WWCodeHackathon

Women Who Code for Social Good Hackathon Project focusing on financial inclusion and behavioral metrics

Timeline

Oct 12 - Opening

• have a list of variables to regress against payability (pred. lending, payday loan, low credit risk) -> have a clear hypothesis

Oct 14-15

• do expoloratory analysis on our own

Oct 16-20 (work week)

• block hours 9am-12pm (CST) / 7am-10am (PST)
• work on code, final product, demo

Oct 21: finalize end product Oct 22: deadline

=======

To do:

• Oct 14:
  • decide on end product: visualization of the regression, data exploratory maps and graphs that proves our hypothesis (interactive web app from R, i.e. Shiny App)
• Oct 15 & 16:
  • dive deep into the data: picking variables regarding payability (Vy & Tanvi)
  • decide on the details of the demographic characteristics of each personas (Tanvi)
• Oct 17
  • data exploration: different maps and graphs (Vy ?)
• Oct 18 & 19:
  • run regressions for each personas (Vy & Tanvi, split personas half half)
  • iterate regression -> visualize
  • visualize the code on Shiny app (Tanvi & Vy ?)
• Oct 20:
  • demo/walkthrough (Tanvi / Vy split ?)

=======

Personas: (low and income, multicultural - non-white, multigenerational, number of the dependants, presence of 65+ age,

• A: likely rejected
• B: complete opposite of A
• C: basically the same, good candidate, 1 factor cannot control (race, etc.) -> get access or no access
• D: fall short -> give benefit of the doubt

=======

Hypothesis:

1. worst persona better off by looking at the financial history using our regression -> we prove our model is better than the credit evaluation traditional model.
2. millions of america are subjected to snap shot of behaviors for lending decision. We believe that looking at one's financial history and behaviors over time is a better indicator and a more comprehensive evaluation of lendabiliy

=======

Methodology

1. Identify the elements that go into credit score model these days

• 1.1 Payment History: This is one of the most significant factors in your credit score. It assesses whether you have paid your bills on time and includes information about late payments, accounts in collections, and any public records like bankruptcies or tax liens.
• 1.2 Credit Utilization: This factor looks at the percentage of your available credit that you are currently using. Keeping your credit card balances low relative to your credit limits can positively impact your score.
• 1.3 Length of Credit History: This considers the age of your credit accounts, including the average age of your accounts and the age of your oldest and newest accounts. A longer credit history often leads to a higher score.
• 1.4 Types of Credit: Lenders want to see that you can handle different types of credit responsibly. This factor takes into account credit cards, installment loans, mortgages, and other types of credit accounts.
• 1.5 New Credit Inquiries: Opening multiple new credit accounts in a short period can negatively affect your score. Each hard inquiry that occurs when you apply for credit can impact your score slightly.
• 1.6 Public Records: This includes information about bankruptcies, tax liens, and civil judgments. These items can have a significant negative impact on your credit score.
• 1.7. Total Debt: The total amount of debt you owe, particularly in relation to your credit limits, can influence your score.
• 1.8 Available Credit: The total amount of credit available to you can affect your score. Having a higher credit limit, while maintaining low balances, is generally favorable.
• 1.9 Payment Patterns: Some credit scoring models also consider the patterns of your payment history, such as the frequency and consistency of payments.
• 1.10 Credit Mix: A diverse mix of credit types (credit cards, loans, mortgages) can positively impact your credit score

3. Recreate the credit score model in terms of payability: Regress those credit score to a variable of payability
4. Regress our chosen alternative variables against payability

