Hwf2105059536idv 15013408524309796pdf Hi0001 [600783]

For Peer Review Only

Technical analysis in estimating currency risk port folios:
Case Study: commercial banks in Romania

Journal: Economic Research-Ekonomska Istraživanja
Manuscript ID RERO-2016-0191
Manuscript Type: Full Article
Keywords: Value – at Risk (VaR) method, , method based on his torical simulations,
rate of exchange, standard deviation, maximum loss, currency net
position, financial instruments
Abstract: Loss estimates resulting from foreign currency port folios was one of the
challenges for bank management, numerous techniques and management
methods of foreign exchange risk being developed, t he method Value at
Risk (VaR) becoming one of the most used tools for management and
administration of foreign currency portfolios. With in commercial banks
operating in the Romanian banking system, the main denomination
currency of balance sheet and off-balance sheet ite ms is euro. The study
aims to estimate the maximum loss for this currency , due to exchange rate
volatility by establishing VaR, starting from the m ethod based on historical
simulations. The sample of the research consists of Romanian Commercial
Bank SA, Romanian Development Bank – Groupe Societe Generale,
Transylvania Bank SA and Carpatica Bank), based o n a number of 757
observations corresponding to the working days in t he period 01.01.2012 –
31.12.2014. The results obtained from the research showed that at BT,
BRD, BCC and BCR the maximum anticipated loss in a future time horizon
of 10 days, with a relevance of 1%, does not exceed 2% of the net
positions on the euro currency. The information ava ilable for the four
commercial banks include only EURO, RON and other c urrencies net
positions.

URL: https://mc.manuscriptcentral.com/rero E-mail: [anonimizat] Research-Ekonomska Istra?ivanja

For Peer Review OnlyTechnical analysis in estimating currency risk port folios: Case Study:
commercial banks in Romania

Abstract : Loss estimates resulting from foreign currency port folios was
one of the challenges for bank management, numerous techniques and
management methods of foreign exchange risk being d eveloped, the method
Value at Risk (VaR) becoming one of the most used t ools for management and
administration of foreign currency portfolios. With in commercial banks
operating in the Romanian banking system, the main denomination currency of
balance sheet and offGbalance sheet items is euro. The study aims to estimate the
maximum loss for this currency, due to exchange rat e volatility by establishing
VaR, starting from the method based on historical s imulations. The sample of the
research consists of the four commercial banks list ed on the Bucharest Stock
Exchange (BSE), respectively: Banca Comercială Româ nă SA G Romanian
Commercial Bank SA (member of Erste– BCR Group), Ba nca Română pentru
Dezvoltare G Romanian Development Bank G GroupeSoci eteGenerale (BRD),
Banca Transilvania – Transylvania Bank SA (BT) and Ba ncaCarpatica –
Carpatica Bank (BCC), based on a number of 757 obse rvations corresponding to
the working days in the period 01.01.2012 G 31.12.2 014. The results obtained
from the research showed that at Banca Transilvania SA (BT), the Romanian
Development Bank – Groupe Societe Generale (BRD), B anca Carpatica (BCC)
and the Romanian Commercial Bank SA (member of ERST EGBCR Group) the
maximum anticipated loss in a future time horizon o f 10 days, with a relevance of
1%, does not exceed 2% of the net positions on the euro currency. The
information available for the four commercial banks include only EURO, RON
and other currencies net positions, which has not a llowed us to extend the study
to other currencies (like USD, CHF etc).
Jel Classification: C11, C13, C15, C33
Key words: Value z at Risk (VaR) method, method bas ed on historical
simulations, rate of exchange, standard deviation, maximum loss, currency net
position, financial instruments.
1. Introduction:
Page 1 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review OnlyThe definition attributed to VaR method explains th e significance of the results,
indicating the estimated maximum loss of a portfoli o of financial instruments on a short
fixed time period and the use of the method involve s arbitrary choice of two parameters:
holding period of the instruments, namely, the timi ng and level of relevance. VaR
method is mostly used in assessing maximum loss res ulting from the manifestation of
various risk categories. Specifically, it is used t o estimate the foreign currency portfolio
depreciation, following an unfavourable exchange ra te variation, credit risk event
(increased nonzperforming loans), operational risk and even liquidity risk .
The history of VaR is related to the investment ban k JP Morgan, whose
president had a decisive role in its development. T he creation of Risk Metrics
department in this bank – specialised only in the s tudy and analysis of the risk z has led
to a measure of risk popularized as the ValuezatzRi sk (VaR). For the first time, this
method has been assigned a particular importance in 1993, in the Report of the Group of
30 .
The importance and success of the method determined the Basel Committee to
issue, in 1996, an amendment to the first Agreement (in force at that time) with the view
to recommend to the commercial banks a way through which they calculate their
minimum capital to cover exposure to market risk. L ater, in the Basel II Agreement
(2004), the recommendation to calculate capital req uirements by taking into account the
outcome of VaR becomes mandatory. Even if this legi slative framework has had
considerable changes made by the Basel III and the Regulation 575/2013 of the
European Parliament on prudential requirements of c redit institutions, VaR calculation
methodology has not undergone essential changes, be ing maintained as a pillar for
setting capital requirements for market risk. Once adopted, the method quickly became
used by commercial banks in the attempt to manage a s efficiently foreign currencies’
portfolios.
VaR ability to early counter financial failures con tributed to increasing the
success of this risk management instrument as is th e Metallgesellschaftz1993 case, the
Baringsz1995 case, the LTCMz1997 case, etc. (Armean u D., Bălu F. O., pg. 83).
The developments of the international banking syste m, the complexity of the
banking activity and of the economic environment ha ve imposed further adaptations of
risk management techniques.
European Parliament Regulation no. 575/2013 on prud ential requirements for
credit institutions recommends VaR calculation by s trict compliance with the following
steps:
a. Determining daily the level of exposure of risk;
b. Establishing confidence interval with a probability of 99%;
c. Withholding period: 10 days;
d. Observation period:1 year minimum;
e. Monthly updating of the database used to determine VaR.

