IV. Special Research of Financial Stability. 8. Methods of Financial Stability Risks Assessment Risk Assessment Map

Transcription

1 IV. Special Research of Financial Sabiliy 8. Mehods of Financial Sabiliy Risks Assessmen Risk Assessmen Map To ensure financial sabiliy, regular monioring and analysis of various vulnerabiliy facors and risks for financial sabiliy shall be provided on basis of a wide lis of macroeconomic and prudenial indicaors. For his purpose Risk Assessmen Map 1, an analyical approach o complex assessmen of financial sabiliy risks, was developed. The Risk Assessmen Map is a schemaic image of a level of risks on quaniaive parameers for a cerain period, which also allows o analyzing in wha way a risk level has changed from he momen of he previous period. Moreover, he Risk Map is a saring poin for profound analysis of risks affecing on financial sabiliy. The Risk Map is creaed on basis of 57 indicaors in view of mehods o assess financial sabiliy indicaors, macroprudenial indicaors, indicaors defining financial crises probabiliy, including macroeconomic indicaors and differen indicaors of global, banking and real economic secors. In he Risk Assessmen Map he indicaors 2 are grouped under 8 parameers, as for heir major feaures: 1. global world (exernal facors) 5. deb burden 2. adequae reserves of he counry 6. liquidiy and sensiiviy of banking secor o he marke risks 3. capial flows 7. qualiy of banking secor asses (credi risks) 4. moneary and fiscal policy 8. profiabiliy of banking secor To avoid he represenaion of he indicaors which idenically characerize a risk source in he sysem, inerrelaion of indicaors wihin he formed groups was checked and indicaors posiive correlaed wih each oher were excluded. This arge was achieved by means of a correlaion marix. Indicaors of formed groups are presened below and a corresponding mark shows wha behaviour of an indicaor demonsraes increase in risks and heir vulnerabiliy facors. Descripion of Risk Assessmen Map parameers 1. Global world. Indicaors of his group allows o esimae probabiliy of negaive impac of exernal environmen facors which may lead o sharp correcion a he domesic marke of he counry in view of curren condiions a he inernaional capial and commodiies markes as well as an esimae of conagion effec probabiliy of difficulies from major rading couner-parners. Global world Forma Mark Inernaional ineres raes (LIBOR 3 monh) change (year o year) + World oil prices change (year o year) - Terms of rade change (year o year) - Indusrial producion of OECD counries change (year o year) - Real GDP of Russia change (year o year) - 2. Adequae reserves of he counry. This parameer includes indicaors assessing probabiliy of decrease in counry s solvency capaciy. Adequae reserves of he counry Forma Mark Coverage of commercial banks foreign liabiliies wih foreign asses raio - of NBRK (including asses of he Naional Fund) Ne foreign asses of NBRK o money supply raio raio - 1 The mehodology has been developed by he Financial Sabiliy Division of he Naional Bank of he Republic of Kazakhsan. Presened Risk Assessmen Map is no final; i will be furher improved wih mehodological approaches refined. 2 The indicaors are presened in differen formas such as absolue value, raio, change a year, rend deviaion and regression model residual argumen. For mos of he indicaors a hisorical series sars from

2 Gross inernaional reserves o impor raio change (year o year) - Raio of inernaional reserves o forhcoming paymens o discharge raio - and service foreign deb 3 3. Capial flows. Indicaors assessing a risk of decrease in capial inflow o he counry economy. Capial flows Forma Mark Curren accoun balance o GDP raio raio - FDI and curren accoun balance o GDP raio raio - Real effecive exchange rae 4 rend deviaion + Expor of goods and services change (year o year) - Impor of goods and services change (year o year) + Real GDP 5 change (year o year) + Capial accoun absolue value - Raio of Errors and Omissions and shor-erm capial o GDP raio - 4. Moneary and fiscal policy. Assessmen of risk of he policy efficiency conduced by public auhoriies. Moneary and fiscal policy Forma Mark Public expenses o GDP raio raio + Fiscal budge balance o GDP raio raio - Credis o economy o GDP raio change (year o year) + Excess money supply in real erm 6 esimae + Inflaion change (year o year) + Money muliplier change (year o year) + 5. Deb burden. Indicaors which allow assessing a risk of increase in dependency on foreign and domesic deb capial and is srucure. Deb burden Forma Mark Raio of foreign deb of he counry o GDP raio + Share of commercial banks liabiliies in he gross foreign deb raio + Share of public deb in he gross foreign deb raio + Public deb o GDP raio raio + Raio of liabiliies o equiy capial of corporae secor raio + Raio of equiy capial o asses and off-balance liabiliies weighed a a risk level of banking secor raio - 6. Liquidiy and sensiiviy of banking secor o he marke risks. Probabiliy assessmen of losses of financial insiuions in liquidiy decreases and foreign exchange and ineres rae risks increase siuaion. Liquidiy and sensiiviy of banking secor o he marke risks Forma Mark Spread beween he highes and lowes inerbank offered rae absolue value + Real deposi ineres rae change (year o year) + Differenial beween domesic and foreign nominal raes 7 absolue value - 3 I s calculaed as raio of inernaional reserves, excluding he gold, o shor-erm deb, plus discharge and service paymens for a long-erm foreign deb for preceding wo years, on average. 4 Increase in real exchange rae index denoes appreciaion of naional currency, and decrease - reducion in is value. To calculae a rend of real effecive exchange rae, Hodrick-Presco filer was used. 5 Real GDP indicaor wih a posiive mark means a change in aggregae demand for impor. 6 Indicaor is defined as deviaion of assessed money demand from observed money supply (expressed as a share of money supply in GDP), i.e. as residuals of he below regression equaion: M GDP = a 0 + a 1 Y + a 2 p + a 3 where M money supply, GDP nominal GDP, Y GDP (in real erms), р CPI, ime. + ε 80

