and regulators to measure and estimate future
risks. After that the regulators can take the right decisions regarding the
financial instruments (Bollerslev et al. [4]). The exposure of market risk was
officially predicted along with the accurate forecasting volatility. The high
fluctuation of stock price highlights the significance of volatility forecast. In
recent years, a long list of volatility and uncertain models have been
proposed in the academic works of literature for testing the fundamental
trade-off between risks and return of financial assets, and for investigating
the causes and consequences of the volatility dynamic in the economy.
Given its importance, GARCH type models are the most commonly used
models that proposed in the literature for that estimating and forecasting
financial market volatility. Based on Engle’s ARCH (Autoregressive
Conditional Heteroskedasticity) in 1982 and Bolllerselv’s GARCH
(Generalized Autoregressive Conditional Heteroskedasticity) models in
1986, several of GARCH models are listed, such as NAGARCH, GJR-
GARCH, FIGARCH, and EGARCH. These models are used in exchange,
bonds, and options markets. A large percentage of studies focused on the
stock market volatility. Trück and Liang examined the performance of
different models (GARCH, TARCH, TGARCH, and ARMA) in 2012.
Researchers found that the TARCH models provided the best results in the
gold market.
 Forecasting of Volatility Risk for Jordanian Banking Sector 3

This paper compares previous volatility approaches and developed them

along with the wavelet transforming model and forecasting of banks’ stocks’
prices in Jordan. Generally, it could be based on ARIMA models, which
received great attention in economic and finance areas. George Box and
Gwilym Jenkins popularized this model in 1970s. Furthermore, ARIMA
model can be defined as the following: ARIMA( p, d , q ) , where: P: order of
autoregressive part (AR), d: degree of first differentiation (I) and q: order of
the first moving part (MA). According to that, ARIMA models create the
best forecast for the time series. This method is not costly compared to other
methods in the quantitative models (Pankratz, 1983).
The rest of the study is structured as follows. Section 2 provides a
literature review. Section 3 provides data description. Section 4 reports the
results from an empirical application of the stock market. Section 5
concludes.

2. Literature Review

2.1. Volatility literature

Modeling and forecasting volatility is a popular research agenda in
financial markets. There is a continuing increase in a number of models and
approaches for forecasting volatility in financial assets. (Gang and Yong [8])
examines the linear models (ARCH and GARCH) and nonlinear models
(EGARCH and exponential GARCH) for the daily price of the three-month
copper futures with sample size 2200. The return is obtained by calculating
the first order difference of logarithmic price and the squared return is taken
as the proxy of actual volatility. The researcher used the loss functions for
assessing the predictive power of the GARCH-type models and model
averaging methods. They found that the OLS time-varying weighted model
averaging method can achieve the smallest forecasting error and significantly
reduce the over-prediction percentage. In addition, they found that there
exists GARCH effect in the copper futures market at 5% level. Bentes [3]
studies the adequacy of GARCH-class models in describing the gold
volatility behavior. And, compare the out-of-sample predictive ability models
4 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

(GARCH, IGARCH and FIGARCH) based on three evaluation criterions;