In order to determine VaR (analytic VaR, VaR calcul ated using Monte Carlo
simulation, historical VaR, etc.) various models we re developed, starting from the
financial instruments to which the method can be ap plied; the accuracy ofrisk measures
(including statistical assumptions on which they ar e based); implementation
requirements (risk assessment models, risk decompos ition, data availability); necessary
informatics systems; ease of communication of resul ts to users (Moinescu B., Codîrlașu
A., 2009).

2. Specialty literature: Page 2 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only
The wide use of VaR method is closely related on th e one hand to the
recommendations of the Basel Committee, which requi re the calculation of capital
requirements for market risk based on the results o btained, and, on the other hand, to the
possibility of its application in determining the l oss from various areas of activity in the
loan portfolios, foreign exchange portfolios and se curities portfolios.
The researcher Chowdhury (1993) studied the impact of exchange rate volatility
on trade flows of the G7, using an error correction model, noting that there is a strong
and negative correlation between the two variables.
Duncan Wilson (1995), through the study undertaken stressed VaR applicability
in operational risk control.
Boudoukh (1998), after using the historical VaR, ha s concluded that the new
information on future risks have a much higher info rmational content than the former
ones, which is the why it is considered proper to t ake into consideration the share of P/L
values (profit / loss), depending on their maturity (the newer they are, the higher their
share should).
In the years 2000 and 2001, studies conducted showe d significant differences
resulting from the use of different methodologies f or calculating VaR. Therefore,
Alexander and Baptista have studied the use of the Mean VaR method compared to the
variance z covariance method. Whereas the applicati on of VaR implies assuming some
restrictions, they were considered by Gourieroux an d Manfort (2001) in the study on the
portfolios’ effectiveness analysis.
According to the researcher Medova, a lowzfrequency event has a low
probability of generating operational risk.
The study conducted by experts of the International Monetary Fund (2004)
considered closely the factors affecting the exchan ge rate volatility, focusing on the
structural characteristics of the foreign exchange regime.
The researchers’ conclusions (CanaleszKriljenko, H abermeier K., 2004) showed
that decentralized markets, regulations on the use of national currency by nonzresidents,
accepting limits on currency positions of banks are associated with reduced exchange
rate volatility.
VaR method is easily transmissible to the board of a company in the context of
the assumption of a normal daily distribution of th e return on assets (Fen Ying C.,
2010).
Simone Varotto (2011) indicated that losses associa ted with market risk are ten
times higher than the cost of additional risk (taki ng into account the characteristics of
portfolios).
Although VaR method to estimate resulting loss of t he currency portfolio is
popular and widely used at commercial banks, many r esearchers recommending the
CVaR (ConditionalzVaR) method, arguing the ease in usage and in determining those
strategies to optimize the various portfolios.
Studies conducted by AlfarozCid, BaixaulizSoler, Fe rnándezzBlanco (2011)
demonstrated that the use of variance provides the same results as CVaR model.
Canova and Ciccarelli (2009, 2013) have shown that in the panel VaR models
statistical and dynamic interrelations are captured simultaneously, making it possible to
take into account a dynamic heterogeneity of the da ta series.
The applicability of VaR method in the Romanian fin ancialzbanking system was
studied by researchers such as:
z Trenca I Mutu S, Dezsi E (2011) on the share pric e of 6 companies listed on
