thick description in empirical economics - world...

Thick description in empirical economics

Jonathan Morduch

New York University ABCDE, World Bank

June 20, 2016

Thick description: Explains context of behavior, to give it meaning. Observe actions and, from those, piece together meanings Immersive Interpretive [Inductive] (Geertz 1973, Ryle 1971)

Paradox of social science

• We are becoming more “scientific” – Causality – Heterogeneity, LATE, the contingency of results – Pre-analysis plans, protocols – “High-altitude data”

• Need complementary data that explains context – Systematic, clear economic definitions – Probe aggregation – Preferences, constraints and mechanisms – “Low-altitude”

Opportunity

• Researchers are in the field more than ever • How to best exploit that opportunity?

How do you live on

$1.90 a day?

The Economist. October 2015. “Poor statistics. The tricky work of measuring falling global poverty. The number of poor people is declining, but the data are fuzzy” http://www.economist.com/news/finance-economics/21673530-number-poor-people-declining-data-are-fuzzy-tricky-work-measuring-falling

$2…

Can’t even buy a small coffee

If you earn $2 a day, it’s easy to assume…

You live hand-to-mouth

You can’t plan for the future You can’t save

You can’t have much of a

financial life

By Daryl Collins, Jonathan Morduch, Stuart Rutherford and Orlanda Ruthven

www. USfinancialdiaries.org | ©, 2014

US Financial Diaries

• Jonathan Morduch, Professor of Public Policy and Economics, New York University

• Rachel Schneider, Senior Vice President, Insights and Analytics, Center for Financial Services innovation

Principal Investigators

The US financial diaries were created jointly by the NYU Financial Access Initiative, the Center for Financial Services Innovation, and Bankable Frontier Associates.


Support

Leadership support for the project is provided by the Ford Foundation and the Citi Foundation, with additional support

and guidance from the Omidyar Network.

U.S. Financial Diaries

PRELIMINARY DATA - DO NOT CITE

235 households allowed us to track every dollar they earned, saved, borrowed, shared, and spent for a full year.

Macon and Columbus, August 2013, Photo: Whitten Sabbatini

Financial Diaries • Atlanta • Glasgow • Tokyo • NYC, Mississippi, Kentucky, Ohio, California • Indian SME • Bangladesh (Hrisipara) • Myanmar • Kenya • Mexico

• Bankable Frontier Associates • Microfinance Opportunities • Independent researchers

Methodology and Data

Why digital data aggregation would miss a lot Fed Cash Product Office (April 2014) Diary of Consumer Payment Choice

USFD: 40% on average

USFD: 50% on average

Empirical Progression

INCOME ASSETS CASH FLOW

244 Households 316,763 cash flows 100 spending categories 38 income types 69 financial instruments

High-frequency data Households surveyed every 2-4 weeks

US Financial Diaries

Village studies • ICRISAT Village-level studies. 6 villages. 1974-85, 2001-4 • Palanpur, 1 village in Uttar Pradesh. 112 households.

1957/58, 1963/64, 1974/5, 1983/84, 2008-10 • Townsend Thai monthly village surveys. 1998 onward

• Insight: agricultural household model • Utility (home-grown good, market good, leisure)

maximized subject to – Cash income constraint – Time constraint – Production function

Structure • Corporate financial tools

– Balance sheets – Income statements – Cash flow statements

• Adding up constraint: inflows vs outflows

Krislert Samphantharak and Robert Townsend. 2009. Households as

Corporate Firms: An Analysis of Household Finance Using Integrated Household Surveys and Corporate Financial Accounting. Cambridge [Econometric Society Monographs].