mean absolute error (MAE), root mean squared error (RMSE) and Theil’s
inequality coefficient (TIC) among a long time frame of over 39 years in
order to provide a general picture of the overall behavior of the time series.
He found that FIGARCH (1, d , 1) is the best forecasting model for the return
series based on either MAE and RMSE criteria, and when taking into account
the TIC criterion, IGARCH (1,1) works the best. While for all criteria select
GARCH (1,1) as the worst forecasting model. Moreover, the MAE, RMSE,
and TIC each select the FIGARCH (1, d , 1) as the best forecasting model but
evidence regarding the worst is mixed. (Joukar and Nahmens [11]) used
ARCH and GARCH methods to determine the persistency of volatilities in
the case of external shocks (economic shocks or political shocks) for the
period of January 1978 to July 2014. They detect salient features of the
economic series via calculating the variance of construction cost index (CCI).
The authors noticed a decreasing trend over the past 4 years for the CCI
volatility, which suggests less likelihood of overestimation or
underestimation of the construction cost estimation; the contractors could
consider lower premium in fixed price contracts. (Kambouroudis and
McMillan [12]) explored the effect of including the volatility index (VIX) as
a benchmark of expected short-term market (options and futures), and trading
volume (VO) within the volatility forecasting model, with the explicit
purpose of seeing whether their inclusion improves forecast performance.
Researchers used six GARCH models to check the ability of each model to
capture different features of the data for the three markets of the US, the UK
and France, such as volatility clustering, asymmetry, and long memory. They
found that VIX and volume help to produce more accurate volatility forecasts
in comparison to the GARCH forecasts alone. On the basis of the MZ
regressions, the results showed that an increase in R2 can be achieved.
Moreover, they found that the forecast models perform slightly better when
VIX and volume are included as independent or forecasted variables,
although the results are less clear-cut when they are included as regressors in
the GARCH variance equation, and that provided evidence that VIX and
volume have additional explanatory power in forecasting returns volatility.
 Forecasting of Volatility Risk for Jordanian Banking Sector 5

Kristjanpoller and McMillan [12] examine the improvement of accuracy in

forecasting using a hybrid model as opposed to traditional GARCH models.
They demonstrated the greater accuracy of the hybrid Artificial Neural
Network-GARCH forecasts of gold price volatility for different periods and
for spot and future gold prices over the GARCH alone. They found that the
ANN-GARCH model improves the forecast results comparing with that of
the GARCH model by 25% for the gold spot price volatility and by 38% for
the gold future price volatility. Abounoori et al. [1] have done a research
paper about forecasting the volatility of Tehran stock exchange (TSE). The
data consist of 3067 daily observations of the closing prices of the Tehran
stock exchange for 13 years. This paper concentrated on the empirical
estimates of ARMA, single-regime GARCH, and MRS-GARCH models,
together with the in the sample and the out-of-sample forecast evaluation.
They founded that leverage effect exists in Tehran stock market, since the
effects of bad news on volatility are larger than the effects of good news on
volatility, and the differences between the good and bad news coefficients
are not significant for GJR-N model but it is significant at 5% confidence
level for GJR models with t-student and GED distributions. In addition, the
researcher rejected the null hypothesis of no difference in the accuracy of the
two competing forecasts at 1, 5 and 10 days period horizon. They used the
DM test for the equal predictive ability of AR(2)-MRSGARCH with GED
innovations against predictive ability of other GARCH and MRSGARCH
models, and the risk management at 95% VaR indicated the same; that not
clear to find the best model. Byun [5] investigates potential channels through
cross-sectional information for predict aggregate volatility, and the
researcher developed a model of individual returns to the study of volatility
in any financial assets, such as the S & P 500 Composite index at the stock
market. There is a total of 1545 individual stocks that appear in the index. He
found that GARCH-X specifications could improve in-sample forecasting
accuracy, and cross-sectional dispersion does not appear to be useful for out-
of-sample forecasting. GARCH-X with cross-sectional dispersion failed to
provide more accurate out-of-sample volatility forecasts than GARCH. In
addition, he found that cross-sectional dispersion improves the accuracy of
aggregate volatility forecasts both in the sample and out of sample by
6 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

