ATM Refilling Model for a Bank with External ATM Refilling Agency

Banks Desire
Minimum money should be kept in ATMs Minimum refilling servicing as bank has to pay for it By keeping little money in ATMs bank increases servicing costs but also increases investment return profits

Ideal Situation
All ATMs will dry up as expected in the model

ATM Model Rules & Desired Solution

Bank has employed an external servicing agency to refill ATMs similar to what most banks in India do1 ATMs should be refilled with money amount enough to last till next servicing schedule Model should find optimize & find two variables:
Number of Refilling of an ATM per week Refill money amount for that ATM

People Desire
ATM should always be running

Model
Bank informs agency of drying up of ATM Bank informs agency how much it has to be filled Bank gives the agency its ATM servicing schedule Servicing Costs to Bank charged by Agency per ATM = Rs. 4001 Servicing Time - 24x7

Required Calculations
ATM transactions/money amount requirement analysis People cash spend pattern analysis may be a good reflection of this

Model by: Prashant Kumar

ATM Refilling Model

Study of past records

People Spending

ATM Users

ATM Location

Financials

Study of last 3 year transaction records of all ATMs, all Savings/ Current Accounts at all Branches of the Bank and all Card transaction s of select cities/town s whose data can be taken as Sample data for cities/town s similar to them to find weights of all the parameters defined in this model

Income related to balance in savings account in bank branch

Higher Spending Days

Payment Method market region/ merchant specific Malls, supermarkets & High Society Markets prefer Card, rest all markets prefer Cash

Age

Profession

Economy outlook & Employme nt level

Ratio of Current: Savings Account at bank branch

Pref Fem erre ale: d Mal Tran Card e sacti Pen ratio etra on of tion via Good card Cash card own / accep ers Card tabilit
y indica tes that Good peopl mean e s prefe wider r accep using tabilit card y of for card trans actio ns or ATM for withd rawal s Cash Prefe rence will incre ase withd rawal s from ATMs

Offic e Area

Market Area

Residential Area

Special Locations Seasonal Demand

ATM Accessibilit y, Visibility, Location from demand area

Time/Cost required to refill ATM

Earnings of Bank using liquid money Profit = Earning* Profit Margin

Max imu ATM m Cash daily Carr with ying dra Cap wal acity limit

Low amount in Savings High amou Account in nt in Banks Savin Branch gs either people dont have Accou nt in money or Branc people prefer h other banks over ours

Festiv e Salary Seaso Week Day end n Mont Spen spend h ding - ing/s End/F Diwal hoppi irst i/ Eid/ ng Week New Year

Medium & Big Markets Non Malls

Small Markets

stude Salari nt/re Busin ed ess cent peopl 30-60 earne Retire peopl e d e rhave gener have gener have defin ally many less ally deal ed & needs deal in needs in varia cheq cash ble ue needs

Good

Bad

Good more busin ess peopl e

Bad less busin ess peopl e

Cash Spen d: Cash Very Cash Spen Cash Spen Low d: high Spen High PerPer- amou d: d: high nts high capita capita high amou very amou inco amou inco nts high high me me nts nts high regio regio frequ high frequ n n frequ ency frequ ency seaso ency ency 1st nal week variat ion Cash Cash Spen Spen d: d: medi low um amou amou nts nts low medi frequ um ency frequ ency

Cash Spend: high amoun ts high freque ncy

Cash Spend: low amoun ts high freque ncy

Cash Cash Cash Spen Spen Spen d: d: d: low low low amou amou amou nts nts nts very high low low frequ frequ frequ ency ency ency

Cash Cash Spen Spen d: d: high low amou amou nts nts very low high frequ frequ ency ency

Cash Spend: high amount s high frequen cy

Cash Spend: low amount s low frequen cy

Cash Cash Spen Spen d: d: low high amou amou nts nts low high frequ frequ ency ency

Regul ar usage gener ally Colleg Dema Very nd High es by High defin dema low High demand retire dema during month d ed nd usage nd Touri end to pay peopl durin and durin st durin salaries of regul g eg Areas g ar festiv sum helpers and take week out mers/ food/ e home rents ends highe winte trans seaso r port n rs amou nts at Cash Cash a Spen Spen Cash Cash Cash Spen time d: d in Spen Spen d: very Cash Mark Cash d: low d: low Cash medi Spen high Spen et: Spen amou amou d: d: um to amou high d: low nts nts nts amou high high high amou low high amou amou very amou nts frequ frequ nts nts high high nts nts ency ency low low high frequ high frequ frequ 1st, seaso frequ frequ ency frequ ency ency 2nd, nal seaso ency seaso ency ency 3rd variat 4th nal nal week ion week variat variat ion ion

Good

Bad

Branc h attac hed ATM

Standalone ATM

Rs. loan Share 50k, retur s, Max ns bonds 40 10% , notes to assets in a 20% - say single say 25% trans 15% P.A. actio P.A. n1

Small Mach ine : Rs. 20 lakhs Big Mach ine : Rs. 35 lakh2

Cash Spend: high amount s high frequen cy

Cash Spend: low amount s low frequen cy

Labou r charg e

Dista Vehicl nce No. of e& ATMs of Labou requir ATM r ing from charg refill Branc e h

