streamlining individual lending evaluations at … standardized margins and household expenses for...
TRANSCRIPT
Swadhaar Finserve Pvt. Ltd.
Evaluations at SwadhaarAppendix 2 to
Streamlining Individual LendingEvaluations Project Report
Streamlining Individual Lending
2
• The Need for an Individual Microenterprise Loan Product
• Swadhaar’s Individual Lending Process
• Swadhaar – Unitus Project Objectives
• Analysis of Key Tools / Processes
o Tailoring Products to suit Business Needs
o Developing a Preliminary Client Scoring Model
o Standardization of key metrics in capacity evaluation
o Client’s business growth indicators
Page 2
Contents
3
The microenterprise segment is not being served effectively by current banking OR group lending products – thus, there exists a large gap for flexible individual business loan products
* For further information, please refer to APPENDIX 1 - Background Note on Individual Lending
Micro entrepreneurs need… Commercial Banks offer… Microfinance - Group Loans offer…
Flexible loan amounts between
Rs 25,000 – Rs 2,00,000
Loan amounts typically starting at > Rs 5,00,000
Rigid loans based on cycle, between Rs 5,000 – Rs 30,000
Shorter loan tenure for working capital requirement
Loan terms typically 12 months or greater
Loan terms typically 12-24 months
Option to foreclose loans based on business seasonality and cash flows
Available, generally for a penalty/fee
Renewal loans only when the whole group closes, and not as per individual requirement
Fewer documentation requirements Require several documents such as IT Returns, and often collateral
Low documentation requirement
Flexibility in payment mode – Doorstep cash collection/ PDC/ ECS
Require Post Dated Cheques or bank transfers
Need to participate in time-consuming group meetings
Page 3
The Need for an Individual Microenterprise Loan Product – Serving the Missing Middle*
4
Initiate Renewal
Loan
Evaluation
Loan
Approval
Loan
Disbursement
Loan
Repayment
Willingness to Pay
Capacity to Pay
Customer Relationship Officer (CRO)
CRO & Branch Manager
(BM)
Branch Mgr/ Area Mgr
Admin Assistant
CRO
20-30 mins 2 – 2.5 hours 15-20 mins 20-30 mins 5-10 mins per
collection
Loan evaluation is the most critical and difficult part of the lending process to
cover individual risk – it is important to explore tools to optimize this process
Promotion
Page 4
Swadhaar’s Individual Lending Process
5
Objectives of the Swadhaar-Unitus Project
• The primary objective of the Swadhaar -Unitus project was to analyze a selection of available data from Swadhaar’s existing client base (across 4 years of operations) and develop tools to simplify and standardize the individual lending evaluation process with an aim to increase efficiencies and enhance risk mitigation.
• The key tools/processes that were explored in detail as part of this project were:
1. Tailoring products to suit the needs of specific business sectors
2. Identifying a suitable methodology to implement a preliminary credit scoring model for better client selection
3. Using standardized margins and household expenses for client’s capacity evaluation
4. Using key business indicators to track client’s business growth across cycles
• Each of these processes is discussed in greater detail in the following sections
Page 5
1. Tailoring products to suit the needs of specific business sectors
2. Identifying a suitable methodology to design a preliminary credit scoring model for better client selection
3. Using standardized margins and household expenses for client’s capacity evaluation
4. Using key business indicators to track client’s growth across cycles
Analysis of Key Tools / Processes
Page 6
7
Benefits of tailored products
There are several benefits of tailoring products to the needs of specific business sectors:
• Effective Marketing
• Field officers can readily identify the target segment for the product and can deliver a more focused/relevant sales speech
• Reduced evaluation time
• As critical aspects of the business are already considered in the product features, evaluation processes can be shorter
• Increased client satisfaction
• Products are more customized to client’s business needs, thus leading to increased satisfaction levels
• Clients are more likely to avail repeat loans, thus improving customer retention
• Minimized Risk
• Specific business risks are addressed through product eligibility criteria and other loan features
Page 7
8
Findings: Business Categories by sector and activity
Based on similarities found, businesses were categorized across six major business types.