Tracking the “Modes” of Payments

Jan received a paycheck and cashed it the same day Richard received wages onto prepaid card EBT card balance increased automatically Paid rent Paid vehicle loan, rent-to-own, recreation, alcohol Bought food, gasoline, clothing, cigarettes

$470

$800

$317

-$625

-$461

-$212

First Weekend

in June 1st-3rd

Received child care payment and cashed it Bought food Paid for vehicle maintenance Bought gasoline, food Bought clothing

$270

-$110

-$477

-$37 -$10

First Full Week

(Mon 4th –Sun 10th)

Jan received a paycheck and cashed it the same day Richard received wages onto prepaid card Paid loan, utilities Paid loan, insurance, bought food, prescription drugs

$781 $722

-$749

-$611

Second Week

Received money from mother Bought cigarettes, gasoline Bought food, housekeeping supplies

$30

-$23 -$56

Third Week

Received money from friend $30 Fourth Week

Household Example: “Jan and Richard” from June 1 to June 30 Amount

Margin of error US Financial Diaries

0%

20%

40%

60%

80%

100%

120%

140%

1 6 11 16 21 26Interview Number

67%

12%

Margin of error

Missing outflows + inflows divided by total outflows

It took 6 interviews (3+ months) to bring margin of error down substantially

$0$500

$1,000$1,500$2,000$2,500$3,000$3,500

25thPercentile

Median 75thPercentile

Income Distributions Earlier and Later in the Study

Initial QuestionnairesEnd-of-Study Cash Flow…

Precision over time

25%-37% increase

$0.78 per person per day

x 2.67 = $PPP 2.08 per person per day

Microfinance savings account

Saving with a moneyguard

Home savings

Life insurance

Remittance to home village

Loans to others

Cash in hand

Microfinance loan

Interest free loan from neighbor

Wage advance

Savings held for neighbors

Shopkeeper credit

Rent arrears

Volatility


$2 a day is just an average…… Seasonal variations in monthly income

$0

$20

$40

$60

$80

$100

$120

$140

$160

$180

$200

Aug Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug

Traders

Small Farmers

Appendix Figure A.1

-0.726398441

-0.9199011636

-3.4291159702

-0.6103714945

-1.5955796526

-0.3834176526

-0.0545323493

-0.3044022478

-0.0817491755

-0.2180559357

-0.2964965682

-0.0637464449

-0.3225552049

0.0124516578

0.0810273982

-0.0076981393

-0.0197246812

0.0301176016

0.019425984

-0.0819182362

0.0480258092

-0.1001486823

0.0264796019

0.0891200749

0.1569383251

0.1090021209

0.1094809482

0.4836879491

0.7164715026

0.3333333333

After about 6 interviews, the margin of error of data collection decreases to an average of 6%.

Number of interviews

Figure 5.3

End-DecEnd-Dec

JanJan

FebFeb

MarMar

AprApr

MayMay

JunJun

JulJul

AugAug

SepSep

OctOct

NovNov

DecDec

But saving at the ASCA savings rate of 30% per month gives you a return of 734% over the year, or $1322!