3 STB credis o STB deposis raio change (year o year) + STB deposis (in real erms) change (year o year) - Credis of cenral bank o he banking sysem change (year o year) + Highly liquidiy asses o oal asses raio raio - Highly liquidiy asses o shor-erm liabiliies raio raio - Foreign currency liabiliies o oal liabiliies raios raio + Bank-o-bank liabiliies change (year o year) - Ne open posiion in FX (in module) o equiy capial raio raio + Volailiy of exchange rae of enge o US dollar 8 absolue value + 7. Qualiy of banking secor asses. Probabiliy assessmen of financial insiuions losses as a resul of defaul on liabiliies by heir couner-parners and cliens. Qualiy of banking secor asses Forma Mark Non performing loans 9 o oal loans raio raio + Morgage credis o oal credis raio raio + Consumer credis o oal credis raio raio + Foreign currency credis o oal credis raio raio + Sae secor credis o GDP raio raio + Trade secor credis o oal credis raio raio + Indusrial secor credis o oal credis raio raio + Consrucion secor credis o oal credis raio raio + Reurn on equiy of corporae secor absolue value - 8. Profiabiliy of banking secor. Indicaors assessing a risk of losses by financial insiuions in case of decrease in yield and efficiency of heir asse managemen. Profiabiliy of banking secor Forma Mark Ne ineres income o gross income raio raio - Non-ineres expenses o gross income raio raio + Operaional expenses o gross income raio raio + Reurn on asses absolue value - Differenial beween credi and deposi ineres raes absolue value - In he Risk Assessmen Map indicaors are ranked according o he risk groups. The risk groups are formed on basis of a percenile scale which was creaed in view of hisorical values of each indicaor by ascending / descending soring in accordance wih he direcion of deerioraion of such indicaor. Range limis of he risk groups are calculaed as 25%, 50%, 75% and 100% percenile of indicaor s value disribuion. Thus, any change of indicaor s value o he increase of risk and vulnerabiliy will demonsrae a high risk level wihin an inerval beween he values of 75% and 100% percenile of disribuion and vice versa. Thus, he firs risk group (75%-100%) denoes high risk; he second risk group (50%-75%) - moderae risk; he hird risk group (25%- 50%) insufficien and he fourh risk group (less han 25%) low risk. A binary sysem is used o deermine if a curren indicaor eners in any risk group, i.e. if a value of such indicaors is wihin he limis of he firs risk group, i s equal o 1 and 0 for oher groups. Then he indicaors values are summed up for each of he risk group and a resuling amoun is weighed in he following way. Amoun of values from he firs and fourh risk groups is weighed by 1, and amoun from he moderae and insufficien groups - 0,5. As a resul, he final value of he risk level is calculaed by subracion of he amoun of values of he firs and second groups, from amoun of he hird and fourh risk groups. 7 Differenial beween average weighed raes for credis in naional currency adjused for change in exchange rae, and LIBOR 3 monh in US dollars. As agains an early warning sysem (Secion 9), he indicaor deerioraion is assessed when a differenial reduce, i.e. demonsraes growh of ineres expenses on deb servicing as compared wih ineres income. 8 Indicaor of exchange rae risk assessmen, calculaed as a sandard deviaion of exchange rae of enge o US dollar. 9 Amoun of loans classified as doubful loans of caegories 2, 4, 5 and bad loans. 81

4 Values for each parameer of he risk source are normalized as is share in number of oal indicaors of each parameer ha le o sandardize differences in number of indicaors in each subgroup of he risk source in he Risk Assessmen Map. The Risk Assessmen Map enables no only o assess a curren risk level on various parameers of heir sources bu also o rack changes in condiions for formaion of he risks as compared wih a previous period. Moreover, every ime a new period is added o he assessmen range of limi values of he risk groups. For example, i is possible o compare a risk source level change in he second quarer of 2007 wih he end of ha provides easy visual demonsraion of changes for a half-year. 82

5 9. Early Warning Sysem of Financial Crisis for Kazakhsan Today he issues of mainenance of financial sabiliy in he world have a special imporance. Severe consequences of financial crises for sable economic developmen make people non only undersand heir naure and origins bu develop cerain approaches o deec early facors of vulnerabiliy of he counry o financial crises, wha pracically can be achieved by creaing differen early warning sysems of financial crises. As a mehodical basis of he early warning sysem for boh currency and banking crises had been chosen a signal approach 10. The signal approach for early warning sysems 11 implies monioring of dynamics of cerain economic variables which end o unusual sysemaic behavior during a precrisis period. Deviaion of any variable from is normal level is considered as a signal of a poenial financial crisis which may occur during a cerain fuure period. The signal approach as a mehodical basis was preferred because predicive power assessmen of each indicaor is provided on individual basis ha allows o range and find a signaling economic variable which is a poenial source of he vulnerabiliy facor. Analysis of reasons for he variable signaling is one of he mos imporan informaion blocks o decide on and choose any applicable prevenive and correcive measures of he naional economic policy. Fundamenals of Financial Crises Early Warning Sysem Sage 1. Idenificaion of Crisis Siuaion: Exchange Marke Pressure Index In empirical researches a concep of a so-called speculaive pressure is applied for purpose of deerminaion a currency crisis. According o he concep, a crisis as well as a speculaive aack is defined as a period when speculaive pressure a he foreign exchange marke reached is criical values. Exchange Marke Pressure Index (hereinafer referred o as EMPI ) used o assess sabiliy of a naional currency exchange rae is consruced on basis of rae of change of he exchange rae and inernaional reserves as agains o a previous period: EMPI 1) (( e / e ) 1) ( σ / σ ) ((Re s / Re s ) 1 (1) ( = 1 e Re s 1 As an alernaive, he EMPI is consruced on basis of changes per year: EMPI ( 2) = (( e / e 12 ) 1) ( σ e / σ Re s ) ((Re s / Re s 12 ) 1 (2) where: e - an official exchange rae; Res inernaional reserves; σe, σ Res sandard deviaions of he exchange rae and inernaional reserves respecively. A currency crisis is considered o have occurred when EMPI crosses specified hreshold value of is arihmeic mean plus hree imes sandard deviaion of he index. The crisis siuaion is defined by ransforming index behavior o he binary sysem: To consrucion of EMPI, monhly daa of gross inernaional reserves of he Naional Bank of he Republic of Kazakhsan 12 and official exchange raes of enge o US dollar as for he end of he period from January 1997 o Ocober are used. 10 Mehodology for Early Warning Sysem of financial crises for Kazakhsan was developed by he Financial Sabiliy Division of he Naional Bank of he Republic of Kazakhsan using many sudies by research insiues, invesmen and cenral banks, IMF and oher inernaional financial insiuions. 11 The researches being he basis for his approach were carried ou by such auhors as G. Kaminsky, K. Reinhar, S. Lisondo, M. Goldsein and H. Edison. 12 Gross inernaional reserves in US dollars a he end of period, excep he gold. 83