Sector Business Type Example of Businesses
Trader and Service
1. Semi mobile businesses
- Selling vegetables, fish, flowers, toys, plastic - Selling ice cream, candies, juices - Food stalls (chat, samosas, Chinese, vada pav, tea etc)
Trader 2. Small businesses but at fixed location:
- Selling: grocery items, footwear, masala, meats (chicken, mutton, eggs, etc), cutlery items, sari
Service 3. Small businesses but at fixed location: Services related
- Services: beauty parlor, barber shops, service stations, repair shops (watches, electronics etc.), tailor, photography
Service 4. Small businesses but at fixed location: Food related
- Fixed food outlets: restaurants, tea shops, sweets
Trader 5. Wholesalers/large retailers
- Selling: clothes (cut pieces, materials, readymade garments, lingerie, socks, etc.), Kirana shops, footwear
Manufacturer 6. Home based businesses/small factory
- Manufacture of shoes, bangles, imitation jewelry, piece-rate tailors
Page 8
Trading
9
Findings: Business Characteristics
Business Type Monthly Sales
(in INR)
Monthly Cash Flows:
Inventory rotation
Business Assets #of employees
on Cash on Credit Inventory Furniture/Tools Others
1. Semi mobile businesses
30,000 - 60,000
100% 0% 3 days - 1 week
80%
10% 10% None (mostly family members)
2. Small businesses but at fixed
50,000 - 1,00,000
~80%
<20% 1 week - 3 weeks
70% 20% 10% One to two employee + family members
3. Small businesses but at fixed location: Services related
30,000 - 50,000
~90% <10% Own Inventory: negligible
10% 80% 10% None
4. Small businesses but at fixed location: Food related
50,000 - 1,00,000
80% <20% 1 week - 3 weeks
60% 30% 10% None
5. Wholesalers /large retailers
1,00,000 - 3,00,000
~60% <40% 4 weeks - 6 weeks
90% 10% NA One to three employees
6. Home based businesses/small factory
50,000 - 3,00,000
<10% ~90% 4 weeks - 6 weeks
80% 10% 10% Up to 10-20 workers paid on per piece basis
Page 9
location: Trading
10
Recommendations: Product Features
Business Type Loan Requirem
ents
Cycle Loan Amount (in INR)
Loan Term (months)
Seasonality/Moratorium
Repayment Frequency & Method
Document Proof
Collateral
1. Semi mobile businesses
Working capital loans and business asset loan
First 10,000 to 15,000
6-9
None Daily or Weekly, repayment via cash
ID proof Residence proof
Co-signor required
Second 15,000 to 20,000
9-12
Third 15,000 to 20,000
9-12
2. Small businesses but at fixed location:
Working capital loans and business asset loan
First 15,000 6-9
None Weekly or Fortnightly, repayment via cash
ID proof and any address proof (residence/business)
Co-signor required
Second 20,000 to 25,000
9-12
Third 25,000 9-12
Based on a detailed analysis of business cash flows, and using Swadhaar’s current experience in individual lending, standardized product features were developed for each category:
Page 10
Trading
11
Recommendations: Product Features
Business Type Loan Requirem
ents
Cycle Loan Amount (in INR)
Loan Term
(months)
Seasonality/Moratorium
Repayment Frequency & Method
Document Proof
Collateral
3. Small businesses but at fixed location: Services related
Working capital loans
First 15,000 6-9
None Weekly or Fortnightly, repayment via cash
ID proof Residence proof
Co-signor required
Second 20,000 to 25,000
9-12
Third 25,000 9-12
Building business assets
First None NA
Second 20,000 to 30,000
9-15
Third 30,000 9-18
4. Small businesses but at fixed
related
Working capital loans
First 15,000 6-9
None Weekly or Fortnightly, repayment via cash
ID proof Residence proof
Co-signor required
Second 20,000 to 25,000
9-12
Third 25,000 9-12
Building business assets
First None NA
Second 20,000 to 30,000
9-15
Third 30,000 9-18
Page 11
location: Food
12
Recommendations: Product Features
Business Type Loan Requirem
ents
Cycle Loan Amount (in INR)
Loan Term
(months)
Seasonality/Moratorium
Repayment Frequency & Method
Document Proof
Collateral
5. Wholesale/ Working capital loans
First 30,000 to 50,000
6 None Monthly, repayment via cash/cheque/ECS. From second cycle onwards via cheque/ECS)
ID proof and any address proof (residence or business)
10% upfront margin
Second Upto 65,000 6-9
Third Upto 80,000 6-12
Fourth Up to 1 lakh 6-12
Building business assets
None (No significant demand for an asset loan)
6. Manufacturing: Home based
factory
Working capital loans
First 30,000 to 50,000
6 During lean period or at beginning of loan term: only interest has to be paid, principal to be adjusted in remaining months. Moratorium maximum of 3 months.