the Bucharest Stock Exchange (Erste Group Bank, SIF Banat Crișana, SIF Page 3 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review OnlyMoldova, SIF Transilvania, SIF Muntenia, SIF Olteni a), by the comparative
usage of the different techniques for determining t he VaR;
z Trenca I., Mutu S (2011) on interbank interest ra tes;
z Trenca I., Pece A, Mihuț I. showed that losses in the portfolio are determined
mainly by increased volatility in the currency mark et;
z Bobeică G, Țimurlea M., (2012) following the mode lling of a series of data
EUR/RON, for a period of five years from 2007 to 20 12 (data on a daily basis)
showed that at the level of a portfolio consisting of euro currency, the maximum
loss was of 2.4%, on a time horizon of 10 days, at a significance level of 1%;
z Armeanu D. and Bălu F., (2006) estimated VaR reco rded in the portfolio of
currencies held by a commercial bank for two years, while emphasizing that
final VaR is lower than the individual VaR of a por tfolio of currencies, therefore
considering the correlation coefficients between th em.
An actual calculation of VaR implies the reference to two parameters, namely:
z time horizon (h) for which the risk assessment is carried out;
z the confidence coefficient (p) or risk tolerance coefficient (1z p) for which the
risk is calculated.
In addition to the recommendations of the Basel Com mittee, which involves a
time horizon of 10 days and a probability of 99% co nfidence, the methodology
developed by the Risk Metrics department is also re commended at which time horizon
shrinks to 1 day, and the probability of confidence drops to 95%.
After 1993, commercial banks have developed their o wn methods of assessing
market risk, the selection of either of them being carried out according to the purpose of
research, and the merits and disadvantages of each:
a. Parametric method (the method variancezcovariance/a nalytic VaR) which
requires making three assumptions, namely: a normal distribution of returns of a
portfolio’s values’ instruments; constant maintenan ce of initial shares granted to
instruments of the portfolio; completely ignoring t he presence of “fat tails” in
the distribution of probability. The main drawback of the method is derived from
the actual statistical hypothesis, according to whi ch financial asset’s price
development shows a normal distribution (rare case in practice). Moreover, the
method is not recommended for portfolios containing discontinuous payoff
(barrier options);
b. Historical simulation method which, unlike the para metric method, makes no
assumption regarding the distribution of profitabil ity, only empirical distribution
being used (resulting from the analysis of the data series). The method has
disadvantages in terms of estimating future values that are based on the past
ones and also in terms of dependency of the data se t used. For example, in the
situation of a determination of VaR on unusually ca lm markets (or volatile), but
whose conditions were later changed, historical sim ulation is likely to generate
too small estimates of the VaR for current risks. H istorical simulation has
difficulties in terms of taking into consideration the changes in market
developments that occurred in the period considered and the VaR measures
obtained through historical simulation do not captu re the risk associated to
plausible events in the future, but which have not happened in the past.
c. Monte Carlo Simulation Method, showing great flexib ility compared to the other
two methods, in terms of the portfolio’s return dis tribution, which is obtained by
generating various scenarios for risk factors consi dered.

3. Methodology of research: Page 4 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only For estimating VaR, researcher Down (2002), propose s determining the
resulting loss or gain, as the difference between t he original value of the portfolio and
the value expected after h days.