6 As a rule, subsanial flucuaions of EMPI dynamics during a precrisis period demonsrae cerain probabiliy of speculaive aacks increase on he exchange rae, bu only exceeding he hreshold value can be evidence of currency crisis occurrence. When hreshold values are imposed on a index curve i can be seen ha EMPI (1) and EMPI (2) has exceeded he upper limi once in April 1999 (Figure 1, Figure 2). Exchange Marke Pressure Index (monh-o-monh change) Figure 1 Figure 2 Exchange Marke Pressure Index (year-o-year change) % of changes EMPI Threshold (3хsd) % of changes EMPI Threshold (3хsd) Sage 2. Leading Indicaors 2.1. Selecion and Calculaion of Early Warning Indicaors Based on he analysis of empirical researches for he early warning sysem of Kazakhsan 17 variables were seleced as he mos high qualiy indicaors ha demonsrae differen aspecs of rising economic risks. Quarerly daa from 2000 when negaive consequences of 1999 calmed down, o 3rd quarer were used for he early warning indicaors in he sysem. Values of he indicaors were calculaed as a percenage change o a relevan period of a previous year, wha allows eliminaing he seasonal effecs. Two formas of he daa for he early warning indicaors are provided: sandard presenaion (growh rae of he indicaor, raio of absolue values) and nonsandard presenaion (deviaion from rend of real effecive exchange rae 15 ; indicaors calculaed on basis of ineres raes; residue argumen of excess money supply model in real erms 16 ) Signal Exracion Mehod According o he signal approach, each indicaor shall be analyzed separaely wihin a single-variable mehod o deermine a signal. Behavior of each indicaor is observed o deermine when a variable deviaes from is normal behavior above a cerain hreshold. The indicaor signals when is value exceeds a criical level. where, S,j - a signal of j indicaor during period, Х,j - value of j indicaor during period. 13 To calculae EMPI(1), exreme daa (hree poins for April, May and June 1999) were excluded from he series of exchange rae and inernaional reserves; for EMPI(2)- exreme values from April 1999 o May For some indicaors of Q3 2007, esimaed daa are provided. 15 Hodrick-Presco filer was used o calculae he real effecive exchange rae rend. 16 Indicaor is defined as deviaion of assessed demand for money from money supply observed (expressed as a share of money supply in GDP), i.e. as residuals of he below regression equaion: M GDP = a 0 + a 1 Y + a 2 p + a 3 where M money supply, GDP nominal GDP, Y GDP (in real erms), р CPI, ime. + ε 84

7 I should be aken ino accoun ha for some indicaors, any value above he hreshold means increase of poenial crisis, while for oher indicaors vice versa (Table 1). Table 1 Indicaor s forma and direcion relevan hreshold crossing Iem Indicaor signals relaive o is Indicaors Forma No. hreshold value, if i is 1 Gross inernaional reserves of he NBRK growh rae lower 2 Curren accoun balance/gdp absolue value lower 3 Differenial beween domesic and foreign 17 nominal absolue value ineres raes higher 4 Real effecive exchange rae rend deviaion higher 5 Expor of goods and services growh rae lower 6 Impor of goods and services growh rae higher 7 Terms of rade growh rae lower 8 Shor-erm foreign deb share in GDP absolue value higher 9 Credis o economy /GDP growh rae higher 10 Money muliplier growh rae higher 11 Real ineres rae on deposis absolue value higher 12 Excess money supply in real erms esimaion higher 13 Money supply / inernaional reserves growh rae higher Differenial beween ineres raes for credis and absolue value 14 deposis 18 higher 15 STB deposis growh rae lower 16 Real GDP growh rae lower 17 Fiscal budge balance/gdp absolue value lower Threshold values for each indicaor were calculaed on basis of a percenile scale. A percenile of empirical disribuion of he indicaor which minimized raio of false o good signals 19, was defined as an opimal hreshold level. The acual hreshold value of he indicaor varies from counry o counry, however, perceniles are permanen. The false o good signals raio (noise-osignal raio, NRS) has primary imporance in he early warning sysem: he lower NRS, he more effecive forecas he indicaor provides, and herefore any indicaors wih NRS more han 1 are reaed as unessenial. In he empirical researches NRS values were calculaed on basis of financial crises daa of many counries and, hus, for calculaion NRS i is required o have a relevan hisory of financial crises. Neverheless, in he absence of sufficien hisory of financial crises in Kazakhsan, daa of he various empirical researches were used ha had given an opporuniy o exrapolae hem for he sysem of early warning indicaors of Kazakhsan. As he mos opimal varian of hreshold value was used a percenile of disribuion which was equal o he arihmeic mean of values available from he various researches (Table 2). Implemenaion of such approach can be proved by fac ha in considered sudies was researched daa of many developed and developing counries which experienced boh currency and banking crises. Table 2 Threshold values of indicaors for currency and banking crises (in perceniles of disribuion) Indicaors Average hreshold value for currency crisis Average hreshold value for banking crisis Indicaors Average hreshold value for currency crisis Average hreshold value for banking crisis 17 I demonsraes counry risk assessmen. 18 I demonsraes credi expansion and growh of ineres raes for credis wih regard o deposis for compensaion possible losses as a resul of porfolio deerioraion. 19 A signal is considered as false, if an indicaor had given he signal bu a crisis didn occur, or an indicaor had given no signal bu a crisis occurred (wihin 24 monhs) and vice versa. 85

