Measuring Poverty

LECTURE 14

Training 1

Inequality and poverty measurement

1) a measure of living standards

persons
2) high-quality data on households’ living
standards
3) a distribution of living standards
(inequality)
4) a critical level (a poverty line) below
which individuals are classified as
“poor” living standard
5) one or more poverty measures
Training 2

Poverty lines

Training 3

1
 How to draw a poverty line? An Overview

DCI
Subjective Relative
Orshansky
Objective Absolute
FEI
Hybrid
CBN

Training

Subjective poverty lines – I/III

§ Poverty lines are inherently subjective judgments people make about

what constitutes a socially acceptable minimum standard of living in a
particular society at a given time (Ravallion 1994: 42).

§ The subjective poverty approach is based on the self-assessed

adequacy of a family's food, housing, and clothing.

§ How are poverty lines estimated, in practice?

Training 5

Subjective poverty lines – II/III

§ A surveys can ask the Minimum Income Question (MIQ):

“What income level do you personally consider to be absolutely
minimal? That is to say that with less you could not make ends meet?”

§ Another possibility is the Economic Ladder Question (ELQ):

“Imagine six steps, where on the bottom, the first step, stand the
poorest people, and on the highest step, the sith, stand the rich (show
a picture of the steps). On which step are you today?”

Training 6

2
 Subjective poverty lines – III/III
World Bank (2017)

§ Recommendation 4:

“The World Bank should explore the use of

subjective assessments of personal poverty
status (in “quick” surveys of poverty), and of
the minimum consumption considered
necessary to avoid extreme poverty, as an
aid to interpreting the conclusions drawn
from the global poverty estimates”.

Training 7

Objective poverty lines

§ An objective poverty line is one

based on some objective metric,
such as consumption or income.

Training 8

Absolute poverty lines

§ An absolute poverty line is one which is fixed in terms of the average

standard of living (or welfare).

§ Example: cost of a bundle containing “basic commodities”, however defined.

§ Note 1: ‘fixed’ is a false friend. An absolute poverty is defined in a specific
context and time, that is is fully historically determined. Fixed ≠ unchanging.

§ Note 2: ‘absolute’ is not a synonym of ‘low’ – an absolute poverty line can

be as generous as the analyst or the society wishes.

Training 9

3
 Relative poverty lines

§ A relative poverty line is one which varies with the average standard of living.
§ Example: half the mean of per capita income.

§ The EU definition of relative poverty line:

“Low income rate after transfers with low-income threshold set at 60% of median
[equivalized] income, with breakdowns by gender, age (...)”
§ Question: why 60%? Why this specific number?
§ Answer: who knows?
Indicator 11 (“Dispersion around the low income threshold”). Three thresholds:
40, 50 and 70% of the median income.
Training

10

The problem with relative poverty: the richer…the

poorer?
poverty line
x1 x2 x3 x4 x5 total mean (50% of the poor
mean)

2 2 16 20 60 100 20 10 40%

3 3 24 170 300 500 100 50 60%

An awkward feature of relative poverty lines is that a policy which raises the living
standards of all, but proportionally more those of the rich, will increase poverty,
notwithstanding the fact that the absolute living standard of the poor has increased!
D u clo s J.-Y . (2 0 0 2 ) P o v e rty a n d e q u ity: th e o ry a n d e stim a tio n . M im e o

Training

11

Relative poverty lines

an assessment

§ In short, relative poverty = inequality

§ In lecture 13 we discussed inequality measures at length – we have better

tools for measuring inequality than ‘relative poverty’ measures
§ However… World Bank (2017):

§ Recommendation 16: The World Bank should introduce a “societal”

headcount ratio measure of global consumption poverty that takes account,
above an appropriate level, of the standard of living in the country in
question, thus combining fixed and relative elements of poverty

Training 12

12

4
 Absolute poverty lines
many popular methods but one key idea: food is the anchor

1) Direct Calorie Intake (DCI)

Kakwani (2003)

2) Food Energy Intake (FEI)

Dandekar & Rath (1971) + Greer & Thorbecke (1986)

3) Food-share
Orshansky (1963, 1965)
4) Cost of Basic Needs (CBN)
Rowntree (1901) + Ravallion (1994)

Training

13

The Cost of Basic Needs (CBN) method

§ In a nutshell: estimate the cost of a consumption bundle adequate to meet basic

consumption needs.
§ Question
What constitutes a ‘basic need’ and what does not?
§ Constraint
The choice of the basic-needs bundle should reflect local perceptions of what
constitutes poverty (specificity).
§ Solution
A safe start consists in including foodstuffs among the basic needs. After, we’ll
think of how to add an allowance for consumption of non-food goods/services.

Training 14

14

The CBN method: A strategy

§ Three steps:
1) Estimate the cost of a ‘basic food bundle’: this gives the food poverty
line
2) Estimate the allowance for ‘basic non-food goods’

3) Add 2) to 1): this gives the (total) poverty line

Training 15

15

5
 The food poverty line (FPL)

§ How to define a ‘basic food bundle’?