checking the possibility of accurate estimation of model parameters using

individual stocks jointly. Furthermore, out-of-sample volatility forecasts
from the bivariate GARCH model are shown to be more accurate than those
from GARCH in times of NBER recessions as well as during the periods
with negative S&P Index returns. Thus, the researcher concluded that cross-
sectional dispersion helps to predict volatility forecasts indirectly by helping
to estimate parameters. Bollerslev et al. [4] worked on a new class of
volatility forecasting models. Those models use a realistic theory for high-
frequency realized volatility estimation to improve the accuracy of the
forecasts and to generate more responsive forecasts when the measurement
error is relatively low. The researchers implemented the new models for the
S&P 500 (high-frequency futures prices) equity index and the individual
constituents of the Dow Jones industrial average for the period 1997-201.
Moreover, they noticed significant improvements in the accuracy of the
resulting forecasts compared to the forecasts from some of the most popular
existing models that implicitly ignore the temporal variation in the magnitude
of the realized volatility measurement errors. They found that the HARQ
model is slightly more subtle while the HAR places greater weight on the
weekly and monthly lags, which are less ability to measurement errors than
the daily lag, but also further in the past. This was indicated that for the one-
day forecasts, the daily lag is generally the most important in the estimated
HARQ models.
However, wavelet transform (WT) is a mathematical model that used to
convert the original observation into a different domain (Daubechies [7]);
(Chang and Moretin [6]); (Gencay et al. [9]). The WT model is suitable for
financial data that is utmost non-stationary (Al Wadi et al. 2013). (Al-
Khazaleh et al. [2]) used financial data from Amman stock exchange (ASE)
to test three methods (box plot, Z-score, wavelet transform asymmetric
winsorized mean (WTAWM)) for outlier detection. He found that the
wavelet transform asymmetric winsorized mean (WTAWN) is the best model
in outlier detections. However, the box plot method could not give
significant solutions and Z-score could not give any outlier if the data could
not follow the normal distribution.
 Forecasting of Volatility Risk for Jordanian Banking Sector 7

3. Mathematically Formulations

In this section, the mathematical formulas will be discussed. The

definition of volatility will be mentioned in the first part. Wavelet transform
will be discussed in the second part. Then ARIMA formula will be
discussing in the third part. After that, the several approaches to accuracy
will be listed.

3.1. Volatility formula

In risk management, volatility is defined as the standard deviation of the

continuously compounded return per day. Define Si as the close price at the
end of day i. The continuously compounded return per day for the close price
on day i as

Si
R = ln .
Si − 1

A variable’s volatility, σt , is defined as the standard deviation of Ri′ s at

time t.

∑i =1 (Ri − R )2 ,
n

σt =
n −1

where R is the arithmetic mean of the Ri′ s (Hull, 2012).

3.2. Wavlet transform formula

DWT has defined by (Chiann and Moretin [6]), (Gencay et al. [9]) as:

j
ψ j , k (t ) = 2 ψ(2 j t
2 − k ), j , k ∈ Z ; z = {0, 1, 2, "},

where ψ is defined as a real-valued function having compactly supported,

∞
and ∫ − ∞ ψ(t ) dt = 0. Generally, the wavelet transforms were evaluated by
8 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

using dilation equations and defined as:

φ(t ) = 2 ∑ lk φ(2t − k ),
k

ψ(t ) = 2 ∑ hk φ(2t − k ),
k

where φ(2t − k ) represents the father wavelet, and ψ(t ) represents the
mother wavelet. Father wavelet gives the high scale approximation
components of the signal, while the mother wavelet shows the deviations
from the approximation components. The father wavelet generates the
scaling coefficients, while mother wavelet evaluates the differencing
coefficients. Father wavelet defines high pass filters coefficients (lk ) and the
lower pass filter coefficients (hk ) are defined respectively as follow (Gencay
et al, [9]):
∞
lk = 2 ∫ − ∞ φ(t ) φ(2t − k ) dt,
∞
hk = 2 ∫ − ∞ ψ(t ) ψ(2t − k ) dt.
Whereas HWT is the oldest and simplest example in the wavelet transforms
and it can be defined as:

⎧1, 0≤t ≤
1
⎪ 2
⎪ 1
ψ (t ) = ⎨−1,
H
≤ t ≤1
⎪ 2
⎪0, Otherwise
⎩
For the HWT:
∞ ⎧1, 0 ≤ t ≤ 1
lk = 2 ∫ −∞
φ(t ) φ(2t − k ) dt = ⎨
⎩0, otherwise
,

for N = 2, lk = ⎧⎨
1 1 ⎫ ⎧ 1 , − 1 ⎫.
, ⎬, hk = ⎨ ⎬
⎩ 2 2⎭ ⎩ 2 2⎭
 Forecasting of Volatility Risk for Jordanian Banking Sector 9

The mother wavelet satisfies the following two conditions:

∞ ∞ ∞ ψ1(ω) 2
∫ −∞ ψ(t ) dt = 0, ∫ −∞ ψ(t ) < ∞, ∫ −∞ ω
dω < ∞,

where ψ1(ω) presents the wavelet transform.