8 Gross inernaional reserves of he NBRK Money muliplier Curren accoun balance/gdp Real ineres rae on deposis in enge Differenial beween domesic and Excess money supply in real foreign nominal ineres raes erms Real effecive exchange rae Money supply / inernaional reserves Expor of goods and services Differenial beween ineres raes for credis and deposis Impor of goods and services STB deposis Terms of rade Real GDP Shor-erm foreign deb in GDP Fiscal budge balance/gdp Credis o economy /GDP Criical levels specific for our indicaors were calculaed based on he obained hreshold values expressed in percenile of disribuion and crossing heir value were recorded wih he binary sysem using formula (3). Sage 3. Complex Crisis Vulnerabiliy Assessmen of Financial Sysem of Kazakhsan 3.1. Composie Vulnerabiliy Index I is difficul o judge abou a probabiliy of a financial crisis based on signals given by each individual indicaor, as hey do no provide overall complex assessmen of financial insabiliy. Therefore, a direc mehod o cover all vulnerabiliy facors is o combine of signals ino a single main channel, i.e. consrucion of a composie vulnerabiliy facor. Composie vulnerabiliy facor gives a possibiliy o combine informaion obained from each indicaor as he more indicaors signal of a forhcoming crisis, he more probabiliy of is acual occurrence. In he course of his invesigaion, a weighed composie index was consruced being a basis for deerminaion of condiional probabiliy of a crisis in fuure. The advanage of his composie index is ha i akes ino accoun predicive power of each individual indicaor and combines informaion of heir signals by heir weighing wih inverse value of noise-o-signal raio. I is essenial o remember ha indicaors wih he leas value of he noise-o-signal raio pracically proved o be high qualiy early warning indicaors. Therefore, when hey are weighed in consrucion of he composie index, hey are given he highes weighs. The composie index is calculaed according o he following formula: I = where, S - a signal of j variable during period, w noise-o-signal raio for j variable. For consrucion of he composie index i is required values of noise-o-signal raios of each of indicaors. As i has been already noed, i is impossible o calculae hese raio values in he Kazakhsan siuaion, herefore a echnique similar o definiion of hreshold levels for indicaors was applied (Table 3). Table 3 Weighs for signals of early warning indicaors Indicaors Weighs for signals (currency crisis) n j= 1 S w j j (4) Indicaors Weighs for signals (banking crisis) Real effecive exchange rae 3,58 Real effecive exchange rae 2,74 Fiscal budge balance / GDP 2,50 Fiscal budge balance / GDP 2,48 Curren accoun balance/gdp 2,44 Real GDP 2,22 Expor of goods and services 2,30 Curren accoun balance/gdp 2,00 Money supply/inernaional reserves 2,12 Money muliplier 1,90 Gross inernaional reserves of he NBRK 1,89 Real ineres rae on deposis 1,90 Excess money supply in real erms 1,76 Expor of goods and services 1,88 86

9 Real GDP 1,72 Differenial beween domesic and foreign nominal ineres raes Credis o economy/gdp 1,56 Gross inernaional reserves of he NBRK 1,56 Real ineres rae on deposis 1,48 Money supply/inernaional reserves 1,54 Terms of rade 1,37 Credis o economy/gdp 1,44 Money muliplier 1,30 STB deposis 1,39 STB deposis 1,17 Excess money supply in real erms 1,20 Shor-erm foreign deb in GFD 1,11 Terms of rade 1,15 Differenial beween domesic and foreign nominal ineres raes Impor of goods and services 0,89 Differenial beween ineres raes for credis and deposis 1,71 1,01 Shor-erm foreign deb in GFD 1,11 Differenial beween ineres raes for credis and deposis 0,87 0,69 Impor of goods and services 0,52 The mos predicive power in forecasing of boh he currency and banking crises has he real effecive exchange rae indicaor; he leas predicive power are: for he currency crisis - he differenial beween ineres raes for credis and deposis, and for he banking crisis impor of goods and services. Dynamics of he weighed composie vulnerabiliy index for Kazakhsan calculaed on basis of signals of seveneen early warning indicaors by crossing of he specified hreshold values are shown in Figure 3. Figure 3 Composie index for currency and banking crises Q Q Q Q Q Q Q Q Q. 4 Q. 3 Q Composie index for currency crisis Composie index for banking crisis 3.2. Condiional Probabiliy Assessmen of Financial Crises Based on he resuls of he composie index, i is possible o assess probabiliy of financial crises in Kazakhsan. To assess when economy of he counry is more or less vulnerable o any financial crises, he composie index values should be conneced wih condiional probabiliy of a crisis. G. Kaminsky, M. Goldsein and K. Reinhar calculaed condiional probabiliy of a currency and banking crises for heir huge sample. Analogous o he argumenaion above, i is impossible o deermine he condiional probabiliy for Kazakhsan; however, in he course of his invesigaion he probabiliies were esimaed by linear ransformaion of he inervals analyzed by G. Kaminsky. For linear ransformaion i is necessary o calculae he heoreical maximum of he composie index for a cerain period of ime which is a sum of all he weighs ha can be calculaed, if all he indicaors will give signals a he cerain momen of ime. The heoreical maximum of he weighed composie index for he currency crisis is and for he composie index of he banking crisis To compare all he pracical values of he composie index wih is maximum, Figure 4 demonsraes rescaled graph of he composie indices o assess a real siuaion, if, in heory, all he indicaors will give a signal. Figure 4 87