∆W h = W 07W h, (1)
In which:
W0 – initial value of the portfolio;
Wh z portfolio value after „h” days
∆W h – loss/profit resulted after h days (W 0<W h – indicates registration of loss in
the foreign currency portfolio after h days).
Recommendation of Basel Committee shall require VaR by applying the
following model:
VaR i=7W i,0 *α*σ i*√/g2190, (2)
Where:
Wi,0 – net foreign currency position for the currency co nsidered in RON
equivalent;
α – confidence coefficient;
σi – daily volatilityof currency;
h – withholding period (set timeframe).
From this model, it can be concluded that an increa se in portfolio volatility will
cause the flattening of the curve (distribution of returns), therefore resulting in an
increase of the VaR and the increase of confidence sets an increase in VaR.
Romanian supervisory authorities (NBR Regulation no . 5/2013 and the NBR
Regulation no. 22/2011) and the European supervisor y authorities (Regulation no.
575/2013 of the European Parliament on prudential r equirements for credit institutions,
Basel III) have decided to use within the model, th e correlation between the confidence
coefficient and the probability of simulation. Thus , as recommended by Basel III, at a
probability of 99%, it is appropriate to use a conf idence coefficient of 2.33, while by the
recommendation of the Risk Metrics department, at a probability of 95%, it is indicated
to use a confidence coefficient of 1.65.
Backztesting plays an important role in validating the performance of the model
and procedures for risk, although, by using backzte sting errors may occur, such as:
incorrect historical data/ estimated incorrectly pa rameters, incorrect estimation of
standard deviation of portfolio or of excessive non linearity, abnormal distribution of
returns (generating an incorrect VaR) etc.
The primary objective of this study is to identify the maximum estimated loss of
exposures denominated in euro currency to commercia l banks listed on the Bucharest
Stock Exchange. Also, starting from foreign currenc y portfolio structure, we intended to
identify the most vulnerable commercial bank in ter ms of EURO/RON rate of exchange
fluctuation.
As already mentioned, the study of exchange rate vo latility was conducted on a
set of data on a daily basis (757 observations, the equivalent of the 757 working days in
the period 01.01.2012 z 31.12.2014). The database w as developed, starting from the
value of net recorded positions of the four banks, registered on 31.12.2014 and the
amounts of EUR/RON exchange rates communicated by t he National Bank of Romania.
In determining VaR, the historical simulation metho d was used and data
processing and value at risk calculation was perfor med using the statistical and
mathematical software Excel. In this study, we used the method proposed by the Basel
Committee (approved by the National Bank of Romania ), respectively: 99% probability,
associated with a confidence level of 2.33. Page 5 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review OnlyCarrying out the research involves taking the follo wing steps:
1. Taking from the annual financial statements of t he four commercial banks the
value of the euro currency position (long or short position);
2. Identifying the daily EUR/RON exchange rate, in the period 01.01.2012z
31.12.2014 period (757 days working);
3. VaR determination, namely:
a. Daily return calculation, according to the model :
Ri,t = ln /g3017/g2919,/g2930
/g3017/g3036,/g3047/g2879/g2869;

Based on the daily rates of exchange, determining t he continuous daily
return(757 returns calculated with a value lower th an 1);
b. Determination of volatility expressed as standar d deviation (Standard
Deviation) for the euro currency, based on the foll owing models:

σi=√∑/g4666/g3019/g3036/g2879/g3019/g3040/g3036 /g4667/g2870/g3289
/g3284/g3128/g3116
/g3021/g2879/g2869; Rm i=∑/g3019/g3036/g3047 /g3289
/g3284/g3128/g3116
/g3021

Where:
σi – daily volatility of currency;
Ri – daily return;
Rmi – daily average return;
T – time period.

c. Determination of the final VaR value, by applyin g the model:
VaR i=zW i,0 *α*σ i *√ℎ,

Limitations of the research:

The research could not be extended to other currenc ies, given the fact that banks
did not report the structure of the balance sheet a nd net positions denominated in
currencies in which most operations were done (EUR and USD). As for the commercial
banks included in the survey, most operations were denominated in euros. Net foreign
currency position denominated in USD could not be b rought into the study whereas at
BRD it has not been included in the financial state ments. For this reason, in conjunction
with the desire to ensure the maintenance of homoge nous data, it was decided to study
the maximum loss that the four banks may incur, due to variation of the RON/EURO
rate of exchange.

4. Results of the research:

The share of foreign currency positions in euro in total foreign currency position
on 31.12.2014 is presented in table no. 1.

Table 1 – The share of foreign currency positions o n the euro component at 31.12.2014
Commercial Bank Currency position in euro
at 31.12.2014 Total currency
position at
31.12.2014 Share of foreign
currency position in
total foreign currency
position at
31.12.2014
0 1 2 3=1/2 x 100
Transylvania Bank 159.250 3.385.549 4,70% Page 6 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only(thousands of lei)
Romanian
Development Bank
(thousands of lei) 3.957.863 7.915.725 50,00%
Romanian
Commercial Bank
(thousands of lei) 19.057 52.225 36,50%
Carpatica Commercial
Bank (lei) 23.396 190.061 12,31%
Source: Banks’ financial reporting at 31.12.2014