The HWT was improved and developed the frequency-domain

characteristics by DWT. However, there is no specific formula for this
method of the wavelet transform. Thus, we tend to use the square gain
function of their scaling filter, it is defined as (Gencay et al, [9] and Alrumaih
and Al- Fawzan, (2002)):
l
−1
2 ⎛ l ⎞
g ( f ) = 2 cost (πf ) ∑ ⎜⎜⎝ 2 − l1 + l ⎟⎟⎠ sin 2l (πf ),
l =0

where l: Positive number and represents the length of the filter,

For more details and examples see (Gencay et al. [9]), (Alrumaih and
Al- Fawzan, 2002), (Manchanda et al. 2007), (Motohiro, 2008), (Struzik,
2001), (Janacek and Swift, 1993) (Daubechies [7]) and (Cai 2005).

3.3. Autoregressive integrated moving-average model (ARIMA (p, d, q))

The auto-regressive moving average (ARMA) models are used in time

series analysis to describe stationary time series. The ARMA model is a
combination of an autoregressive (AR) model and a moving average (MA)
model. A time series {et } is said to be a white noise (WN) process, {Yt } is

called Gaussian process iff for all t, et is iid N (0, σ 2 ).

A time series {Yt } is said to follow the ARMA (p, q) model if:

Yt = ì + φ1Yt −1 + φ2Yt − 2 " + φ pYt − p + et − è1et −1 − è2et − 2 " èq et − q

Where p and q are non-negative integers, p represents order of autoregressive

part (AR), q is defined as order of the first moving part (MA) and {et } is the
white noise (WN) process.
10 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

An extension of the ordinary ARMA model is the auto-regressive

integrated moving-average model (ARIMA (p, d, q)) given by:

φ p ( B ) (1 − B )d Yt = è0 + èq ( B ) et ,

where p, d and q denote orders of auto-regression, integration (differencing)

and moving average, respectively.
When d = 0, the ARIMA model reduces to the ordinary ARMA model

ARIMA model is the most popular way of forecasting since there is no

need for any assumptions and it is not limited to specific type of pattern.
These models can be fitted to any set of time series data (stationary or non-
stationary) by estimating the parameters p, d, and q to be suitable with the
required dataset.
3.4. Accuracy criteria
The criteria that have been used to make a fair comparison between
ARIMA and ARIMA-WTcan be presented in this section. We have been
adopted three types of accuracy criteria; Root means squared error (RMSE),
Akaike information criterion (AIC), and Bayesian information criterion
(BIC). The framework comparison can be presented with more details as
follows:
(1) Root mean squared error (RMSE).
N
∑ (actual value-predicted value)2
RMSE = i =1 ,
N
where
N represents the number of observations.
(2) Bayesian information criterion (BIC).
BIC is based on the maximum likelihood estimates of the model
parameters (Schwarz, 1978)

BIC = −2∗ log − likelihood + k ∗npar ,

 Forecasting of Volatility Risk for Jordanian Banking Sector 11

where

npar represents the number of parameters in the fitted model, k = log( n ) ,

and n being the number of observations.
(3) Akaike information criterion (AIC).
Also, AIC is based on the maximum likelihood estimates of the model
parameters (Akaike, 1969)

AIC = −2∗ log − likelihood + k ∗npur ,

where npar represents the number of parameters in the fitted model, k = 2,

and n being the number of observations.