10 currency crisis 28,0 24,0 20,0 16,0 12,0 8,0 4,0 Composie index for currency and banking crises (new scale) 27,6 22,6 17,6 12,6 7,6 2,6 banking crisis 0,0 1 Q Q Q Q Q Q Q Q Q. 4 Q. 3 Q ,4 Composie index for currency crisis Composie index for banking crisis The nex sep is o ransform he composie index values via linear ransformaion o condiional probabiliies of he currency and banking crises. According o he linear ransformaion resuls he probabiliy of a currency crisis reaches almos 100% when he composie akes on a value of 13 and above. However, an analogous composie index value of he banking crisis equal o 12.5 and above defines he maximal probabiliy of banking crisis occurrence only a a level abou 40 %. I can be explained by he fac ha according o G. Kaminsky research hese indicaors are more effecive for esimaion of currency crisis probabiliy han for he banking crisis. A final sep of he complex assessmen of he financial sysem vulnerabiliy is esimaion of probabiliy for he currency and banking crises in Kazakhsan based on composie index values for a period from 2000 o 3rd quarer For calculaion of condiional probabiliy for Kazakhsan, he composie index values were compared o corresponding inerval of values of condiional probabiliy (Figure 5). Figure 5 Condiional probabiliy of currency and banking crises in Kazakhsan currency crisis 1,00 0,75 0,50 0,25 0,00 1 Q Q Q Q Q Q Q Q Q. 4 Q. 3 Q ,40 0,35 0,30 0,25 0,20 0,15 0,10 0,05 0,00 banking crisis Probabiliy of currency crisis Probabiliy of banking crisis Probabiliy dynamics of he currency and banking crises has he similar endencies of increase and decrease. High probabiliy of he financial crisis during and some periods of is observed. Thus, a forecas period is considered as advanced monhs. I is possible o see, ha dynamics of condiional probabiliy of financial crises flucuaes, showing high probabiliy of he financial crisis during one period and a very low level during anoher one. In his case i should be aken ino accoun boh he naional economic policy providing for cerain correcive measures and exising general condiions a he domesic and foreign markes. For example, high urbulence of he financial marke siuaion in 3rd quarer 2007, decrease of inernaional reserves, deerioraion of parameers of balance of paymen and subsequen reducion of he money supply were refleced in general dynamics of condiional probabiliy. Thus, he early warning sysem of he financial crisis developed for Kazakhsan represens adaped o he feaures of he counry he signal approach for early warning of financial crises. The 88

11 fac of ha early warning sysem is based on inegraed experience of empirical researches on se of many financial crises ha occurred during he differen periods, does no reduce is value, and on he conrary, gives confidence in is effecive applicaion for financial sabiliy analysis along wih oher analyical ools. However, if in forecasing he currency crises he indicaors demonsrae raher cerain probabiliy, as for he banking crises he analysis should be expanded wih addiional mehods of vulnerabiliy assessmen of he financial secor due o average probabiliy of he banking crisis a he maximal value of composie index. 89

12 10. Risk Disribuion in Corporae Secor Financial condiion of non-financial organizaions has a direc impac on sabiliy of he whole economy. The corporae secor, being a principal debor of banks, plays one of he key roles in developmen of all financial sysem. The purpose of his analysis is o deermine risks of he corporae secor hrough analysis of annual resuls of for large and medium-sized businesses amouning o 75% of oal asses of he corporae secor. The research was conduced on he base of disribuion of financial sabiliy indices of enerprises in erms of profiabiliy, deb burden and liquidiy; hrough analysis of financial indicaors consancy; and idenificaion of businesses exposed o defaul risks 20. Financial Saus Indicaors of Corporae Secor Profiabiliy indicaors (reurn on equiy and sales) ROE is a profiabiliy indicaor, which is criical o business sabiliy. I shows an abiliy of an enerprise o increase capial, resis adverse condiions and o pay off he debs. ROE= Pre-ax reurn /Average equiy capial On he whole, as for large and medium sized businesses, an increase of reurn on equiy was observed in. The reurn on equiy amouned o 49.3% in comparison wih 40.8% in This endency is confirmed by percenile disribuion of he index where a median value demonsraes some increase in comparison wih he previous year. However, classificaion of businesses by percenile groups is raher erraic. The 50 h percenile group includes mos of he businesses where average ROE was 6% in, in oher words 25.3% of he oal number. 27% of asses are allocaed beween businesses included in 75 h percenile, where average ROE is 35% (Figure 1). Moreover, having amouned o 25% of oal businesses in, he businesses included in 10 h and 25 h percenile groups sill have negaive profiabiliy. This business group is of he mos ineres as enerprises of his group are subjec o high defaul probabiliy. The mos vulnerable group covers abou 8% of large and medium-sized business asses and 16% indebedness o banks and number of employees. Disribuion characerisics of reurn on sales are similar o reurn on capial dynamics. Reurn on sales index demonsraes capaciy of any enerprises o produce goods or provide services a low coss or high price and i is calculaed using he following formula: Reurn on sales = Pre-ax reurn /Gross income of main aciviy Under general growh of reurns on sales in, 24.7% of oal enerprises were included in a negaive profiabiliy zone (up o 25 h percenile); a share of he enerprises changed a lile in comparison wih 2005 (Figure 2). Indebedness of hese enerprises o banks is abou 16% of oal indebedness for he whole secor. 20 The Mehods are based on Guidelines of Financial Sabiliy Indicaors (FSI) issued by he Inernaional Moneary Fund and oher sudies of cenral banks, he Bank of Ireland, in paricular. Analysis and calculaions are made by he Financial Sabiliy Division of he Naional Bank on basis of primary daa provided by Saisics Agency of RK. 90