The analysis of the foreign currency position’s str ucture lead us to the
conclusion that if commercial banks holding equity majority foreign (Romanian
Commercial Bank SA, Romanian Development Bank SA), the share of foreign currency
position denominated in euro is clearly higher than that recorded by the commercial
banks with full Romanian capital (Banca Transilvani a SA, Banca Carpatica SA).
At the end of 2014, the four commercial banks inclu ded in the research reported
the following net position on euro currency (table no. 2):

Table no. 2z Net currency position at 31.12.2014
Banking
institution Net currency position
(thousands of RON) RON /EURO rate of exchange at 31.12.2014
BCC 23.396 4,4821
BRD 3.957.863 4,4821
BCR z19.057 4,4821
BT z159.250 4,4821
Source: Banks’ financial reporting at 31.12.2014

It appears that Banca Transilvania SA (Transylvani a Bank SA) and Banca
Comercială Română (Romanian Commercial Bank SA) hol d a short foreign currency
position, while Banca Comercială Carpatica SA (Carp atica Commercial Bank) and
Banca Română de Dezvoltare SA (Romanian Development Bank SA) recorded at the
end of 2014 a long foreign exchange position. In ac cordance with NBR Norm no
9/1993, holding of a long foreign exchange position indicates that the volume of claims
exceeds commitments undertaken by them. Prudential rules recommends that at the end
of a working day, total long foreign currency posit ion does not exceed 25% of own
funds, while short foreign currency position reache s a maximum of 10% of own funds.
Over the period analysed, the evolution of RON/EURO was relatively steady,
given that it ranged between 4.3072 lei/euro and 4. 6481 lei/euro, the maximum
appreciation of RON/EURO over three years being 8%.
Daily evolution of VaR and the descriptive statisti cs of data series for the period
January 1 st , 2012zDecember 31 st , 2014 at the Romanian Commercial Bank SA z a
member of Erste Group z (the first largest bank in the Romanian banking system,
appreciated by assets) is shown in figure no. 1.
Page 7 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only
Descriptive Statistics
Mean z114,48
Standard Error 0,06
Median z114,40
Mode z114,23
Standard Deviation 1,58
Sample Variance 2,49
Kurtosis 70,110768519
Skewness 70,36907897
Range 8,79
Maximum z119,85
Minimum z111,06
Sum z86662,02
Count 757
Figure no. 1– Daily VaR evolution at BCR and descri ptive statistics, period01.01.2012z31.12.2014
Source: author’s calculations based on the foreign currency position and daily profitability rate
developments

From figure no.1, a bimodal distribution of daily VaR can be noticed, indicating
the grouping of frequencies into two modal classes, separated in their turn from classes
with a low frequency. The relationship resulting fr om the descriptive statistics
concerning the series of data, between mean, median and mode indicates a normal
distribution. The value obtained for theindex Kurto sis (<3) shows that the distribution is
also platykurtic, indicating a lower probability fo r the extreme values than the normal
distribution. Even if the mean of the data series i s not equal to its module, a left
asymmetry/skewness can be noticed, suggesting that most of the extreme values are
oriented to the left, showing the appreciation tren d of the RON/EUR rate of exchange.
The analysis of daily VaR descriptive statistics in dicates that at BCR, for a period of
three years, a daily average loss of 114 480 lei is estimated, with a minimum of 111.06
thousands of lei and a maximum of 119.85 thousands of lei. By weighting the daily
maximum loss with the value recorded after applying the radical in the time of
withholding the structure of foreign currency posit ion (10 days), the maximum loss
resulting is of 378.99 thousand lei , representing 1.9887% of the foreign currency
position (exposure the euro currency). Through the 757 simulations, the daily loss with
the highest frequency was identified (5 appearances ), of 114.23 thousand lei, while the
Standard Deviation indicator’s value of 1.58 shows avery small error probability of the
estimator.
In order to determine VaR at Banca Transilvania, th e standard deviation of profit
or loss arising on the euro component was initially calculated, using the historical
simulationmethod. Evolution of daily VaR with a rel evance of 1%, determined over a
period of three years is shown in figure no. 2.
z122 z120 z118 z116 z114 z112 z110 z108 z106 2012 2013 2014 Daily VaR Page 8 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only

Descriptive Statistics
Mean z957,28
Standard Error 0,48
Median z956,62
Mode z955,24
Standard Deviation 13,19
Sample Variance 173,97
Kurtosis 70,11
Skewness 70,37
Range 73,50
Maximum z1002,17
Minimum z928,67
Sum z723704,21
Count 756
Confidence Level (95,0%) 0,94 z1020.00000 z1000.00000 z980.00000 z960.00000 z940.00000 z920.00000 z900.00000 z880.00000 2012 2013 2014 Daily VaR Page 9 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review OnlyFigure no. 2– Daily VaR evolution at Banca Transilv ania and descriptive statistics, period
01.01.2012z31.12.2014
Source: author’s calculations based on the foreign currency position and daily
profitability rate developments

