Post on 18-Dec-2015
SME lending in aretail bank
Roberto GiannantoniExperian Scorex
Origination scoringOrigination scoring
Behavioural scoringBehavioural scoring
Customer scoringCustomer scoring
Origination scoringOrigination scoring
Personal Personal customerscustomers
Small businessSmall business customerscustomers
Customer scoringCustomer scoring
Background
Evolution of risk modelling within a Retail Bank
Background
Why small business lending is complex
• Great variation in the trading entities
• Infrequent (and late) production of formal financial details
• Risk assessment is only half the problem
Personal customersPersonal customers CommercialCommercialSmall Small businessesbusinesses
RulesRules
PrescriptivePrescriptivetreatmenttreatment
ExpertExperttoolstools
Hard dataHard dataHard weightsHard weights
Soft + hard dataSoft + hard dataSoft weightsSoft weights
Prescriptive decisions versus experts
PortfolioPortfolio
Complex relationshipComplex relationship(Group limit exists)(Group limit exists)
OUT OF SCOPEOUT OF SCOPE
ElseElse
Turnover > £1m paTurnover > £1m paOR OR
Borrowings > £100kBorrowings > £100kOUT OF SCOPEOUT OF SCOPE
ElseElse
Existing customerExisting customerNew customerNew customer
Start upStart upEXPERTEXPERT
SwitcherSwitcher WeakWeakrelationshiprelationship
StrongStrongrelationshiprelationship
EstablishedEstablishedUnestablishedUnestablishedEXPERTEXPERT
Primary segments
Portfolio segmentation
0%0%
5%5%
10%10%
15%15%
20%20%
25%25%
0-0-£25k£25k
£26k-£26k-£50k£50k
£51k-£51k-£100k£100k
£101k-£101k-£200k£200k
£201k-£201k-£500k£500k
£501k-£501k-£1m£1m
ApplicationApplicationvolumesvolumes
Annual turnoverAnnual turnover
Profile of small businesses
(Overdrafts + short term loans on equal footing. Includes existing borrowings)(Overdrafts + short term loans on equal footing. Includes existing borrowings)
0%0%
5%5%
10%10%
15%15%
20%20%
25%25%
30%30%
0-0-£5k£5k
£6k-£6k-£10k£10k
£11k-£11k-£25k£25k
£26k-£26k-£50k£50k
£51k-£51k-£100k£100k
Total borrowingsTotal borrowings
Profile of small businesses
ApplicationApplicationvolumesvolumes
Existing: strongExisting: strong(65%)(65%)
Existing: weakExisting: weak(20%)(20%)
Switcher:Switcher:establishedestablished(15%)(15%)
Profile of small businesses
Proportion ofProportion ofapplicationsapplications
(Contribution to model: (Contribution to model: + + weak weak ++++ medium medium ++++++ strong) strong)
Type of DataType of Data
Small business behav. dataSmall business behav. data
Key personnel bureau dataKey personnel bureau data
Key personnel behav. dataKey personnel behav. data
Commercial bureau dataCommercial bureau data
App. form details - financialsApp. form details - financials
Previous bank statementsPrevious bank statements
App. form details - otherApp. form details - other
Switcher:Switcher:EstablishedEstablished
Existing:Existing:WeakWeak
Existing:Existing:StrongStrong
++++ ++++++
++++++ ++++
++++ ++++ ++
++++ ++++ ++
++++
++++
++++
++
++
++
++++++
++++
Data sources for key segments
Gini coefficientGini coefficient
Score percentile rangeScore percentile rangeSwitcher:Switcher:
EstablishedEstablished(Good/bad odds)(Good/bad odds)
Existing:Existing:WeakWeak
(Good/bad odds)(Good/bad odds)
1 - 51 - 5 6 - 106 - 10 11 - 1511 - 15
....
....
....
....
....
....
....
....
.... 86 - 9086 - 90 91 - 9591 - 95 96 - 10096 - 100
Existing:Existing:StrongStrong
(Good/bad odds)(Good/bad odds)
0.6 : 10.6 : 1 1.0 : 11.0 : 1 1.6 : 11.6 : 1
....
....
....
....
....
....
....
....
.... 7 : 17 : 1 9 : 19 : 1
10 : 110 : 1
0.4 : 10.4 : 1 0.7 : 10.7 : 1 0.9 : 10.9 : 1
....
....
....
....
....
....
....
....
.... 15 : 115 : 1 20 : 120 : 1 40 : 140 : 1
0.7 : 10.7 : 1 1.1 : 11.1 : 1 2.0 : 12.0 : 1
....
....
....
....
....
....
....
....