13 Figure 1 Figure 2 140% 120% 100% 80% 60% 40% ROE disribuion on percenile % % 55.29% 57.86% 30% 20% 10% 0% -10% Reurn on sales disribuion on perceniles -7.44% 20.45% 8.29% 1.14% 24.98% 9.37% 1.34% -6.62% 20% 0% -20% -40% 14.28% 17.07% -1.37% -0.69% % % h percenile 25h percenile Median 50h percenile 75h percenile 90h percenile -20% -30% -40% % h percenile Median 50h percenile 90h percenile 25h percenile 75h percenile % Business raing as o percenile groups may change from year o year. A reason for such changes is ha businesses, being he mos profiable during a cerain period, may pass o a group of he leas profiable businesses in a relaively shor period. In his case defaul probabiliy for he enerprise increases a lo. To clarify his fac, a ransiion marix should be drawn. This marix demonsraes quarile 21 classificaion of businesses subjec o heir profiabiliy levels, and consancy of disribuion by periods. A purpose of he ransiion marix is o deermine shares of enerprises for each quarile in comparison wih a previous period (he more quarile value is, he higher degree of consancy is applied). Inclusion of business in low-profi quariles can be explained wih one ou of wo reasons: emporary or permanen low profiabiliy. In case of he emporary low profiabiliy, a consancy value is low and defaul probabiliy for an enerprise is low as well, since a profiabiliy index can increase in he following period. In he second case a consancy value is considered o be higher and defaul probabiliy increases a lo. On he whole, a he end of abou 49% of enerprises had negaive reurn on equiy (Table 1). A he same ime, reurn volailiy is raher high as abou 50% of enerprises pass from he 1 s quarile o oher quarile groups. Table 1 Transiion marix of reurn on equiy ROE (%) 2005 Quarile 1 Quarile 2 Quarile 3 Quarile 4 Quarile 1 49% 18% 12% 20% Quarile 2 21% 54% 19% 6% Quarile 3 15% 21% 45% 20% Quarile 4 15% 7% 24% 54% Moreover, 20% enerprises of a risk group, mainly, in agriculure and consrucion secors, could jump o 4 h quarile and 23% of high-profi enerprises moved from he 4 h quarile o he 1 s and 2 nd. Thus, reurn on equiy may be subjec o considerable flucuaions during shor periods. Reurn on sales demonsraes higher consancy level (Table 2). The dominan share of enerprises which had been in a risk zone in 2005, kep negaive profi in. Mos of enerprises which moved o a group of enerprises wih maximum profiabiliy were agriculural businesses. Enerprises ha moved from he 4 h quarile in 2005 o he 1 s quarile in included 13% of 21 Quarile is deermined by 25 perceniles inerval. 91

14 consrucion companies. On he whole, high-profi enerprises probably will keep high profiabiliy in he nex period in comparison wih a risk group ha subjec o exernal facors deermining financial and economic aciviy of enerprises. Table 2 Transiion marix of reurn on sales Profiabiliy (%) 2005 Quarile 1 Quarile 2 Quarile 3 Quarile 4 Quarile 1 58% 18% 11% 14% Quarile 2 21% 50% 19% 10% Quarile 3 12% 23% 49% 16% Quarile 4 10% 9% 22% 60% Deb burden indicaors (leverage and ineres rae coverage raio) The financial leverage raio shows he exen he enerprises rely on deb financing raher han equiy capial. This raio is an insrumen deermining defaul probabiliy of an enerprise under is credi conracs. The bigger an enerprise deb is, he more probabiliy of defaul is o occur since oo big debs resul in insolvency and financial difficulies. The leverage is calculaed using he following formula: Leverage = Toal liabiliies / Equiy capial Leverage dynamics demonsraes ha enerprises wih high deb burden increased heir obligaions even more, having increased heir deb burden o a greaer exen hrough bank loans (Figure 3). Figure Disribuion of leverage on perceniles h percenile Median 50h percenile 90h percenile 25h percenile 75h percenile A share of obligaions under he highes percenile group has increased by 31% o 60% of whole obligaions of enerprises; and indebedness of his percenile group o banks increased from 22% o 55% of oal indebedness o banks. A number of enerprises wih leverage values equal or more han he 90 h percenile, amouns o 25% of Kazakh enerprises. The leverage ransiion marix shows ha 67% of enerprises are in he 4 h quarile having sayed a risk zone from year o year (Table 3). Abou 16% of enerprises moved from he risk zone o a non-risk zone o he 1 s quarile. Abou 35% of such enerprises included consrucion and rading companies. Majoriy of enerprises ha moved from he non-risk zone o a higher-risk caegory were agriculural businesses having amouned o 21% of he 4 h quarile enerprises. 92