Classifications
ATM Location: City Type Metro / City / Town / Village Tourist City No / Tourism Season / Tourism Off Season Location Office Area / Market Big & Medium / Market Small / Residential Area / College ATM Visibility & Accessibility Good / OK / Bad
Time:
ARBITRARY ALLOTMENT OF WEIGHTS
City Type Metro City Town 1 0.2 0.02 0.001 Tourist City No Tourism Season Tourism Off Season 1 10 3 Location Office Area Market Big & Medium Market Small Residential Area College 0.05 1 0.5 0.01 0.001 ATM Visibility & Accessibility Good OK Bad 1 0.85 0.75

First Week of Month Yes / No Weekend Yes / No Long Weekend Yes / No Big Festive Season Yes / No

Village

First Week of Month Yes 1 No 0.85

Weekend Yes Long 1 Yes Normal No 0.9 0.75

Big Festive Season Yes 1 No 0.75

People Information: Average Per Capita Income High / Good / Average / Low Use of Cards Good / OK / Low Preferred Transaction Type Card / Cash General Information: Economy & Employment Outlook Good / OK / Bad Approach of Calculations: Assuming that biggest spend will be with the following parameters: Metro, No, Market Big, Good, Yes, Yes, Yes, Yes, High, Good, Cash, Good

Average Per Capita Income High 1 Good 0.65 Average 0.1 Low 0.01

Good OK Low

Use of Cards 1 0.5 0.1

Card Cash

Preferred Transaction Type 0.5 1

Good OK Bad

Economy & Employment Outlook 1 0.75 0.3

Maximum Cash Withdrawal Case

Case Parameters
Festive Season Maximum Card Penetration Female : Male Card Ownership Ratio = 1 Market: Medium to Big, Non-mall People/ Merchants prefer Cash Transactions
Assumed Average Top 5% High Worth Withdrawals Assumed Average Top 10% High Worth Withdrawals Assumed Average next 15% High Worth Withdrawals Assumed Average next 50% Withdrawal s Assumed Average lowest 20% Withdrawals Maximum number of transactions in a single day 1 Withdrawal in a single day Max ATM Capacity of Big ATM Max ATM Capacity of Small ATM No. of days a completely filled Big ATM can run on max load No. of days a completely filled Small ATM can run on max load 5% 10% 15% 50% 20% Rs. 25,000 Rs. 10,000 Rs. 5,000 Rs. 2,000 Rs. 500 600 Rs. 24,60,000 Rs. 35,00,000 Rs. 20,00,000 1.4 days 0.8 days

Calculations
Maximum number of transactions in a single day = 6001 Big ATM Max Capacity = Rs. 35 lakh2 Small ATM Max Capacity = Rs. 20 lakh2

Expected Loss in Earnings of Bank by filling cash in ATMs

Assumptions
Assuming annual loan returns = 15%PA Assuming annual shares & bonds returns = 25%PA Assuming bank will earn 20% returns over all investments PA Assuming half is profit to bank after removing recurring costs, we find that bank earns profit of 10% on investment per annum ATM is refilled on the day it is expected to get dry

Calculations
Annual profit on investment = 10% Therefore, per day profit on investment = (10/365)% = 0.027397 % Per day profit lost on keeping money in ATMs = 0.027397% of money kept in ATM Loss will be calculated over a week Loss will be calculated on the cash amount left in the ATM at the end of each day of the week For ease of calculations, it is assumed that while refilling, the ATM cash level is taken till the desired optimized cash level only

Calculation of Running Cost of ATM per week

Let ATM be refilled at start of week Let amount of cash put in ATM be x Rs. Let the per day consumption rate be y Rs./day Let the cost of refilling the ATM be z Rs./refill This consumption rate has already been calculated from a consumption model made above. Therefore, this is a known constant. The no. of days it will last = x/y days Loss on this cash will be calculated at the end of the day, meaning, the cash which is left at the end of the day in the ATM is the extra cash which the bank could have invested and earned a profit on. Summing the daily losses up, we get the loss in keeping money in ATM for x/y days = [x^2/y (x/2)(x/y+1)]/3650 Total no. of refills required in the whole week = 7*y/x Therefore, Total refill costs RF_Cost = 7*y/x*z Rs. For ease of calculations, assuming that we refill the ATM back to the desired level of cash, i.e., Rs. x Therefore, again, sum of daily losses incurred from here till the next refill will be = [x^2/y (x/2)(x/y+1)]/3650 Total no. of such losses will be equal to the number of refills, therefore, total losses in a week will be = {7*y/x} * {[x^2/y (x/2)(x/y+1)]/3650} on simplification Loss = 7/7300*(x-y) Total Costs = RF_Cost + Loss => Total Costs = 7*y/x*z + 7/7300*(x-y) Here only x is variable and both y and z are constants. Therefore, differentiating this w.r.t. x and equating that to zero, gives us the optimized value of x Optimized Refill Level x = sqrt(7300*y*z) Optimized no. of refills per week = 7*y/x Total Costs of running an ATM Total Costs = (7*y/x)*z + (7/7300)*(x-y)

Reference
1. 2. http://www.dnaindia.com/money/report_atm-refilling-costs-up-as-withdrawal-limit-is-hiked_1391225 http://articles.timesofindia.indiatimes.com/2011-07-11/mumbai/29760322_1_atms-bank-officials-refilling