Monthly repayment via cash/cheque/ECS. From second cycle onwards via cheque/ECS)
ID proof and any address proof (residence or business)
10% upfront margin
Second Upto 65,000 6-9
Third Upto 80,000 6-12
Fourth Up to 1 lakh 6-12
Building business assets
First and Second
None NA
Third 50,000 to 1 lakh
6-12
Fourth 50,000 to 1 lakh
9-18
Page 12
businesses/ Small
Large retailers
1. Tailoring products to suit the needs of specific business sectors
2. Identifying a suitable methodology to design a preliminary credit scoring model for better client selection
3. Using standardized margins and household expenses for client’s capacity evaluation
4. Using key business indicators to track client’s growth across cycles
Analysis of Key Tools / Processes
Page 13
14
Overview of Credit Scoring
• Why Scoring
• Swadhaar is looking for ways to further streamline its credit process from the application stage to the disbursement stage to make its products/processes customer friendly and efficient, and to mitigate risk in client selection
• Where does Scoring fit in the Credit Process
• Scoring of clients would enable Swadhaar to prioritize client visits for renewals and collections and to offer products based on their risk categories. Scoring can also be used to select clients at time of application
• Swadhaar’s experience with scoring • Swadhaar has implemented a Renewal Score as part of the process to select clients
for renewals. However, it hasn’t been able to implement a score for client selection as existing clients haven't completed sufficient number of cycles with Swadhaar to build a large enough database
Page 14
15
Benefits of Credit Scoring
A credit score would allow Swadhaar to categorize its customers as high, medium or low risk. This information could then be used to:
•Determine what products to offer
• Loan amount, term, interest rate, etc. when an applicant applies for a loan
•Prioritize visits to be made
• At the time of renewals, our credit score informs us which clients are our ‘best’ clients and hence can be renewed earlier
•Improve loan officer productivity and efficiency
• An MIS segregating our clients allows for closer monitoring of renewals and provide better service to more deserving clients
Page 15
16
Methodology to develop a Credit Scorecard
Credit Scoring – Three types of methods may be used
1. Expert Methodology
Built using the knowledge of an ‘expert’ or a group of ‘experts’. Based on their past experience/learning, experts identify variables and assign weights to these variables
2. Combination of Expert selection and Data analysis/validation
Expert selection card is used to identify variables and data analysis to check, correct/adjust and validate the score
3. Statistical Methodology Using a regression model to examine how a given variable is affected by another
variable(s)
Page 16
17
Development of selection scorecard using the expert methodology
Steps followed in this process are as follows
1. Identification of variables: A group of experts is asked to identify variables to be
considered when evaluating a client’s willingness to pay
2. Assigning weights to variables:
a. Each variable is given a score from 1-7, (1=insignificant, 4=neutral, 7=critical) by
each expert; the weighted score from all experts gives the significance of the
variable.
b. Variables are selected and weighted based on the weighted expert scores.
c. The attributes (“values”) of each variable are then scored by the experts to
identify how they affect the client score (e.g. if education is the variable, then its
attributes would be illiterate, public school, private school or college); each
different attribute would get a score such as -10, 0, 5, 10.
3. Calculating client score: Expert scores are constructed viz:
a. For a client, the scores from all variables are added to get the final score.
b. The higher the client’s final score, the lower the risk category s/he represents
Page 17
18
Development of the selection scorecard (1/2)
1. Identification of variables: 29 variables were proposed, e.g. Demographics (e.g. age, civil status, education, gender), Geo-demographic stability (e.g. business & home
2. Assigning weights to variables: Given below is an example how the final score is calculated for each variable and feature/attribute:
a. All variables received a score from the experts; e.g. the age variable received the average score 4.8 by the experts (4 being neutral, meaning that age is significant)
b. 11 variables were selected for the final score, e.g. the age variable was selected for the score based on the score given; variables selected included
• Age and gender
• Civil Status
• Home ownership
• Number of dependents
• Etc.
c. Attributes and scores were assigned for each variable;
e.g. age attributes/scores:
Age attribute range
Score
18 – 30 yrs 10
31 – 50 yrs 50
> 50 yrs 10
Page 18
ownership), Family stability (e.g. household size, children’s education), etc.
19
Development of the selection scorecard (2/2)
3. Calculating client score:
a. Final client score is obtained by adding scores from all variables:
b. Segmentation of clients into categories based on the score:
Client score range (example) Category
>= 350 AA-Low Risk or Best Clients
150 to 349 A- Low to Medium Risk Clients
0 to 149 B- Medium Risk Clients
-500 to -1 C- High Risk Clients
<-500 D- Unable to score
Variable Attribute Score
Age 35 50
Civil Status Single -40
Gender Male 0
Home ownership Own house 70
… … …
Total 105
Page 19
20
Methodology to develop a Renewal Scorecard
When developing the renewal score at Swadhaar a combination of the expert methodology and data analysis was used to check and validate the expert score
Steps followed in this process were as follows
1. Identify variables: A number of repayment performance related variables were identified during the analysis.
• Scores are assigned for each attribute (e.g. value) of a variable based on past experience and expert opinion.
• Based on aggregate weighted scores across all the variables, the score range for segmentation of clients into categories is defined.