=======

Variables

Year: 2009-2021

FAMILY LEVEL DATA

Income and Expenditure

a) Static Income and Expenditure Profile

• ER78091 A31 DOLLARS RENT
• ER78093 ACCURACY OF RENT
• ER78098 A33 GOVT PAY PART RENT?
• ER78114 A42A FUEL EXPENSE PER
• ER78117 A42C COMBINED GAS/ELECT EXPENSE
• ER78122 A44 TELEPHONE EXPENSE PER
• ER78125 A45B TOTAL OTR UTILITIES
• R81914 TOTAL EXPENDITURE (mortgage & property tax - own a home)
• ER81915 TOTAL CONSUMPTION WITH RENTAL VALUE (rent value - rent) -> compare with household income
• ER81775 TOTAL FAMILY INCOME-2020
• ER81630 OVERTIME INCOME OF REF PERSON-2020
• ER81631 ACC OVERTIME INCOME OF RP-2020 -> hustle (extra money)
• ER79460 G44D WTR ALIMONY INCOME-RP
• ER79508 G44G WTR ANY OTHER INCOME-RP
• ER72940 F82 WTR SCHOOL EXPENSES
• ER81354 M4 WTR DONATED TO ORGANIZATION FOR NEEDY

Financial Status

a) Static Financial profile

• ER81836 IMP WEALTH W/O EQUITY (WEALTH1) 2021 -> change over year (up and down)
• ER81837 ACC WEALTH W/O EQUITY (WEALTH1) 2021
• ER78951 F67 LOAN PMT AMT PER #1
• ER78950 F67 LOAN PAYMENT AMT #1
• ER80001 W38A WTR HAVE CREDIT/STORE CARD DEBT
• ER79062 GCOVID16 HOW MNG FINAN - TAKE OUT A LOAN
• ER77487 IMP WTR STUDENT LOAN DEBT (W38B1) 2019
• ER77488 ACC WTR STUDENT LOAN DEBT (W38B1) 2019
• ER77489 IMP VAL STUDENT LOAN DEBT (W39B1) 2019
• ER77490 ACC VAL STUDENT LOAN DEBT (W39B1) 2019

b) Financial Behavior

• ER81354 M4 WTR DONATED TO ORGANIZATION FOR NEEDY
• ER34995 H5N7/H50G WTR CHNGE HNDLNG MONEY ISSUE21
• H11E WTR DIFFICULT MANAGE MONEY-SP
• ER80697 H11E WTR DIFFICULT MANAGE MONEY-RP
• ER73374 G44E WTR HELP FROM RELATIVES-RP
• ER73376 AMOUNT HELP FROM RELATIVES PER-RP
• ER73375 AMOUNT HELP FROM RELATIVES-RP
• ER73377 ACCURACY OF HELP FROM RELATIVES-RP
• ER73888 W38B WTR HAS LOANS FROM RELATIVES
• ER73905 W39B4 AMOUNT LOANS FROM RELATIVES
• ER72983 GSD10 WTR PUT OFF PAYING BILLS -> this variable is only specific to 2019 because of 2018 govt shut down

Education and Employment

a) Employment Behavior

• ER78408 BC8 WTR UNEMPLOYED(RP) -> unemployed during any time in 2020
• ER78427 C7 WTR OUT OF LABOR FORCE (RP) -> data exists for every 2 years since '03
• ER78477 BC68 CKPT: WTR NOT WRKING/NOT LOOKING
• ER81189 L70 #YRS WRKD SINCE 18-RP
• ER79561 WTR WORK HRS PRO PRACTICE/TRADE-SPOUSE
• ER79562 G18GIGA WTR GIG WORK - SP
• ER79933 W10 WTR OWN BUSINESS/FARM

b) Education

• ER76916 L49 GRADE OF SCHOOL FINISHED-RP

• R81336 IMMSTATUS - SP

• ER76690 H61D2 WTR ANY FU MEMBER HLTH INSURANCE

• ER76692 H61J PER FU INSURANCE PREMIUMS

• ER72413 BC62 WTR EVER WORKED

• ER72407 BC60ACKPT WTR CURRENTLY WORKING

• ER72986 GSD13 WTR SOLD BELONGINGS

• ER72985 GSD12 WTR USED MONEY FR RETIREMENT ACCTS -> this variable is only specific to 2019 because of 2018 govt shut down

• ER72968 F91 COST OF OTR RECREATION LAST YEAR -> this variable is only specific to 2019 because of 2018 govt shut down