From figure no. 2, a bimodal distribution of daily VaR can be noticed,
indicating the grouping of frequencies into two mod al classes, separated in
their turn from classes with a low frequency. The r elationship resulting from
the descriptive statistics concerning the series of data, between mean, median
and mode indicates a normal distribution. The value obtained for the index
Kurtosis (<3) shows that the distribution is also p latykurtic, indicating a
lower probability for the extreme values than the n ormal distribution. Even if
the mean of the data series is not equal to its mod ule, a left
asymmetry/skewness can be noticed, suggesting that most of the extreme
values are oriented to the left, showing the apprec iation trend of the
RON/EUR rate of exchange . The analysis of daily VaR descriptive statistics
indicates that at Transylvania Bank, for a period o f three years, a daily
average loss of 957.28 thousands of lei is estimate d, with a minimum of
928.67 thousands of lei and a maximum of 1002.17 th ousands of lei. By
weighting the daily maximum loss with the value rec orded after applying the
radical in the time of withholding the structure of foreign currency position
(10 days), the maximum loss resulting is of 3.169,15 thousand lei ,
representing 1.99% of the foreign currency position (exposure the euro
currency). Through the 757 simulations, the daily l oss with the highest
frequency was identified (5 appearances), of955.24 thousand lei, while the
Standard Deviation indicator’s value of 13.19 shows a very small error
probability of the estimator.
Banca Comerciala Carpatica SA, a mediumzsmall sized bank with a
Romanianshareholding, considering its nature, regis tered a negative gap on
the foreign currency component (the volume of recei vables is lower than the
volume of the debts in euro currency). To determine the maximum loss over
a period of 10 days, the historical VaR method was used. The summary of
daily VaR evolution is shown in figure no. 3.
Page 10 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only

Figure no. 3– Daily VaR evolution at BCC and descri ptive statistics, period01.01.2012z31.12.2014
Source: author’s calculations based on the foreign currency position and daily profitability rate
developments

From figure no. 3, a bimodal distribution of daily VaR can be noticed, indicating
the grouping of frequencies into two modal classes, separated in their turn from classes
with a low frequency. The relationship resulting fr om the descriptive statistics
concerning the series of data, between mean, median and mode indicates a normal
distribution. The value obtained for the index Kurt osis (<3) shows that the distribution
is also platykurtic, indicating a lower probability for the extreme values than the normal
distribution. Even if the mean of the data series i s not equal to its module, a left
asymmetry/skewness can be noticed, suggesting that most of the extreme values are
oriented to the left, showing the appreciation tren d of the RON/EUR rate of exchange .
The analysis of daily VaR descriptive statistics in dicates that at Carpatica Bank, for a
period of three years, a daily average loss of 140. 544,77lei is estimated, with a
minimum of 136.344,70 lei and a maximum of 147.135, 91 lei. By weighting the daily
maximum loss with the value recorded after applying the radical in the time of
withholding the structure of foreign currency posit ion (10 days), the maximum loss
resulting is of 465.281,61 lei , representing 1.9887% of the foreign currency posi tion
(exposure the euro currency). Through the 757 simul ations, the daily loss with the
highest frequency was identified (5 appearances), o f140.244,60 lei, while the Standard
Deviation indicator’s value of 70.43 shows a very s mall error probability of the
estimator.
Romanian Development Bank – Groupe Societe Generale SA held at the end of
financial year 2014 a short foreigncurrency positio n on their foreign currency portfolio
in euroscomponent. Although it recorded a positive gap on the euro currency, the
objective proposed in the research was to identify the maximum loss the bank may feel,
due to the structure of its loan portfolio. Therefo re, the first step considered determining
and analysing the distribution of the daily rates o f return. The evolution of the exchange
rate yield is summarized in the chart no. 4.

z148000 z146000 z144000 z142000 z140000 z138000 z136000 z134000 z132000 z130000 2012 2013 2014 Daily VaR Descriptive Statistics
Mean z140544,77
Standard Error 70,43
Median z140447,20
Mode z140244,60
Standard Deviation 1936,47
Sample Variance 3749911,11
Kurtosis 70,11
Skewness 70,37
Range 10791,21
Maximum z147.135,91
Minimum z136.344,70
Sum z106251843,84
Count 756
Confidence Level (95,0%) 138,26 lei Page 11 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Only
Mean z23775,72
Standard Error 11,91
Median z23759,21
Mode z23724,94
Standard Deviation 327,59
Sample Variance 107314,64
Kurtosis 70,11
Skewness 70,37
Range 1825,53
Maximum z24890,73
Minimum z23065,20
Sum z17974444,25
Count 756
Confidence Level (95,0%) 23,39