3. Back-testing: Historic client data is then used to score a sample of existing clients. This helps:
• To validate the performance of clients post renewal (to see if client behaved the way we thought s/he would)
• To check that the bands and variables are stable
4. Adjustment: Adjustment of the variables and scores assigned to them is done at this stage based on the results of the back-testing
5. Validation: Once the final adjusted score is arrived at, it is validated against another sample of clients from the database to ensure that the results are the stable
Page 20
2. Analysis:
1 Ratio Days Late is defined as the ratio of the maximum days late during the month in the term of the loan and the term of the loan given 2 A higher score means a better / lower risk client
21
Methodology used to develop a Renewal Scorecard (1/2)
1. Identify variables: Some of the repayment performance related variables identified were:
• Average days past due during loan term,
• Ratio of maximum days late to loan term1,
• Vintage of the client with the institution,
• Resting time between loans,
• Etc.
2. Data analysis (1):
An example of a score band for the variable “Ratio of maximum days late1“ to loan term:
Score band for Value of Variable “Ratio Maximum Days Late1”
Score2
Equal to zero 100
Greater than 0 and less than equal to 0.6 50
Greater than 0.6 and less than equal to 1 0
Greater than 1 and less than 6 -50
Greater than 6 -100
Page 21
22
Methodology used to develop a Renewal Scorecard (2/2)
2. Data Analysis (2): Based on aggregate weighted client score ranges (e.g. from ≤ -500
to ≥ +500), clients are segmented into categories with a recommendation for the type of renewal evaluation and eligibility for loan amount increases in the next cycle:
Client score range (example)
Category Type of evaluation for renewal loan
Recommended increase in loan amount
>= 350 AA-Low Risk or Best Clients No evaluation required Upto 50%
150 to 349 A- Low to Medium Risk Clients
Simplified Evaluation Upto 25%
0 to 149 B- Medium Risk Clients Full Evaluation Same as previous loan
-500 to -1 C- High Risk Clients Reject NA
<= -500 D- Unable to score on account of insufficient information
Full evaluation Same as previous loan
Page 22
1. Tailoring products to suit the needs of specific business sectors
2. Identifying a suitable methodology to design a preliminary credit scoring model for better client selection
3. Using standardized margins and household expenses for client’s capacity evaluation
4. Using key business indicators to track client’s growth across cycles
Analysis of Key Tools / Processes
Page 23
24
Standardization of key metrics in Swadhaar's
Client Capacity Evaluation
A. Issues in Current Process and Benefits of Standardization
B. Standardization Methodology
C. Swadhaar Findings
D. Standardization Review
Page 24
25
Issues affecting loan officer productivity • Individual Loan officers handle multiple
responsibilities – marketing/promotions, client evaluations, and installment collections
• A major part of the loan officer’s time is spent in client’s financial capacity evaluation as he/she collects detailed data on client’s business and household income and expenditure items
• Limitations
• Growth to client base is limited due to excessive time spent on evaluation instead of promotions
• Lack of experience/skill level of loan officer resulting in underestimation/ overestimation of client capacity
• Deliberate manipulation of client evaluation to fulfill loan requirement
• Clients are uncomfortable with detailed questioning on their income and expenses
The Loan officer markets the individual loan product, conducts client
evaluations and collections – hence optimizing his/her time and resources
directly enhances profitability
Page 25
26
Benefits of moving to a more Standardized Process
To overcome these limitations, we decided to use our existing knowledge base and experience to identify commonalities for business and household segments to create benchmarks of key metrics. This helped us in the following ways:
• Increased efficiency and consistency of evaluations, leading to reduced time (~30% lower) in client evaluation
• Ability to scale and replicate the individual lending model to different geographic regions and/or business types due to reduced training time
• Enhanced understanding of market segments and business types, leading to more tailored product offering increasing client satisfaction
• Minimize variability in loan evaluation across loan officers; from a risk perspective, this allows us to manage our portfolio risk across geographies more consistently
• Effectively evaluate client’s financial capacity, thus offering optimum loan terms to the client and improving portfolio quality
• Simplify loan evaluation process could potentially result in 2-3 additional disbursements per month; this would result in an increase in portfolio by Rs 3.0 lakh per loan officer
Caution: Expected benefits from standardization depend on the organization’s strategic focus and operating policies/processes – it is important to assess benefits from the
organization’s viewpoint before undertaking the standardization exercise.
Page 26
27
Sales
Cost of Goods Sold
Gross Profit
Operational Expenses
Net Business Income
Other Household Income
Total Household Income
Detailed Financial Evaluation Process
- =
- =
+ =
Education
Household Surplus
-
-
- -
=
At Swadhaar, several business and household (HH) metrics are used to determine the HH surplus, which then determines clients’ loan amount and monthly installment eligibility.
However, calculating each metric individually takes a large amount of the loan officers time, which adversely affects our customer acquisition rate, and our ability to monitor the accuracy of each income and expense item.
Other Expenses
Rent
Food
Page 27
28
a. Decided the metrics to be standardized
b. Decided the sample size and sampling method for data collection
c. Collected data for our samples from various geographies
d. Checked the quality of the data
e. Performed analysis on the collected data
f. Reported the findings
Each of these steps is discussed in detail in the Appendix - Standardization Methodology
The sales margins and HH expenses at Swadhaar were standardized in the following manner..