• ER72727 F1H HOW OFTEN INTERACT W/OTHERS-RP

• ER72728 F1I HOW OFTEN PHYSICAL ACTIVITIES-RP

• ER72729 F1J HOW OFTEN MENTAL ACTIVITIES-RP

• ER72726 F1G LEISURE HRS-REF PERSON

• ER72725 F1F EDUCATIONAL ACTIVITY HRS-REF PERSON

• ER72724 F1E VOLUNTEERING HRS-REF PERSON

• ER72723 F1D2 ADULT CARE HRS-REF PERSON

• ER72722 F1D CHILD CARE HRS-REF PERSON

• ER72721 F1C SHOPPING HRS-REF PERSON

• ER72720 F1B PERSONAL CARE HRS-REF PERSON

• ER72794 F19 WTR FOOD DELIVERED TO HOME -> used food stamp last month

• ER72804 F23 WTR FOOD DELIVERED TO HOME -> not used food stamp last month

• ER72790 F17 WTR BUY FOOD TO USE AT HOME

• ER72792 F18 COST OF FOOD AT HOME PER -> used food stamp last month

• ER72802 F22 COST OF FOOD AT HOME PER -> not used food stamp last month

• ER72796 F20 COST OF DELIVERED FOOD PER (time unit) -> used food stamp last month

• ER72806 F24 COST OF DELIVERED FOOD PER (time unit) -> used food stamp last month

• ER72799 F21 COST OF FOOD EATEN OUT PER (time unit) -> used food stamp last month

• ER72809 F25 COST OF FOOD EATEN OUT PER (time unit) -> used food stamp last month

• ER72958 F89 COST OF CLOTHING LAST YEAR

• ER72963 F90 COST OF TRIPS, VACATIONS LAST YEAR

• ER72964 F90 TIME UNIT FOR TRIPS, VACATIONS

• ER72968 F91 COST OF OTR RECREATION LAST YEAR

• ER72953 F88 COST OF HHOLD FURNISHINGS LAST YEAR

Demographic:

• ER76960 L68A RELIGIOUS PREFERENCE-RP
• R76897 L40 RACE OF REFERENCE PERSON-MENTION 1
• ER76898 L40 RACE OF REFERENCE PERSON-MENTION 2
• ER76899 L40 RACE OF REFERENCE PERSON-MENTION 3
• ER76900 L40 RACE OF REFERENCE PERSON-MENTION 4
• ER76901 L40A ASIAN ETHNICITY OF REFERENCE PERSON
• ER76902 L41 ETHNIC GROUP-RP
• ER76906 YEAR HIGHEST EDUCATION UPDATED-RP
• ER72747 F6CCKPT WTR FU MEMBER UNDER 16 LAST YR
• ER72745 F7 CKPT: WTR CHILD 5-18 IN FU LAST YEAR
• ER72767 F7CCKPT WTR FU MEMBER 60+ LAST YR

Key words: Loan/lending:

• Mortgage
• Refinance
• foreclosure, bankrupt

Living expense

• rent, grocery, insurance, gas
• utilities, utility, eletricity, water, heating, wifi
• house, mortgage, propterty tax,

======= 10/16/23

• Problem: only have variables about repayment loans for car and mortgage, no variables on repayment of other loans, no variables on paying utilities
• Solutions:
  • Look at behavioral of an individual over time! (finish highschool, move out by 18, etc.)
  • Research on Behavioral changes as indicator of trustfulness, trustworthiness, etc. that relate to payability: Character Lab (UPenn) -> child dev -> prediction of child flourish, psychology of money (Book)
  • Taking a multicultural, multigenerationl perspective, justifying the take on behavioral and cultural turn
  • Controlling: first gen (white, hispanic, etc.) control for baseline characteristics to make arg tighter, this person have the non-control feature -> always the preferred candidate A person's payability depends on family level payability (multicultural and multigenerational household) How should financial institution go about looking at the financial data of a family

Credit payment: Ability and willingness to pay to debt (equally determine how they pay for that):

• Impatience, impulsiveness (presence bias, discount future, borrow today, harm future by taking on debt today) -> high borrow today = high default in the future Risk tolerance and trustworthiness
• self eval fin health -> strong presence bias -> more risky Variables:
• borrow for post secondary education
• laid off