DI Fund Suffcency Evaluaton Methodologcal Recommendatons and DIA Russa Practce Andre G. Melnkov Deputy General Drector DIA Russa THE DEPOSIT INSURANCE CONFERENCE IN THE MENA REGION AMMAN-JORDAN, 18 20 NOVEMBER, 2009
Two basc methods of evaluaton of DI Fund suffcency (n practce) 1. On the bass of expert opnons on suffcency sze of DI Fund (wthout estmaton of PD of member banks and DI Fund cover losses) Ideas of some respected experts about «margn of safety» whch the DI Fund should have 2. On the bass of rsk analyss Estmaton of PD of member banks and DI Fund cover losses ٢
4 STEP Procedure of estmaton of DI Fund suffcency STEP-1 Assgnng the mpled level of DIS fnancal relablty n correspondence wth the soveregn credt ratng STEP-3 Estmaton of expected and unexpected losses of DI Fund wth a certan probablty STEP-4 Evaluaton of DIF suffcency STEP-2 Determnng the lst of too bg to fal banks and excludng them from the bass of evaluaton ٣
STEP - 1 Orentaton on the mpled level of DIS fnancal relablty A general ndcator of fnancal relablty s a credt ratng For Depost Insurer t should be a modelng or so-called mpled credt ratng Impled ratng can be assgned by mappng procedure, whch gves the correspondence between credt ratngs and values of PD ٤
Correlaton of credt ratng and hstorcal frequency of default on the example of DIA, Russa Standard & Poor s Ratng Hstorcal frequency of default, % duraton perod, 1 year duraton perod, 5 years A 0,06 0,60 A- 0,07 0,73 BBB+ 0,15 1,74 BBB 0,23 1,95 BBB- 0,31 3,74 BB+ 0,52 5,41 BB 0,81 8,38 BB- 1,44 12,32 B+ 2,53 17,65 B 6,27 23,84 B- 9,06 29,44 CCC C 25,59 44,50 ٥
STEP - 2 excludng too bg to fal banks from the estmaton bass When these banks meet dffcultes, the State undertakes a set of specal measures for ther support Excludng too bg to fal banks from evaluaton bass we decrease our depost nsurance labltes by 67% ٦
STEP - 3 Approaches to estmatons of expected (EL) and unexpected losses (UL) of DI Fund CL = EL + UL EL = EAD PD LGD - Expected Losses EAD nsured deposts n a member bank PD probablty of default of a member bank LGD share of non-recoverable resources from the bankruptcy estate of a lqudated bank Value of Unexpected Losses (UL) does not have a smple analytcal expresson. The easest way to estmate UL s to use statstcal smulaton method (Monte Carlo). ٧
Estmatons of EAD (nsured deposts n a member bank) EL = EAD PD LGD To assess the varable EAD - we analyze the dynamcs of growth of household deposts (.e. nsured deposts n a member bank) RUR bln.. 12 000,0 10 000,0 Dynamcs of growth of household deposts n 2008-2009 RUR deposts foregn currency deposts 8 000,0 6 000,0 4 000,0 2 000,0 Forecast 0,0 01.2008 07.2008 01.2009 07.2009 01.2010 07.2010 01.2011 07.2011 01.2012٨
Estmatons of LGD (share of non-recoverable resources from the bankruptcy estate of a lqudated bank ) EL = EAD PD LGD To assess the varable LGD we use collected statstcal data from all bankruptcy cases Snce 2004, the DIA, Russa has been fulfllng the functons of the bankruptcy trustee n 224 banks. In 137 lqudaton proceedngs have come to the end, n 87 cases are stll n progress. ٩
Approaches to estmatons of PD (probablty of default of a member bank) EL = EAD PD Three man approaches to estmaton of probablty of default (PD) of member banks LGD 1. On the bass of credt ratngs of member banks (Standard Approach) 2. On the bass of econometrcal models (Improved Approach) 3. On the bass of market-data models (Advanced Approach) ١٠
PD estmaton on the bass of econometrcal model The model of a bnary choce sutes best of all PD(Y=1)= f(β0+ β1*x1+ + βk*xk) where f(..) functon of logstc dstrbuton xk ndependent varables havng an nfluence on the event of bank default βk coeffcents ١١
## Sgnfcant varables (Xk) value 1 ROE (return on equty) -0,023 2 captal adequacy 3 nterest cost of labltes 4 yeld of promssory notes 5 revenue performance of loan portfolo excl. promssory notes 6 workng credt 7 lqudty cushon 8 provsons for bad debts 9 lqud assets 10 marketable securtes (resdents) -0,249-0,036-0,039-0,566-5,927-0,264 3,831-0,041-0,197 ١٢
PD estmaton on the bass of market-data model PD s estmated not on the bass of prevous hstory of defaults of smlar member banks but takng nto consderaton current state of each real member bank n current condtons of bankng sector and economy as a whole PD of largest banks whch are the most dangerous can be adequately estmated only by market models In practce, two man types of market-data models are the most developed: - Structural Model - PDs are estmated on the bass of current market prces of shares ssued by DIS members - Reduced Form Model - PDs are estmated on the bass of current market prces of bonds, ssued by DIS members ١٣
Densty of dstrbuton of DI Fund losses Densty of dstrbuton 0.35 0.3 0.25 Confdence nterval 99.7 % 99 % 95 % 0.2 0.15 0.1 0.05 0-10 0 10 20 30 40 50 60 Losses of the Fund ١٤
Theory & Practce at Losses Estmaton 100,0 90,0 86,0 89,0 80,0 RUR bln 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0 40,6 unexpected losses 5,2 0,3 61,6 unexpected losses 12,6 16,1 71,5 18,0 as of 1.11.09 5,2 34,0 2007 2008 2009 2010 unexpected losses unexpected losses 5% forecasted expected losses ncurred losses recalculated ncurred losses ١٥