Page 28
29
Household (HH) Expenses • The following HH expenses are more uniform and can be standardized – food, clothing,
transportation, health, utilities, education, rent
• HH Expenses was directly proportionate to the number of HH members
• Values varied based on geographic region (e.g. – Mumbai’s rental expenses were ~ 50% higher than Pune, Nasik, and Gujarat)
• We found that HH Expenses varied with the following factors –
Swadhaar’s Findings – Household Expenses
Expense Category Range of Values (INR)
Food, clothing, utilities, health, transportation – 1st HH member 4,000 – 8,000
Food, clothing, utilities, health, transportation – Additional HH member
500 – 1,000 / per member
Rent 1,500 – 4,000
Education -Private School 500 – 1,000 / per student
Education - College 500 – 1,000 / per student
Education - Government School 300 – 600 / per student
Savings, debt payments, insurance payments Vary significantly, capture on actual basis as they cannot be standardized
Page 29
30
Swadhaar’s Findings – Gross Margins
*The variations in margin are caused due to shop location (home
based/residential area/market area) OR product range offered
and similar factors
Different businesses within the sample group were analyzed to arrive at the
following margins -
Business Type Gross Margin*
General Store 15-30%
Kirana Store 12-20%
Cloth Merchant/Garmet Seller/Sari Seller 20-40%
Footware Store 30-40%
Electric/Hardware Store 35-55%
Imitation Jewelry/Cosmetic Items Store 40-50%
Vegetable Vendor 20-35%
Mess/Tiffin Service/Food Stall 45-60%
Restaurant 30-50%
Garage/Automobile Repair 60-80%
Saloon/Beauty Parlour 70-90%
Tailoring 55-80%
Laundry 45-65%
Fabrication 30-60%
Handwork 45-60%
Page 30
31
Swadhaar’s Observations on Gross Margins
Several other salient characteristics of gross margins were found, as listed below -
• The standardized gross margins include the cost of goods sold, but exclude indirect costs such as rent, salaries, utility payments, etc.
• Margins were found to be uniform in the range of business sizes within the sample for businesses of the same type – margins may differ for businesses which are larger or smaller than those considered in the sample
• Margins in the service sector are typically much higher than in trading/manufacturing
• Margins can be standardized across regions, as they do not seem to vary significantly across geographies
• Diversified product offering in the same type of business does not make a major difference in sales margin (For example – cloth sellers, garment sellers, sari sellers had similar margins)
• For business types where sufficient sample size was not available, margins were not standardized – these businesses will have to be relooked at a later stage when more data samples are available
• Caution: If a business location houses two different types of standardized businesses (e.g. xerox and general store), then it is recommended to calculate the business margin rather than using the standardized methodology
Page 31
32
Review of Standardized Values
A process for validation and regular review of the standardized values needs to be in place (review at least once a year)
A review of the standardized values is further needed when:
• Your operations expand to different areas
• Your target client segment changes (E.g. different income group)
• There are changes in the macroeconomic environment (Inflation, regulations, culture, etc.)
• There are changes in the local environment (Free education declared in certain pockets, product demand fluctuates, natural calamities, etc.)
• CAUTION: An incorrect assessment of the current environment and its impact on standardized margins can have adverse consequences on portfolio quality
You can ensure accuracy of your standardized metrics by:
• Consistently using on-the-ground knowledge and assessments of field staff to tweak standardized metrics
• Periodically deploying an independent market research and assessment team to survey various business types and market environments
Your organization can maximize automation and standardization of loan approval process
• By extending standardization of margins and metrics to more business types, segments within business types, etc
Page 32
1. Tailoring products to suit the needs of specific business sectors
2. Identifying a suitable methodology to design a preliminary credit scoring model for better client selection
3. Using standardized margins and household expenses for client’s capacity evaluation
4. Using key business indicators to track client’s growth across cycles
Analysis of Key Tools / Processes
Page 33
34
Tracking client evaluations across cycles is important for …
• Designing product features tailored to client business needs across cycles
• Ensuring that the loan increment policy is based on typical sales/inventory growth across cycles
• Potentially catering to additional loan requirements in renewal cycles (E.g. Asset loans in addition to working capital loans)
• Improved client segmentation
• Client’s ability to translate our loan into visible business growth, may be used as one of the parameters to determine exposure to various business segments for the institution
• Assessing social impact
• Ensuring that a loan in increases client’s business income and also the client’s disposable income
• Assessing increased personal savings/investments
• Understanding impact of loan on household spending
Page 34
35
Swadhaar Individual Client History
• Swadhaar launched its Individual Lending (IL) product in 2007 and has disbursed over 17,000 loans worth INR 35 crore (~ $ 7 million to date)
• IL Product was launched in Mumbai in 2007, where we have upto 6-7 cycles of client history
• IL Product was launched in Gujarat in 2009, where we have upto 3-4 cycles of client history