Principal

Interest

Principal

Interest

100

0

200

30

300

99

400

218.7

500

404.31

600

675.603

700

1058.2839

800

1585.76907

900

2301.499791

1000

3261.9497283

1100

4540.53464679

1200

6232.695040827

1300

8462.5035530751

Figure 5.2

End-DecEnd-Dec

JanJan

FebFeb

MarMar

AprApr

MayMay

JunJun

JulJul

AugAug

SepSep

OctOct

NovNov

DecDec

Adding $15 per month at 6% per annum would give you a return of 15% or $26

Interest

Principal

Principal

Interest

100

0

200

0.5

300

1.5025

400

3.0100125

500

5.0250625625

600

7.5501878753

700

10.5879388147

800

14.1408785088

900

18.2115829013

1000

22.8026408158

1100

27.9166540199

1200

33.55623729

1300

39.7240184764

Figure 5.1

2

4

5

6

7

8

10

11

12

13

15

16

18

19

22

25

26

30

31

33

35

36

37

41

47

51

52

65

70

79

80

84

90

131

199

202

236

Days

Term for which loans are quoted

3.6242977306

1.5533008589

3.8989627457

0.6

0.690803232

1.088955762

0.7248595461

0.3294816119

0.75

0.3719709711

0.4103656748

0.4530372163

0.4841154824

0.3947368421

0.4991255506

0.3369316877

0.2389693629

0

0.3129686178

0.1769202652

0.1775200982

0.1343228507

0.1385030216

0.1145700612

0.1028855877

0.079935232

0.0929927428

0.1398190386

0.1035513637

0.1224753916

0.0829875341

0.0719769528

0.0833333333

0.0501594781

0.0408640499

-0.0020922023

0.0563555595

Figure 4.2

-31

-31

-3193

-31

-31

-3193

Each month each member gives $31

And then receives $93 when it's her "turn"

And then the cycle repeats again….

Figure 4.1

-9

-9

-9

-9

-9

-9

-9

-9

-9

-9

-999

Every month each member contributes $9

On the final month, the members split the accumulated funds and each receive $99

Figure 2.3

3/20/20043/20/2004

3/26/20043/26/2004

4/2/20044/2/2004

4/8/20044/8/2004

4/16/20044/16/2004

4/23/20044/23/2004

4/30/20044/30/2004

5/8/20045/8/2004

5/15/20045/15/2004

5/21/20045/21/2004

5/29/20045/29/2004

6/4/20046/4/2004

6/11/20046/11/2004

6/17/20046/17/2004

6/24/20046/24/2004

7/2/20047/2/2004

7/9/20047/9/2004

7/23/20047/23/2004

7/30/20047/30/2004

8/6/20048/6/2004

8/13/20048/13/2004

8/20/20048/20/2004

8/27/20048/27/2004

9/4/20049/4/2004

9/10/20049/10/2004

9/17/20049/17/2004

9/25/20049/25/2004

10/2/200410/2/2004

10/9/200410/9/2004

10/16/200410/16/2004

10/22/200410/22/2004

10/30/200410/30/2004

11/6/200411/6/2004

11/12/200411/12/2004

11/19/200411/19/2004

11/27/200411/27/2004

Used money lender loan

Used savings club payout

Revenues plus credit collections

Inventory purchase and expenses

60

53.8461538462

46.9230769231

35.3846153846

46.1538461538

33.8461538462

43.0769230769

27.6923076923

35.3846153846

22.3076923077

60

33.8461538462

46.1538461538

26.1538461538

62.4615384615

35.3846153846

73.8461538462

49.5384615385

78.4615384615

41.5384615385

86.1538461538

51.5384615385

43.0769230769

24

69.2307692308

40

20

10.1538461538

18.4615384615

13.8461538462

23.8461538462

15.3846153846

29.2307692308

9.6923076923

55.3846153846

32.3076923077

21.5384615385

24.6153846154

41.5384615385

24

35.3846153846

30.1538461538

38.4615384615

26.1538461538

36.9230769231

40

63.0769230769

32.3076923077

69.2307692308

50.7692307692

61.5384615385

36.9230769231

70.6923076923

43.6153846154

69.6153846154

43.6153846154

48.0769230769

30.7692307692

87

63.6923076923

71.8461538462

53.3846153846

80.3076923077

38

96.4615384615

52.3846153846

97

50.8461538462

45.6923076923

40.0769230769

97.5384615385

76

Figure 2.2

AugAug

SeptSept

OctOct

NovNov

DecDec

JanJan

FebFeb

MarMar

AprApr

MayMay

JuneJune

JulyJuly

AugAug

&A

Page &P

Small farmers

Traders

25.2522796353

23.094224924

21.8054711246

17.3556231003

10.5410334347

15.7142857143

15.6291793313

57.2188449848

56.7933130699

16.4133738602

23.5319148936

34.4224924012

41.7933130699

46.3404255319

22.641337386

175.0607902736

42.8145896657

135.2127659574

66.2613981763

100.1215805471

79.2978723404

74.8176291793

8.9969604863

10.3951367781

42.1854103343

13.358662614

Figure 2.1

Definition 1: Regular wagesDefinition 1: Regular wages0.1428571429

Definition 2: Regular Wages + Casual WorkDefinition 2: Regular Wages + Casual Work0.4395604396

