Measuring Employment Using Big Data on Electronic Salary

07/07/2017 1

Measuring Employment Using Big Data on Electronic Salary Payments

Dr. Anil Kumar Sharma &

Ravi ShankarReserve Bank of India

[email protected]@rbi.org.in

Disclaimer – views expressed here are of authors and not of the organisation they belong

IPS024: Official Statistics in the Age of Big Data

mailto:[email protected]

mailto:[email protected]

Outline

• Background

• Employment Measures

• Employment Data in India

• Electronic Payments Systems in India – ‘Structured Big-data’ used for Salary Payments

• Text Mining and Index Methodology

• Empirical Results

• Conclusion and Way forward

07/07/2017 IPS024: Official Statistics in the Age of Big Data 2

Background

• As per Census of India 2011, for Age-Group 15-59 years, total persons in India are approx. 730 millions

• Of which,

• Main workers 324 millions (112 million above matric)

• Marginal workers 103 millions (20 million above matric)

• Non-workers 303 millions (seeking/available for work 28 million)

• Seeking/ available for work are approx. 55 millions


Employment Measures• Measuring employment level forms key challenge for policy

makers – monetary policy outcomes

• Mostly employment is measured using households’ surveys(Demand-side) and/or industry surveys (Supply-side) at quarterlyfrequency

• Both approaches are time-consuming for large and diverseeconomy like India

• This paper proposes an alternative approach to construct amonthly indicator as a measure of employment based onelectronic Wage/Salary payments data


Employment Data - India• Official Statistics

• Quinquennial Survey on Employment by NSSO: Comprehensive large household survey (101,724 households), conducted once every 5 years

• Annual Employment and Unemployment Survey by Labour Bureau: Household survey having coverage of 680,000 individuals from 140,000 households

• Annual Survey of Industries: Statutory survey covers all factories (230,435factories having approx. 11 million workers)

• Quarterly Quick Employment Survey by Labour Bureau: Coverage of more than10,000 enterprise units across 18 sub-sectors

• Private Agencies Statistics• CMIE’s Unemployment rate: Large sample of randomly selected households with

all members of over 15 years of age. The 30-day moving average is derived from a sample of about 130,000 individuals from about 39,600 households.

• Job Speak Index by Naukri.com: On the basis of job listings added to the site every month, the data shows an increase or decrease in the index. To calculate the index, job listings added to the site in July 2008 have been taken as 1000.


Electronic Payment Systems: Big Data used for Salary Payments• National Automated Clearing House (NACH) – Run by NPCI

• Web based solution to facilitate interbank, high volume, electronictransactions of repetitive and periodic in nature (More than 2 billiontransactions handled during 2016-17)

• NACH System can be used for making bulk transactions towardsdistribution of subsidies, dividends, interest, salary, pension etc. and alsofor collection of payments e.g. telephone, electricity, water, loans,investments in mutual funds, and insurance premium

• From August 2016, NPCI has introduced a distinct NACH Credit product forthe purpose of crediting employee salary accounts

• National Electronic Fund Transfer (NEFT) – Run by RBI

• NEFT is a nation-wide payment system facilitating one-to-one funds transfer (More than 1.6 billion transactions handled during 2016-17)


Electronic Payment Systems in India

volume value volume value volume value volume value volume value volume valuevolume (in

thousand)

value (in Rs.

thousand)volume value volume value volume value

2012-13 69 6,76,841 394 29,022 275 21,780 1 4 - - - - - - - - - - - -

2013-14 81 7,34,252 661 43,786 591 44,691 15 96 87 215 - - - - - - - - - -

2014-15 93 7,54,032 928 59,804 965 66,770 78 582 340 1,221 - - - - - - - - - -

2015-16 98 10,35,552 1,253 83,273 958 69,889 221 1,622 1,404 3,802 - - - - - - 748 488 389 4,041

2016-17 108 12,53,652 1,622 1,20,040 1,117 74,035 507 4,111 2,057 7,916 1,964 838 977 13,105

Apr-17 10 88512 143 12156 95 6991 65 562 213 905 7 22 189 301650 231 431 89 22 61 1444

May-17 10 90171 156 12411 97 6746 67 586 194 692 9 28 193 316724 233 451 91 25 65 1941

Volume in million, Value in Rs. billion

Table 1: Electronic Payment Systems Data - Recent Trends

Data for the

period

RTGS NEFT CTS* IMPS* NACH* UPI*

USSD**

Debit and

Credit Cards

at POS &

PPI # Mobile Banking

0

500

1,000

1,500

2,000

2,500

2012-13 2013-14 2014-15 2015-16 2016-17

Number of transactions handled through NEFT & NACH Systems (in millions)

NEFT volume NACH* volume

0

20,000

40,000

60,000

80,000

1,00,000

1,20,000

1,40,000

2012-13 2013-14 2014-15 2015-16 2016-17

Value of Transactions handled through NEFT & NACH Systems (in Rs. billion)

NEFT value NACH value


Methodology for Text Mining • NACH system, the following methodology is used

• Transactions considered once in the month during dates range from1st to 10th and 25th to 31st

• Created unique user/employer names by merging (or) deletion of names

• Transactions with Net Salary amount <₹ 2,00,000 and >₹ 2,500

• Bulk payments e.g. Direct Benefit Transfers, Life InsuranceCorporation, finance credits, pension type transactions are excluded

• Unique user/employer name which are spread across all the months

• Further fine-tuning to incorporate transactions failed or returned (returnclearing)


Methodology for Text Mining • In NEFT system, the payment message IFN298N06 in MT

Standard is used which has a provision to record Sender toReceiver Information.

• Date of transaction lies between 25th of the salary month to 10th of the next month.

• Remittance instruction includes words such as `SALARY’, or`salary’ or `Salary’, or `SAL’, or `sal’, or `Sal’ as characters.

• Sender message includes words such as `SALARY’, or `salary’or `Salary’, or `SAL’, or `sal’, or `Sal’ as characters or ‘pay’

• If userid field used for salary disbursal is utilised.


Methodology – Employment Index• Weight: Average Net Salary disbursed during the current

month divided by Six Month Moving Average of Net Salary disbursed in the previous six months (i.e. Chain Base) is taken as weight

• Employment Index:

No. of Employees in Current Month*Weight*100Index value = ----------------------------------------------------------------

Moving Average of No. of Employees in the last 6 Months


Empirical Results

Based on the above formulae employment index for the formalsector based on 371 common entities covering more than 4.7million employees is estimated as below

Month April 2015 Oct 2015 Nov 2015 Dec 2015 Jan 2016 Feb 2016 Mar 2016

Employment Index 100 99 154 140 142 115 117

0

50

100

150

200

Oct-15 Nov-15 Dec-15 Jan-16 Feb-16 Mar-16

Chart 1: Employment Index


Conclusion & Way forward• Salary disbursal data collected from payment systems is available on monthly frequency

with a minimum lag. The coverage under proposed approach is the maximum among allthe prevailing surveys on employment conducted in India.

• Limitations

• Irregular data from some employer

• Issues in identifying employer using names

• Net Salary disbursed could not be used for estimating wage rate e.g. Salary perhour.

• However, Net salary disbursed to employees could also act as a good leading indicatorof disposal income which is received in the hands of Indian households which ultimatelygenerates consumption demand in the economy.

• Based on the experience and empirical evaluations, the methodology to constructindices using alternative weighting diagrams would lead to fine-tuning of methodologygoing forward.


Thank you


Documents

Measuring Employment Using Big Data on Electronic Salary