• IL Product was launched in Pune & Nasik in 2010, where we do not have a significant renewal history of clients
• Snapshot of client historical information with Swadhaar (only records where sufficient information is available on the system are considered)
# of Loan Cycles with Swadhaar # of Clients
1 5468
2 854
3 375
4 135
5 30
6 9
7 1
Total 6872
Page 35
36
Possible Business Growth Indicators
There are several parameters that can be used to assess the impact of individual loans on clients businesses such as:
• Business Indicators
• Monthly Sales
• Net Business Income
• Monthly Household Surplus (Disposable Income)
• Working Capital
• Current Assets
• Inventory
• Each of the above indicators must ideally reflect some growth over cycles
• Non – Business Indicators
• Investment in fixed assets (Household goods, vehicle, real estate)
• Increase in household expenses (Indicating increased purchasing power)
• Increase in savings
• Increased spend on education/health
Page 36
37
Swadhaar observed that Sales and Net Business Income were good indicators for client’s business growth
• The sales and net business income show significant growth from cycle to cycle
• Inventory and current assets remain approximately constant over the first few cycles
* Compared to first cycle
Cycle Average % increase in
Sales *
Average % increase in Net
business income*
Average % increase in Inventory*
Average % increase in
Current Assets*
1 - - - -
2 0% 0% 0% -1%
3 13% 9% -2% -1%
4 25% 14% -2% -1%
5 24% 18% 4% 3%
6 36% 24% Insufficient Data
Insufficient Data
Page 37
38
Findings – Business Growth Indicators
• Growth in the sales/income of the client is observed from the 3rd cycle
• In the initial stages of growth the client may require to make operational adjustments (Eg: Additional salaries, increased transportation, increased overheads) which limits the client’s observed business growth
• It is important to fund the client’s business across multiple cycles for significant business growth
• The net business income increases around 1.5 times slower than the sales
• Some portion of the sales growth is due to inflation in prices, which also affects business costs – hence increases in the net business income is relatively lower
• The increase in the net business income is a better indicator to the business growth rather than sales as it nets out the effect of inflation
• Trading businesses showed a steady growth in sales/income compared to manufacturing businesses (Refer to Business Growth Indicators – Appendix for details)
• As several manufacturing businesses have seasonal orders, the financial figures would depend upon the timing of the evaluation
• In order to evaluate the growth in manufacturing businesses, it is advisable to observe the client’s business at the same time over consecutive years (Eg: Marriage season, etc)
• Inventory and current assets do not vary significantly until the 5th cycle
• Clients may utilize the business loan to clear payables OR purchase goods on cash at discounted rates - This increases the business profitability but does not reflect in the client’s current assets
• Client can support a limited increase in sales with the same level of inventory and current assets
• An increase in inventory and current assets may be observed in higher cycles
Page 38
39
Appendices
Page 39
40
• In order to make the evaluation process more standard and less time consuming, Swadhaar decided to standardize two key metrics in the financial evaluation process to avoid extensive item-by-item data gathering
• Sales Margin (Gross Margin)
• Calculated as (Gross profit/Sales), gives us the gross margin on the business. It includes the cost of goods sold but excludes indirect costs such as rent, salaries, utility payments, etc
• Swadhaar found that while sales vary vastly based on the scale of the business, sales margins themselves can be standardized based on business types
• Manual calculation of sales margin is very time consuming and error prone as it involves collecting information on all the different products sold by the client – quantities sold, cost price, sales price, wastage.
• Clients are not comfortable with sharing such detailed business information
• Household Expenses
• Total of the expenses on food, clothing, rent, utilities, transport, health and education
• Swadhaar found that clients in similar income range largely follow a uniform spending pattern, thus a large part of the household expenses can be standardized
• Clients often underestimate their household expenses, as they typically do not have full information about the household spend; clients are also not comfortable sharing their personal expenses for a business loan
• Standardizing household expenses would give a more correct estimate of household expenses
Swadhaar’s Standardization Methodology a. Decide the Metrics for Standardization
Page 40
41
Caution!
•Organizations need to balance scientific accuracy with resources available while deciding on sample size & methodology (For other methods see Data Sampling Techniques)
•Ensure that sample bias based on geography, business type/scale, community, etc is avoided with an appropriate sampling method and size.
Swadhaar Standardization Methodology
b. Determine appropriate sample size
Note: There is a tradeoff between practicality and scientific accuracy of the results. Each MFI should determine how to allocate resources accordingly.
• Swadhaar’s Sampling Process
• ~5000 Households (HH) to analyze trends in general HH expenses, education expenses and rent expenses.
• At least 20 samples from each business type to standardize sales margins.
• Used random sampling of clients across Mumbai, Pune, Nasik and Baroda.
General Procedure
•Your organization needs to determine the optimum sample size.