Figure no. 4– Daily VaR evolution at BRD and descri ptive statistics, period 01.01.2012z31.12.2014
Source: author’s calculations based on the foreign currency position and daily profitability rate
developments

As with the other banks subject to the research, fi gure 4 shows the existence of a
bimodal distribution of daily VaR, indicating the g rouping of frequencies into two
modal classes, separated in their turn from classes with a low frequency. The
relationship resulting from the descriptive statist ics concerning the series of data,
between mean, median and mode indicates a normal di stribution. The value obtained for
the index Kurtosis (<3) shows that the distribution is also platykurtic, indicating a lower
probability for the extreme values than the normal distribution. Even if the mean of the
data series is not equal to its module, a left asym metry/skewness can be noticed,
suggesting that most of the extreme values are orie nted to the left, showing the
appreciation trend of the RON/EUR rate of exchange. The analysis of daily VaR
descriptive statistics indicates that at the Romani an Development Bank, for a period of
three years, a daily average loss of 23.777,72 thou sands oflei is estimated, with a
minimum of 23.065,20 thousands of lei and a maximum of 24.890,73 thousands of lei.
By weighting the daily maximum loss with the value recorded after applying the radical
in the time of withholding the structure of foreign currency position (10 days), the
maximum loss resulting is of 78.711,41 thousand lei , representing 1.9887% of the
foreign currency position (exposure the euro curren cy). Through the 757 simulations,
the daily loss with the highest frequency was ident ified (5 appearances), of23.724,94
thousands of lei, while the Standard Deviation indi cator’s value of 327,59 shows a very
small error probability of the estimator (this conc lusion is also supported by the large
number of observations).
The comparative summary of results for the four com mercial banks listed on
BSE indicates that the estimated maximum losses ove r a period of 10 days may be
administered accordingly, not exceeding the 2% of t he foreign currency portfolio.
Although the estimated maximum loss amounts for a p eriod of 10 days are not very
different, the biggest loss is expected to be regis tered at Banca Transilvania SA. In
interpreting this result, we should bear in mind th at, in the total foreign currency
position recorded at this bank, the position denomi nated in RON is 96%, losses being
assessed as having low risk on the bank. The growth trend of the exchange rate and
national currency depreciation has resulted from th e distribution of daily frequencies, as
well as from the resulting skewness coefficient’s v alue. z25500 z25000 z24500 z24000 z23500 z23000 z22500 z22000 2012 2013 2014 Daily VaR Page 12 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review OnlyFrom the research, we have drawn the conclusion acc ording to which holding a
negative foreign currency position does not financi ally unbalance a commercial bank,
given that VaR results show a number that offers ba nks the possibility of an adequate
management of foreign currency risk.
In figure no. 5 there are captured some similaritie s and differences between the
four commercial banks included in the survey on for eign exchange management and
risk assessment.

Fig. no. 5 Similarities and differences between the four banks on foreign currency risk
assessment

5. Conclusions
Although the share of total euro currency in the to tal foreign currency portfolio
of the four commercial banks studied is different, the maximum loss recorded does not
exceed 2% of the total foreign currency portfolio.
In the studied period, the euro currency has underg one appreciations in relation
to RON currency, therefore highlighting the volatil ity of markets and the lack of
financial stability. However, we consider that losses are not likely to unbalance
financially any of the four banks, which demonstrat es the effectiveness of foreign
currency portfolio management. Nonetheless, a continuation of the growth trend of the
RON/EURO exchange rate may be liable to deepen the maximum estimated loss,
indicating the need to increase capital requirement s within these banks and a greater
caution in selecting the techniques to mitigate for eign currency risk.
Our future research will consider extending the his torical period, estimating VaR
through Monte Carlo simulation method and subsequen tly comparing the results
obtained by the two methods.