Definition 3: Regular Wages + Casual Work + Self EmployedDefinition 3: Regular Wages + Casual Work + Self Employed0.8571428571

South Africa

Bangladesh

India

0.4214876033

0.1590909091

0.5592286501

0.3818181818

0.6391184573

0.6909090909


Pumza, South Africa

Net cash flows, aggregated weekly, US$


The challenge of living on $2 a day is that $2 a day is just an average

-$10

$0

$10

$20

$30

$40

$50


Spikes and Dips


Few families have steady income

Average Income

+ 25%

- 25%

Average Income

- 25%

+ 25%

Five spikes/dips per year on average US Financial Diaries

2.7 Spikes

2.7 Dips

Self-reported month to month income variability SHED - Federal Reserve, 2013 Survey of Household Economics and Decisionmaking (7/14)

0%

10%

20%

30%

40%

Under $25,000 $25,000-$49,999 $50,000-$74,999 $75,000-$99,999 $100,000 orHigher

Perc

ent o

f Hou

seho

lds

Annual Income

“Some unusually high or low months”

“Often varies quite a bit from one month to the next”

SHED: Implemented in 9/2013. Nationally-representative sample. Online panel of 50,000 individuals sampled randomly. 6,912 asked to take the survey. About 60% (4,134) agreed. Quick survey (19 minutes median time)

Q: Which of the following is more important to you?

A. Financial Stability B. Moving up the income ladder

Portfolios of the Poor: The triple whammy

Low incomes

Irregular and unpredictable incomes

Lack of financial tools

Poverty

Volatility in Palanpur Poverty assessed in different ways

Apparent Prosperity

Current income

Permanent income

Regular job 24% 13% 30%

Agricultural laborer 76 64 59

Landless 70 44 55

Nicholas Stern and Peter Lanjouw. 1991. “Poverty in Palanpur.” World Bank Economic Review 5(1): 23-55. Table 4, p 38.

Poverty is often a condition of volatility (With opportunities

And challenges)

Charles Booth, Poverty Map of London, late 19th c

Dark Blue

Casual earnings: "very poor" (below 18s. per week for a moderate family). The labourers do not get as much as three days work a week, but it is doubtful if many could or would work full time for long together if they had the opportunity. Class B is not one in which men are born and live and die so much as a deposit of those who from mental, moral and physical reasons are incapable of better work.

Light Blue

Intermittent earnings. 18s to 21s per week for a moderate family. The victims of competition and on them falls with particular severity the weight of recurrent depressions of trade. Labourers, poorer artisans and street sellers. This irregularity of employment may show itself in the week or in the year: stevedores and waterside porters may secure only one of two days' work in a week, whereas labourers in the building trades may get only eight or nine months in a year.

Most of the volatility is within job US Financial Diaries

0

100

200

300

400

500

600

700

800

900

1000Janice's bi-weekly paychecks

Janice is not poor on average, but, based on income, she is poor 5 months of the year.

Sometimes poor

xcz

permanent income = x, consumption = c

Not poor Not poor

Not poor

Episodic poverty, transient poverty, poverty spells

Poor

Poor

Poor

91%

Sometimes poor % of USFD households with annual income above the (supplemental) poverty line that spent 1+ month with income under the poverty line

53

100%

200%

97%

65%

43%

150%

4 months 1.5 months

3 months

Chronic and Episodic poverty, 2009-11 Ashley Edwards. “Dynamics of Economic Well-Being: Poverty, 2009–2011” US Census Bureau. January 2014

29%

45% 9%

3%

90 million 10 million

Spell duration, 2009-11 Ashley Edwards. “Dynamics of Economic Well-Being: Poverty, 2009–2011” US Census Bureau. January 2014