4. Methodology and Results

In order to illustrate volatility and forecasting volatility risk for the

banking sector in the stock market. Daily close price is used from banks after
crises 2006 for the time period from July 2006 until December 2015 that is
selected from Amman Stock Exchange (ASE). We used MATLAB and R-
studio to test the models.
4.1. Comparison with ARIMA and ARIMA-WT and accuracy criteria
The MODWT converts the data into two sets; approximation series (CA1
(n)) and details series (DA1 (n)). These two series presented good behavior
for the data set especially with the insurance data since it is significant
fluctuated. Then, the transformed data can be predicted more accurately. The
reason for the good behavior of these two series is the filtering effect of the
MODWT. Moreover, the approximation series has been used since the series
behave as the main component of the transform. The procedure of
conducting the comparison can be described as follows: Firstly, decompose
through the MODWT, HWT, and DWT the available historical return data.
Secondly, Use specific ARIMA model fitted to each one of the
approximation series to make the forecasting. Finally, this technique is
compared with ARIMA model used directly to forecast the return data series
by using the mentioned criteria.
12 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

Table 1. Comparison with ARIMA and ARIMA-WT models for Jordanian

banks.

In the Table 1 appear comparisons between ARIMA and ARIMA-WT

models for Jordanian banks and accuracy criteria; RMSE, AIC, and BIC for
all banks in Jordan. As indicated from the table above, we found that
ARIMA (1.0,1) with WT is better than ARIMA (2,1,3) directly model
because RMSE (0.034), AIC (-27601.72), and BIC (-27579.1) are lower in
ABCO bank. Also, we can see that ARIMA (1,1,1) with WT is more accurate
than ARIMA(5,1,5) directly in ARBK bank because RMSE (0.021), AIC (-
31090.94), and BIC (-27579.1) are lower. Indeed, ARIMA-WT is more
accurate than direct ARIMA because RMSE, AIC, and BIC are better.

4.2. Daily forecasting volatility of ARIMA-WT

The daily forecasting volatility of ARIMA-WT is figured for each

Jordanian bank for one trading year in this section. Daily trade for one year is
characterized in an x-axis. The forecasting value appears in the y-axis.
Forecasting of Volatility Risk for Jordanian Banking Sector 13
14 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

Figure 1. Forecasting volatility of Jordanian banks with ARIMA-WT for one

trading year.
The daily forecasting volatility of ARIMA-WT for each Jordanian bank
appeared in the previous figure. The percentage of volatility is high in the
first two years because the crisis 2006. After that, we found there is a
stability of volatility for all banks after the crisis. In addition, we found that
index of bank sector is close from zero. In the other words, the volume of
trading after crisis 2006 is lower.
 Forecasting of Volatility Risk for Jordanian Banking Sector 15

Table 2. Forecasting volatility for Jordanian banks for three days by

ARIMA-WT

The estimating point of volatility at 95% confidence interval for each

Jordanian bank appears in Table 2. The percentage of volatility is low for all
Jordanian banks. The forecasting volatility point of SGBJ is 0.032 that is the
highest point in the first day. In addition, the lower point was BOJK and
AHLI banks. However, the forecasting volatility point for bank index is
0.0077 on the first day.

5. Conclusion

ARIMA model is the most general way of forecasting since there is no

need for any assumptions and it is not limited to specific type of pattern.
These models can be fitted to any set of time series data (stationary or non-
stationary) by estimating the parameters p, d, and q to be suitable with the
required dataset (Ahmad Mahir and Al-khazaleh, [15]) . In this study, firstly,
the volatility of the stock price is modeled using wavelet method. Secondly,
16 Jamil J. Jaber, Noriszura Ismail, S. Al Wadi and Mohammad H. Saleh

we compared ARIMA with Wavelet model. Thirdly, we tested the accuracy

of these models by using RMSE, AIC, and BIC assessing functions. Finally,
we estimate the forecasting point of volatility for each ARIMA-wavelet
model for each Jordanian bank. Indeed, we found ARIMA-WT is a suitable
model.