•Then, you need to determine the methodology for sample selection (e.g. Random Sampling)
Page 41
42
Caution!
•Organizations need to balance resources/time to make the data collection process as efficient as possible.
•It is best to give each staff member a list of all business type interviews required, so he/she does not spend too much time in searching for a particular business type
Swadhaar Standardization Methodology
c. Collect data to fulfill sample size
Swadhaar Procedure
•Used existing client files to aggregate data.
•Whenever data was not sufficient, additional client interviews were conducted.
•In cases where data showed divergent trends, expert interviews were conducted.
General procedure
•Data for desired sample size may be collected through:
• Personal interviews
• Focus group discussions
• Expert interviews (E.g. to standardize health expenses, collect data through doctor interviews) – see Quantitative Collection Techniques
Page 42
43
Caution!
•Do not remove statistically significant data; take extra caution in defining what constitutes an outlier.
•If it doesn’t fit a dataset, it may be another subset altogether.
Swadhaar’s Procedure
•Used the quartile method to determine upper and lower bound to eliminate outliers (See details of Quartile Method)
General procedure
•Remove outliers: if a data set has values that vary widely from the central value then:
• Remove the data, OR
• Determine the need to create a separate subset for standardization (see Appendix for Errors)
Swadhaar Standardization Methodology
d. Conduct data quality check
Page 43
44
Swadhaar’s Procedure
•Swadhaar used the median value to standardize household expenses and mean values to standardize sales margins
•Wherever similar business types yielded similar margins, they were grouped under a single head – Eg: All types of food stalls, mess and tiffin services.
Caution!
•Ensure that the statistical method used suits the size and quality of data available – not all methods may yield correct results on a small data sample.
•Look for commonalities among data sets to treat different sub-groups as the same group (E. g: During analysis, it was found traders selling cloth, dress material, saris, ready made garments, handkerchiefs, etc showed similar margins, hence were clubbed together under a single head
• Validate findings with a sample of evaluation forms for any deviations
General procedure
•Use appropriate method to standardize (Calculate average, inter-quartile method, etc)
•Make sure to use an adequate sample size for accuracy
Swadhaar Standardization Methodology
e. Conduct the data analysis
Page 44
45
Appendix: Data Sampling Techniques
• Simple Random Sampling
o This is the easiest and best way to select a data sample
o In this technique, select a group of homogenous subjects (a sample) for study from a larger group (a population)
o Each individual is chosen entirely by chance (E.g. Surveying the first 20 clients that walk into a chosen shop)
o Every possible subject of a given population has the same chance of selection
• Other sampling methodologies may be more appropriate if
o Additional information is available on clients
o Detailed analysis of the sample is expected
o There are stringent accuracy requirements and/or cost/operational concerns (E.g. If only limited staff members are available)
• Systematic Sampling
o A statistical method involving the selection of elements from an ordered sampling frame. The most common form is an equal-probability method, in which every nth element in the frame is selected
o Simple random sampling does not use a systematic method of choosing clients, whereas this method does. E. g. : Choosing every 5th house on a particular street
o Important: This method, along with simple random sampling, should be done in homogenous populations with little variation to ensure the best results
Page 45
46
Appendix : Data Sampling Techniques (contd.)
• Stratified Sampling
– This method is useful when the population varies greatly
– This process involves splitting the data into non-overlapping (Mutually exclusive) sub groups or strata which are homogenous in themselves
– The strata must be collectively exhaustive i.e. No population element can be excluded
– A random or systematic sample is then selected from each stratum
– The sample size of each stratum may be proportionate to the population size of the stratum (Proportionate stratification) or may vary (Disproportionate stratification) based on the variances or costs in different strata
– Eg: To standardize sales margins, its important to get a good sample from the following strata – market shops/residential shops, different business scales, etc
– A stratified sample can provide greater precision than a simple random sample with a smaller sample size. However, the administrative costs are typically higher
• Judgment Sampling
– In this method, sample is obtained based on the judgment of an expert who is familiar with the characteristics of the population
– Eg: A particular locality may be chosen to be a representative of the entire city
– When using this method, the researcher must be confident that the chosen sample is truly representative of the entire population
Page 46
47
Appendix: Quantitative Data Collection Techniques
• Quantitative data collection involves the use of numeric data which can be analysed using statistical techniques. Swadhaar has primarily used the following methods to collect quantitative data from the sample population:
o Anonymous Questionnaires/Surveys
– Can be administered to a large number of people at minimal cost
– People may be more truthful while responding to the questionnaires regarding controversial issues especially if their responses are anonymous
– Drawbacks are that many of the people who get questionnaires often don't return them and those who do might not be representative of the originally selected sample
– Different sources for surveys include: Internet, mail, street, and telephone
o Existing Data
– Swadhaar used its internal client database for an initial sample
– If such information is collected by other MFIs or institutions and is accessible, it may be a viable alternative
– Drawbacks: Data from may not be adequate for your organization’s needs, and may contain potential biases or errors
Page 47
48
Qualitative data collection methods can be used to improve the quality of survey-based quantitative evaluations by helping to strengthen the design of survey questionnaires and expanding or clarifying quantitative evaluation findings. These include:
o Open-ended and less structured protocols (i.e., researchers may change the data collection strategy by adding, refining, or dropping techniques or informants)
o Rely more heavily on interactive interviews; including focus groups, panels, and in-depth individual interviews; respondents may be interviewed several times to follow up on a particular issue, clarify concepts or check the reliability of data
o Use triangulation to increase the credibility of their findings (i.e., researchers rely on multiple data collection methods to check the authenticity of their results)
o Findings cannot be generalized to any specific population. However, each case study produces a single piece of evidence that can be used to seek general patterns among different studies of the same issue
o Qualitative techniques takes a great deal of time, and the researcher needs to record any potentially useful data thoroughly, accurately, and systematically, using field notes, sketches, audiotapes, photographs, and other suitable means.