15 Leverage (%) Leverage ransiion marix 2005 Quarile 1 Quarile 2 Quarile 3 Quarile 4 Quarile 1 76% 9% 3% 11% Quarile 2 1% 77% 19% 3% Quarile 3 3% 16% 62% 19% Quarile 4 16% 2% 15% 67% Table 3 The ineres coverage raio supplemens he leverage analysis wih assessmen of deb service capaciy. If reurn covers ineres paymens, an enerprise is considered sable, even if i s subjec o high deb burden. This raio is calculaed he following way: Ineres rae coverage raio = EBIT 22 /Ineres paid Figure Disribuion of ineres paymen curren raio on perceniles On he whole, as for enerprises, he level of ineres coverage increased slighly. A he same ime, as for enerprises included in he high-risk zone, he raio decreased much (Figure 4). 25% of enerprises lay in group of 10 h and 25 h perceniles wih negaive ineres coverage. The aggregae liabiliies of hese enerprises o banks were abou 17% in. 10h percenile Median 50h percenile 90h percenile 25h percenile 75h percenile Ineres Coverage (%) Transiion marix of ineres coverage raio 2005 Quarile 1 Quarile 2 Quarile 3 Quarile 4 Quarile 1 52% 19% 13% 16% Quarile 2 24% 49% 22% 5% Quarile 3 14% 25% 42% 18% Quarile 4 10% 6% 22% 61% Table 4 According o he ransiion marix of he ineres rae coverage raio (Table 4), 52% of enerprises ha lay in he 1 s quarile of 2005 couldn have improved heir financial saus o he end of, whereas 16% became able o cover heir ineres paymens compleely. Mos of such enerprises were agriculural enerprises (23%). In addiion, a posiive facor is ha 61% of enerprises included in he 4 h quarile during 2005 remained a he same level and only 10% moved o he high-risk zone. 22% of such high-risk enerprises are consrucion businesses. 22 EBIT - Earning before ineress and axes. 93

16 Liquidiy indicaors (curren liquidiy) The curren liquidiy raio demonsraes abiliy of an enerprise o fulfill heir regular financial liabiliies. The more considerable cash flows an enerprise has, he higher possibiliy i will avoid defaul on financial obligaions. The curren liquidiy is calculaed he following way: Curren liquidiy raio = Curren asses/curren liabiliies Figure 5 Disribuion of curren raio on perceniles Disribuion of he curren liquidiy raio demonsraes ha overall growh of liquidiy in 7 in he large and medium-sized business secor o 1.10 was ensured by high-liquidiy enerprises, while oher enerprises had insignifican increase of liquidiy (Figure 5). The ransiion marix for he curren liquidiy raio shows he high consancy level (Table 5). 3 In paricular, only 6% of enerprises being included in he 1 s quarile in 2005 could improve heir aciviies, having achieved he highes resuls in. Mosly, hey were 0 agriculural, ranspor and communicaions h percenile 25h percenile enerprises and enerprises providing services o Median 50h percenile 90h percenile 75h percenile consumers. Enerprises wih decreased liquidiy during he las period should be noed. A share of such enerprises amouned o 4%, one hird of he enerprises are manufacuring enerprises. Liquidiy (%) Transiion marix of curren liquidiy raio 2005 Quarile 1 Quarile 2 Quarile 3 Quarile 4 Quarile 1 67% 19% 8% 6% Quarile 2 21% 54% 18% 7% Quarile 3 8% 20% 52% 20% Quarile 4 4% 7% 22% 67% Table 5 Financial Sabiliy Risks of Corporae Marke The nex sage is assessmen of vulnerabiliy of he banking secor o risks associaed wih borrowers financial saus hrough assessmen of bank obligaions of he enerprises included in he risk zone as o reurn on capial, liquidiy and leverage. A Venn diagram is a graphic represenaion of defaul risks of enerprises. I consolidaes enerprises wih he wors indicaors and groups hem, hus revealing enerprises of he highes risk. As he above analysis demonsraes ha enerprises wih he wors indicaors are included in a group below 25 percenile (low liquidiy and negaive reurn on capial) and above 75 percenile (more han double leverage), relevan hreshold values were used o asses risk groups (Figure 6). 94

17 Disribuion of enerprises in Figure 6 Liquidiy Liquidiy ROE ROE 13.4% 5% 12.7% (9%) (23%) 2% (24%) 6,3% 3,9% (7%) 2,3% (4%) 5% (9%) 2,4% 0,8% 5,9% 7% 14% (25%) 26,5% Leverage Noe: of he RK enerprises; ( ) of he group of enerprises included in he general risk zone Leverage Noe: Liabiliies o banks in oal liabiliies of enerprises o banks in he RK All in all, in a share of enerprises included, a leas, in one of he risk groups amouned o 56% of large and medium-sized businesses in Kazakhsan. Mos of hese enerprises belong o agriculure, manufacuring indusry, consrucion and real esae ransacions and rading secors. Their indebedness o banks is slighly more han 50% of all indebedness. Based on he share of he above enerprises he highes specific weigh of risks accouns for he rading secor: 64% of enerprises of he secor have 74% credis. A he same ime he defaul probabiliy increases, if, a leas, wo risk facors are available: - high indebedness and low liquidiy; - low reurn on capial and low liquidiy; - low reurn on high-leverage capial, or - combinaion of he above hree risk facors. Mos of enerprises are included only in one risk group. Enerprises included ino, a leas, wo of hree risk groups amouned o 16% of enerprises in he Republic of Kazakhsan, and he mos criical zone includes 2.3% enerprises. Thus, i s possible o assess high-risk indebedness, i.e. he share of large and medium-sized business obligaions o banks exposed o high-risk defaul. The indebedness is 11% of oal indebedness of enerprises in he Republic of Kazakhsan. The major groups of enerprises exposed o defaul risk belong o he agriculural secor (22% of secor s obligaions o banks), manufacuring indusry (20%), consrucion (19%), hoel business (19%), real esae ransacions (13%) and rading (7%). In he narrow inerpreaion a high-risk porion of deb is 0.8%, i.e. in case of combinaion of all he risk facors when he defaul risk is he highes. As agains 2005, he highrisk indebedness increased insignificanly, while specific weigh of he enerprises having he above indebedness decreased. 95