Smoothing

Friedman: Permanent Income Little effect of stimulus

Save

Dis-save Income

Spending

Dynamic stochastic consumption, saving/consumption choice

+∑

−

=

−1

00)()1(max

T

tt

t CUE δ

0>tCtt

ttttt

IYA

rCYAA

∈≥

+−+=+

,0

)1)((:tosubject

1

Euler equation Along the optimal path, the marginal value of financial wealth = marginal

utility of consumption

)1/(])1)(([)(

1 δ++= + trCU'ECU'

tt

t

1212 )1/(])1)(([ δ++= + trCU'E tt

120120 )1/(])1)(([ δ++= + trCU'E tt

Next month

Next year

Next decade

Households may be smoothing on some margins but not on others

Soon Now Later

Saving for

Later

Overspending now

Saving for

Soon

Most savings are spent soon US Financial Diaries

85%

7% 5% 3%

44%

17% 16% 23%

0%10%20%30%40%50%60%70%80%90%

Microfinance savings account

Saving with a moneyguard

Home savings

Life insurance

Remittance to home village

Loans to others

Cash in hand

Microfinance loan

Interest free loan from neighbor

Wage advance

Savings held for neighbors

Shopkeeper credit

Rent arrears

From Portfolios of the Poor (2009) 3 needs that drive much of the financial activity: 1. Managing basics: Cash-flow management to transform irregular

income flows into a dependable resource to meet daily needs

2. Coping with risk: Dealing with the emergencies that can derail families with little in reserve

3. Raising lump sums: Seizing opportunities and paying for big-ticket expenses by accumulating usefully large sums of money

Final thoughts

• Not an argument to abandon science, or that qualitative work should be considered more scientific than it is.

• Argument: data collection and qualitative work as theory-building

• Argument: much research is increasingly narrowly scientific but not broadly scientific

• Big data needs little data

Thick description �in empirical economics�Slide Number 2Slide Number 3Paradox of social scienceOpportunitySlide Number 9Slide Number 10$2…If you earn $2 a day, it’s easy to assume…By Daryl Collins, Jonathan Morduch, �Stuart Rutherford and Orlanda Ruthven US Financial DiariesSupportU.S. Financial Diaries235 households allowed us to track every dollar they earned, saved, borrowed, shared, and spent for a full year.Financial DiariesSlide Number 19Why digital data aggregation would miss a lot�Fed Cash Product Office (April 2014)�Diary of Consumer Payment ChoiceEmpirical ProgressionVillage studiesStructureTracking the “Modes” of PaymentsMargin of error�US Financial DiariesPrecision over timeSlide Number 31Slide Number 32Slide Number 33$2 a day is just an average……�Slide Number 36Slide Number 37Spikes and DipsFew families have steady incomeFive spikes/dips per year on average� US Financial Diaries�Self-reported month to month income variability�SHED - Federal Reserve, 2013 Survey of Household Economics and Decisionmaking (7/14)Slide Number 42Portfolios of the Poor:�The triple whammySlide Number 44Volatility in Palanpur�Poverty assessed in different waysSlide Number 46Slide Number 47Slide Number 48Slide Number 49Most of the volatility is within job�US Financial DiariesSlide Number 51Sometimes poorSometimes poor�% of USFD households with annual income above the (supplemental) poverty line that spent 1+ month with income under the poverty lineChronic and Episodic poverty, 2009-11�Ashley Edwards. “Dynamics of Economic Well-Being: Poverty, 2009–2011” US Census Bureau. January 2014Spell duration, 2009-11�Ashley Edwards. “Dynamics of Economic Well-Being: Poverty, 2009–2011” US Census Bureau. January 2014Slide Number 70Friedman: Permanent Income�Little effect of stimulusDynamic stochastic consumption, saving/consumption choiceEuler equation�Along the optimal path, the marginal value of financial wealth = marginal utility of consumptionSlide Number 81Slide Number 82Most savings are spent soon�US Financial DiariesSlide Number 89From Portfolios of the Poor (2009)Final thoughts

thick description in empirical economics - world...

Documents