Appendix: Qualitative Data Collection Techniques
Page 48
49
Appendix: Quartile Method
1. Calculate the median of the data group
1. Calculate the 1st and 3rd quartile of the data group
• The number used for the 1st quartile number used should have 25% of the data below it
• Similarly, the 3rd quartile number used should have 75% of the data below it
2. Subtract the 1st quartile from the 3rd quartile to get the inter-quartile spread
1. To remove outliers, you need to calculate the Upper and Lower Bound (UB & LB)
• UB is calculated by taking the median + 1.5 times the inter-quartile spread
• LB is calculated by taking the median – 1.5 times the inter-quartile spread
2. Treat the UB and LB as the range in which the data set needs to be in. Anything beyond the range is characterized as an outlier
Page 49
50
Appendix: How to Avoid Data Collection Errors
Aside from the sampling error of selection bias, MFIs must be cautious of non-sampling errors, which are caused by other problems in data collection and processing. They include:
• Over-coverage: Inclusion of data from outside of the population
• Under-coverage: Sampling frame does not include elements in the population
• Measurement error: E.g. when respondents misunderstand a question, or find it difficult to answer
• Processing error: Mistakes in data coding
• Non-response: Failure to obtain complete data from all selected individuals
Page 50
51
A steady increase in sales is observed between cycles in almost all businesses
• Trading businesses
typically show a steady
increase in sales.
• Manufacturing
businesses display an
erratic sales growth
through cycles,
primarily due to
seasonal orders
• Sufficient data is not
available in higher
cycles
• Some variation in the
values across cycles may
also be due to change in
the field officer which may
result in a slightly different
estimation
Business Cycle 2 Cycle 3 Cycle 4
General Store 6% 12% 16%
Kirana Store 5% 10% 12%
Cloth Merchant/Garmet Seller/Sari Seller 11% 14% 18%
Electric/Hardware Store 20% -2% -14%
Imitation Jewelry/Cosmetic Items Store 4% 9% 4%
Vegetable Vendor 1% 5% 9%
Mess/Tiffin Service/Food Stall 4% 9% 7%
Restaurant 6% -8% 44%
Garage/Automobile Repair 2% 5% 10%
Saloon/Beauty Parlour 6% 10% 10%
Tailoring 7% 13% 25%
Laundry 7% 9% 16%
Fabrication 5% 6% 20%
Footwear Manufacture 7% -8% 88%
Handwork 6% 44% 50% *Entries in red indicate less than 5 data points
Page 51
52
Growth in net business income is more erratic than sales growth in the first 3 cycles
• While sales shows a relatively
steady growing trend, the
growth trend in the business
income is not very strong in
the initial cycles
• This trend may be due to the
adjustments the micro
entrepreneur needs to make
to support the growing sales
– additional transportation
costs, additional employees,
etc. It may take some time for
the business income to
stabilize
• Inflation affects the sales as
well as cost figures, due to
which the growth in business
income is much lesser than
sales growth
Business Cycle 2 Cycle 3 Cycle 4
General Store 2% 9% 13%
Kirana Store 1% 3% -18%
Cloth Merchant/Garmet Seller/Sari Seller 3% 3% 35%
Electric/Hardware Store -4% -4% -8%
Imitation Jewelry/Cosmetic Items Store -4% 13% 14%
Vegetable Vendor 3% -3% 7%
Mess/Tiffin Service/Food Stall 4% 16% 4%
Restaurant 5% -15%
Garage/Automobile Repair 8% -2% -18%
Saloon/Beauty Parlour -2% 11% 13%
Tailoring 1% 6% 13%
Laundry 5% 5% -
Fabrication -3% - -
Footwear Manufacture 10% 4% 27%
Handwork 1% 13% -14% *Entries in red indicate less than 5 data points
Page 52