18 Summary All in all, profiabiliy and liquidiy indicaors slighly improved along wih he growh of he corporae secor indebedness. A consancy level of major financial indicaors is high, i.e. enerprises wih sable financial indicaors improved heir financial saus; and financial saus of high-risk enerprises did no change or slighly decreased. Defaul probabiliy for bank obligaions is exposed o a moderae risk. The highes risk enerprises are in he agriculural, consrucion and manufacuring indusries. The disribuion of financial indicaors changes in erraic way from year o year, i.e. enerprises of he same secor demonsrae boh considerable increase and decrease of heir financial indicaors. Thus, here were no sysemic risk associaed wih high defaul probabiliy of cerain secors, on he whole, and subsequen considerable decrease of he credi risks due o concenraion of banks asses in agriculure, consrucion and rading, bu poenial problems can occur a individual enerprises. Profiabiliy and liquidiy of agriculural increased significanly, however, aking ino accoun considerable growh of indebedness o banks, his secor is subjec o he highes defaul risk, if environmenal facors change. 96

19 11. Financial Sabiliy Index Consrucion The esimaed composie index of financial sabiliy shows a degree of ension or pressure of differen facors on sabiliy of he banking sysem. This year, posiive dynamics of he index was followed by decrease of is value a he beginning of Ocober Major facors of index downfall are asses qualiy and a liquidiy level. One of he main endencies in recen sudies is developmen of banking secor sabiliy indicaors based on a single composie index helping o monior he banking secor 23. The basis of he above indicaors may be dynamics of he parameers indicaing financial saemen of he banking sysem. These indicaors can be added wih differen marke indicaors bringing leading or predicing specific o he index (for example, based on sock exchange indicaors). In his analysis an aemp was made o assess vulnerabiliy of he banking secor o differen facors of pressure on is sabiliy using he financial sabiliy index on basis of dynamics of differen financial soundness indicaors of he banking sysem 24. In paricular, for index consrucion purpose, variables characerizing impac of asses qualiy, liquidiy level, capial adequacy, profiabiliy and exposure o exchange risks on sabiliy of he banking secor, were seleced. Thus, he composie financial sabiliy index includes he following groups of indicaors: 1. Asses qualiy: - share of nonperforming loans 25 in oal loan porfolio of he banking secor; - raio of nonperforming loans (ne of creaed provisions agains hem) o balance shee capial of banks. 2. Liquidiy: - highly liquid asses in oal asses of banks; - raio of highly liquid asses o shor-erm liabiliies up o 1 year (including demand liabiliies); 3. Capial adequacy: - Tier 1 capial o risk weighed asses. 4. Profiabiliy: - ROA (ne income before ax o oal asses of banks); - ROE (ne income before ax o oal balance shee capial of banks). 5. Exchange rae risk: - Ne open posiion in foreign exchange o regulaory capial of banks. A he iniial sage each variable was normalized by a variance-equal mehod 26. To deermine an index componen value, he average value of normalized values for a group of indicaors included in he componen was calculaed. Afer ha values of all he componens were aggregaed in he composie index. Moreover, o represen adequaely conribuion of each indicaor o he index esimaed, he values of normalized indicaors were convered o posiive or negaive values according o heir impac 27. Thus, he posiive index value reflecs improvemen of he facors influencing sabiliy of he banking sysem and negaive value shows heir deerioraion. 23 An Index of Financial Sress for Canada by Mark Illing and Ying Liu. Bank of Canada Working Paper, 2003; Measuring and forecasing sress in he banking secor: evidence from Swizerland by Elke Hanschel and Pierre Monnin. BIS, Working Paper, The sudies were carried ou by he financial sabiliy division of he Naional Bank of he Republic of Kazakhsan. Experience of he Czech Naional Bank, he Bank of Finland and he Naional Bank of Swizerland was sudied in selecion of he variables as he index componens, index consrucion approach and is inerpreaion. 25 According o adoped loan classificaion rules as nonperforming loans, doubful loans of 2, 4 and 5 caegories and loss loans were included ino calculaion. 26 The daa are normalized using a variance-equal mehod by division deviaion of each value from he mean by sandard deviaion of sample. 27 For insance, if a normalized indicaor (E.g. nonperforming loans) has negaive value, his reflecs relaive decrease of he indicaor and ha has a posiive impac on banks soundness. 97

20 Dynamics of he index reveal he sabiliy facors improvemen from he beginning of wih furher decrease of he index value on Ocober 1, 2007 (Figure 1). So, asses qualiy and liquidiy componens of he index made he major conribuion under idenificaion of negaive facors affecing he banking secor soundness. Moreover, in hisorical rerospecive review, in Q he asses qualiy componen has he bigges negaive value 28, hus increasing credi risk exposure. Negaive impac of he liquidiy componen decreased compare wih previous reporing dae. Influence of profiabiliy, capial adequacy and exchange risk facors compensaing parly more sric requiremens o coverage of credi and liquidiy risks had a posiive impac. Some limiaions in index consrucion should be aken ino accoun. In paricular, one of he exising problems of index calculaion is idenificaion of saisically significan weighs of each componen revealing is relaive conribuion (equal weighs of he componens were used for consrucion of his index). I can be achieved by increase in he number of indicaors and facor analysis implemenaion as well as oher mehods for weigh idenificaion. Moreover, using marke-based indicaors such as prices of banks shares and oher indicaors as componens will conribue forward-looking informaion o he index. The above problems are subjec o furher sudies o improve puriy and qualiy of he index. Table 1 Financial sabiliy index and conribuion of is componens* (2005 Ocober 2007, quarerly, a he beginning of a period) Asse qualiy Liquidiy Capial adequacy Profiabiliy Exchange rae risk Index * calculaed on basis of FSA daa Source: NBRK 28 I also affeced by he ighening of classificaion rules emphasized more on borrower s financial saemen han on collaeral value and, as a resul, provides more adequae credi risk assessmen. 98