References

[1] E. Abounoori, Z. Elmi and Y. Nademic, Forecasting Tehran stock exchange

volatility; Markov switching GARCH approach, Physica A: Statistical Mechanics
and its Applications 445 (2016), 264-282.
[2] A. Al-Khazaleh, S. Alwadi and F. Ababneh, Wavelet Transform Asymmetric
Winsorized Mean In Detecting Outlier Value. Far East Journal of Mathematical
Sciences, 2015.
[3] S. Bentes, Modeling and forecasting volatility in gold returns under the GARCH,
IGARCH and FIGARCH frameworks: New evidence, Physica A: Statistical
Mechanics and its Applications 48 (2015), 355-364.
[4] T. Bollerslev, A. Patton and R. Quaedvlieg, Exploiting the errors: A simple
approach for improved volatility forecasting, Journal of Econometrics 192 (2016),
1-18.
[5] S. Byun, The usefulness of cross-sectional dispersion for forecasting aggregate
stock price volatility, Journal of Empirical Finance 36 (2016), 162-180.
[6] C. Chang and P. Moretin, A Wavelet Analysis for Time Series, Nonparametric
Statistics 10 (1998), 1-46.
[7] I. Daubechies, Ten Lectures on Wavelets, SIAM and Philadelphia, 1992.
[8] L. Gang and L. Yong, Forecasting copper futures volatility under model
uncertainty, Resources Policy: The International Journal of Minerals Policy and
Economics 46 (2015), 167-176.
[9] R. Gencay, F. Seluk and B. Whitcher, An Introduction to Wavelets and Other
Filtering Methods in Finance and Economics, Academic Press, New York, 2002.
[10] J. Jaber and A. Al-khawaldeh, The Impact of Internal and External Factors on
Commercial Bank Profitability in Jordan. International Journal of Business and
Management 9(4) (2014).
[11] A. Joukar and I. Nahmens, Volatility Forecast of Construction Cost Index Using
General Autoregressive Conditional Heteroskedastic Method, Journal of
Construction Engineering and Management 142 (2015), 1943-7862.
 Forecasting of Volatility Risk for Jordanian Banking Sector 17
[12] D. Kambouroudis and D. McMillan, Does VIX or volume improve GARCH
volatility forecasts?, Journal of Applied Economics 48 (2015), 1210-1228.
[13] W. Kristjanpoller and M. Minutolo, Gold price volatility: A forecasting approach
using the Artificial Neural Network-GARCH model, International Journal: Expert
Systems with Applications 42 (2015), 7245-7251.
[14] C. Lanouar, Breaks or long range dependence in the energy futures volatility: Out-
of-sample forecasting and VaR analysis, The International Journal of Theoretical
and Applied Papers on Economic Modeling 53 (2016), 354-374.
[15] A. Mahir and A. Al-Khazaleh, Estimation of missing data by using the filtering
process in a time series modeling. Appl. Math. Sci. 3(35) (2008), 1715-1726.
[16] M. Marcellion, M. Porqueddu and F. Venditti, Short-Term GDP Forecasting With
a Mixed-Frequency Dynamic Factor Model With Stochastic Volatility, Journal of
Business & Economic Statistics 34 (2016), 118-127.
[17] C. Pierdzioch, M. Risse and S. Rohloff, A boosting approach to forecasting the
volatility of gold-price fluctuations under flexible loss, The International Journal
of Minerals Policy and Economics 47 (2016), 95-17.
[18] S. Trück and K. Liang, Modelling and forecasting volatility in the gold market,
Journal of International Banking and Finance 9(1) (2012), 48-80.
[19] W. N. Venables and B. D. Ripley, Modern Applied Statistics with S. New York:
Springer (4th ed), 2002.
[20] S. Wadi and M. Ismail, Selecting wavelet transforms model in forecasting
financial time series data based on ARIMA model. Applied Mathematical
Sciences 5 (7) (2013), 315-326.

Paper No. PPH-1611056-MS
Kindly return the
correction to:
The Publication Manager
Pushpa Publishing House
Vijaya Niwas
198, Mumfordganj
Allahabad-211002 (India)
along with the print charges*
by the fastest mail
*Invoice attached
View publication stats
Proof read by: …………………….…………
Copyright transferred to the Pushpa
Publishing House
proof after
Signature: .………………………..…...……
Date: ……..………………………………….
Tel: ……..……………………….……………
Fax: ………..…………………………………
e-mail: ………..……………..………………
Number of additional reprints required
………………………………………….……
Cost of a set of 25 copies of additional
reprints @ U.S. Dollars 15.00 per page.
(25 copies of reprints are provided to the
corresponding author ex-gratis)