.... 60 : 160 : 1 90 : 190 : 1
200 : 1200 : 1
75%75%65%65%50%50%
Scorecard predictiveness
• For strong relationship existing customers, the drivers for shadow exposure limits are:
turnover Regularity of trading Frequency of credits SIC code Risk
Customer scoringCustomer scoring
Exposure management
Distribution of “overdraft/annual turnover” (= ratio)
0%0%
5%5%
10%10%
15%15%
20%20%
25%25%
to 2
%to
2%
to 6
%to
6%
to 1
0%
to 1
0%
to 1
4%
to 1
4%
to 1
8%
to 1
8%
to 2
2%
to 2
2%
to 2
6%
to 2
6%
to 3
0%
to 3
0%
FrequencyFrequency
Ratio of overdraft to annual turnoverRatio of overdraft to annual turnover
Exposure management
00
2%
4%
6%
8%
10%
12%
14%
AverageAverageratioratio
to £
25k
to £
25k
to £
50k
to £
50k
to £
100k
to £
100k
to £
200k
to £
200k
to £
500k
to £
500k
to £
1m
to £
1m
Annual turnoverAnnual turnover
Exposure management
Impact of “regularity of trading”
00
2%
4%
6%
8%
10%
12%
14%
Very regular trading
Regular trading
Irregular trading
AverageAverageratioratio
to £
25k
to £
25k
to £
50k
to £
50k
to £
100k
to £
100k
to £
200k
to £
200k
to £
500k
to £
500k
to £
1m
to £
1m
Annual turnoverAnnual turnover
Exposure management
0%0%
2%2%
4%4%
6%6%
8%8%
10%10%
Very regular trading
Regular trading
AverageAverageratioratio
High
High
Med
ium
Med
ium
Low
Low
Frequency of creditsFrequency of credits
Impact of “frequency of credits”
Exposure management
Annual turnoverAnnual turnover= £51k - £100k= £51k - £100k
Impact of SIC code
SIC codeSIC code OverdraftOverdraftdemanddemand
Overdraft/Overdraft/turnover %turnover %
RegularityRegularityof trading of trading
FrequencyFrequencyof creditsof credits
Farming - cropsFarming - crops VHiVHi VHiVHi N/AN/AFarming - livestockFarming - livestock VHiVHi VHiVHi AvAv VLowVLowSell carsSell cars VHiVHi AvAv AvAv HiHiRepair carsRepair cars HiHi AvAv VHiVHi HiHiSell petrolSell petrol AvAv VLowVLow VHiVHi VHiVHiW/sale h/hold goodsW/sale h/hold goods AvAv AvAv HiHi AvAvRetail foodRetail food HiHi LowLow VHiVHi VHiVHi
Retail furniture + electricalRetail furniture + electrical HiHi LowLow HiHi VHiVHiRestaurantRestaurant AvAv AvAv VHiVHi HiHiBarBar HiHi VLowVLow VHiVHi
Taxi operationTaxi operation AvAv AvAv AvAv LowLowIT consultancyIT consultancy VLowVLow LowLow LowLow VLowVLow
VLowVLow
HiHi
Exposure management
Exposure management
• For strong relationship existing customers, the drivers for shadow exposure limits are:
Turnover regularity of trading Frequency of credits SIC code Risk
• Significantly prefer loans compared to overdrafts
• Security considerations
• Lower limits for new/weak relationship customers
Customer scoringCustomer scoring
Switcher: establishedSwitcher: established Existing: weakExisting: weak Existing: strongExisting: strong
Comprehensive checking
Robust track Robust track recordrecord
No checking No checking
Know your Know your customer !!customer !!
… … if KYC performed if KYC performed recentlyrecently
Fraud prevention processing
Prescriptive casesPrescriptive cases
Grey area referralsGrey area referrals
Other reason for referralOther reason for referral
TotalTotal
Switcher:Switcher:establishedestablished
Existing:Existing:weakweak
Existing:Existing:strongstrong
60%60% 80%80%
45%45%
25%25% 20%20% 20%20%
100%100% 100%100% 100%100%
20%20%
30%30%
Weighted prescriptive rate = 70%Weighted prescriptive rate = 70%
Time to make prescriptive decisionsTime to make prescriptive decisions 5 min5 min15 min15 min
Degree of prescriptiveness
Conclusions
• Many requests are for small amounts and from small turnover businesses
• Strong scorecards can be developed for the three key segments
• Security can often be waived but is an integral part of the process
• Both experts and underwriters are needed (- evolving rules)
• Prescriptive treatment in ~70% of cases (- especially existing customers with a strong relationship)
• Average time to process an application is decimated
SME lending in aretail bank
Roberto GiannantoniExperian Scorex