Selective references

z Alexander, G.J. and A.M. Baptista, „Economic Implic ations of Using a Meanz
VaR Model for Portfolio Selection: A Comparison wit h MeanzVariance
Analysis”, Working Paper, University of Minnesota. 2000 (December) bimodal and platykurtik distribution
of the four banks' rates of return;
indicating a trend of appreciation of
the exchange rate and recording a
negative impact on the foreign
currency portfolios;
estimated losses are at a manageable
level, not posing a problem to any of
the banks;
low error probability of the estimator.BT shows the highest percentage of
loss of the net position held by the
bank concerned;
in the total sum, the highest estimated
loss is expected to be registered at
BRD, this bank also showing the
highest exposure to currency risk
(50% of its foreign exchange
position)Page 13 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Onlyz AlfarozCid, E., BaixaulizSoler, J. S., & Fernándezz Blanco, M. O., „Minimising
valuezatzrisk in a portfolio optimisation problem u sing a multizobjective genetic
algorithm”, International Journal of Risk Assessmen t and Management, 5(5/6),
2011.
z Armeanu D., Bălu F.z O., „Aplicarea metodologiei Va R portofoliilor valutare
deținute de bănci”, Economie teoretică și aplicată, Vol no 2/2007, 2007.
z BaixaulizSoler, J. S., AlfarozCid, E., & Fernándezz Blanco, M. O., „Several risk
measures in portfolio selection: Is it worthwhile?” , Spanish Journal of Finance
and Accounting, 39(147), 2010.
z BaixaulizSoler, J. S., AlfarozCid, E., & Fernándezz Blanco, M. O., „MeanzVaR
portfolio selection under real constrains”, Computa tional Economics, 37(2),
2011.
z Boudoukh, J.,M. Richardson și R. Whitelaw, „Money, banking and the Financial
System”, West Publishing Company, 1998
z CanaleszKriljenko, Habermeier K., „Structural Facto rs Affecting Exchange Rate
Volatility: A cross section study”, IMF Working Pap er no 147, 2004.
z Canova F., Ciccarelli. M, „Estimating Multicountry VaR Models”; International
Economic Review, 50(3), 2009.
z Chowdhury, A., „Does Exchange Rate Volatility Depre ss Trade Flows Evidence
From ErrorzCorrelation Models”, The Review Of Econo mics And Statistics,
1993.
z Duncan Wilson, „VAR in Operation. Risk”, December ( 1995).
z Down, K., „Measuring Market Risk”, John Wiley of So ns, Chicester, 2002.
z FenzYing C., „Exchange Rate Risk and Jump Risk”, Hi ndawi Publishing
Corporation, Journal of Probability and Statistics, 2010.
z Gourieroux C. and A. Monfort. Pricing with Splines. Working Paper. 2001.
http://www.istfin.eco.usi.ch/seminarzpaperszgour.pd f .
z Medova,E. „Operational Risk Capital Allocation and Integration of Risks: Firmz
wide Issues for Financial Institutions”, Risk Books (ed.), London, 2001.
z Moinescu B., Codîrlașu A., „Strategii și instrument e de administrare a riscurilor
bancare”, Editura ASE, 2009.
z Trenca I., Mutu S., Dezsi E., „Advantages and Limit ations of VaR Models Used
in Managing Market Risk in Banks”, Finance z Challe nges of the Future, 2011,
Vol. 1, Issue 13, indexed in Research Papers in Eco nomics.
z Trenca I., Mutu S., Petria N., „Econometric Models Used For Managing The
Market Risk In The Romanian Banking System”,Analele Științifice ale
Universității "Alexandru Ioan Cuza" din Iasi, 2011, Vol. 2011 SE, Issue July,
indexed in Research Papers in Economics.
z Trenca I, Pece A, Mihut I., „The assessment of mark et risk in the context of the
current financial crisis”, Procedia Economics and F inance 32, 2015.

Bank legislation:
z Basel III – approved by the Basel Committee in Dece mber 2010, revised in June
2011 – available online: www.bis.org ;
z Regulation 575/2013 – on the prudential requirement s for credit institutions and
investment companies – published in the Official Jo urnal of the European Union
L176/1 of 27.06.2013;
z NBR Regulation No. 5/2013 on prudential requirement s for credit institutions z
published in the Official Gazette, Part I, No. 841 on 30.12.2013; Page 14 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

For Peer Review Onlyz OUG 99/2006 – on credit institutions and capital ad equacy z published in the
Official Gazette Part I, No. 1027 on 27.12.2006, as amended and supplemented.

Electronic resources:
z www.bancatransilvania.ro
z www.brd.ro
z www.carpatica.ro
z www.bcr.ro
z www.bnro.ro
z www.bis.org

Page 15 of 15
URL: https://mc.manuscriptcentral.com/rero E-mail: jsetic@unipu.hrEconomic Research-Ekonomska Istra?ivanja
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60