§ The key idea, which does not require any arbitrary assumption on consumption
patterns, is to:
1) estimate the minimum energy requirement for the average individual in the
target population (say 2,000 kcal/person/day)
2) price that amount of calories, using the average cost of one kcal which is
computed using the survey data.

§ A monetary amount is obtained, and that is the food poverty line (FPL)

§ Note that 3) takes account for local tastes (preferences)

Training 16

16

The non-food allowance (NFA)

§ How much is the minimum for non-food necessities?

§ We start by asking the data
§ Focus on a subset of people that are most likely poor, and see how
much they spend on non-food
§ Two ways to define that target population:
1) people whose total expenditure is about as much as the food
poverty line (lower bound)
2) People whose food expenditure is about as much as the food
poverty line (upper bound)
Training 17

17

Allowance for non-food goods

Food
Expenditure 45 #

𝑭
Expenditure = 𝒛

NFA U
Food
𝒛𝑭
NFA L
Food expenditure = 𝒛 𝑭

Expenditure per capita

Lower Upper
Total spending of households having Total spending of households having food
total expenditure = food poverty line expenditure = food poverty line

Training

18

6
 Lower and Upper Bound CBN Poverty Lines
Recap

/0/1002
§ 𝐿𝐵𝑃𝐿 = 𝐹𝑃𝐿 + 𝐸- 𝑥- |𝑥- ≈ 𝐹𝑃𝐿 (lower bound PL)

/0/1002 1002
§ 𝑈𝐵𝑃𝐿 = 𝐹𝑃𝐿 + 𝐸- 𝑥- 6𝑥- ≈ 𝐹𝑃𝐿 (upper bound PL)

§ Which one to choose?

§ It is customary to report results on them all (FPL, LBPL, UBPL), but if
there needs to be one number, it is often based on UBPL

Training

21

Important remark

§ The CBN method hinges on the food poverty line

§ A good food poverty line requires good estimates of calorie intake

§ Good estimates of calorie intake require a well designed questionnaire

(lectures 5-7)

Training 22

22

Zambia, 2015
Living Conditions Monitoring Survey

Food
Poverty
Line
Training 23

23

7
 Zambia, 2015
Living Conditions Monitoring Survey

The non-food allowance was

determined as the average non- 𝐿𝐵𝑃𝐿
food consumption of households 𝐹𝑃𝐿
whose total consumption was Non-food
close to the food poverty line: Allowance
/0/1002
𝐿𝐵𝑃𝐿 = 𝐹𝑃𝐿 + 𝐸- 𝑥 - |𝑥 - ≈ 𝐹𝑃𝐿

Training 24

24

Poverty measures

Training 25

25

Poverty measures
Basic ideas

§ Poverty measures aggregate information.

§ A poverty measure is a function of individual incomes x = (x 1, …, x n) and the poverty line z:

P : Rn ® R+
§ The literature on poverty measures is huge and technical in nature. It deals with the choice
of the functional form of a suitable poverty index.

§ In practice, three indices have taken center stage:

1) the headcount ratio
2) the poverty gap index
3) the poverty gap squared index

Training 26

26

8
 The poverty headcount ratio (H)
Mongolia HSES 2016, Cumulative distribution of per capita consumption (p.10)

§ The headcount ratio is the proportion

of the population that is classified as
Cumulative poor.
Density Function (CDF)
8 :
§ 𝐻 = 9 = 9 ∑9
-<: 𝐼 𝑥- ≤ 𝑧

§ I(·) is an indicator function that is 1 if

its argument is true, 0 otherwise.

§ Interpretation: incidence of poverty

Training

27

The headcount ratio

Discussion

§ Easy to understand
§ Insensitive to:
1) the degree of poverty:
cut in half every poor’s income …H does not change!
2) the distribution of income among the poor:
transfer from a poor person to a not-so-poor person (still poor after the
transfer) … H does not change!
transfer from a very poor person to an ‘almost-not-poor’ person (not poor
after the transfer) ... H decreases!

Training 28

28

The headcount ratio

In terms of policy

§ A transfer to a very poor household would probably leave the

headcount index unchanged (if poor remains below the line) even
though poverty has overall lessened.

§ The easiest way to reduce the headcount index is to target benefits to

people just below the poverty line. Policies based on the headcount
index might be sub-optimal (Lipton, Ravallion 1993: 24)
§ H only shows the effect of poverty-eliminating policies, not poverty-
alleviating policies.

Training 29

29

9
 The Poverty Gap (PG) index

§ The PG index is defined as the

Expenditure
Per capita

Quantile average poverty gap in the

Function
population as a proportion of the
poverty line (where the non-poor
Poverty Gap Index have zero gaps):
9 8
1 𝑧 − 𝑥D 1 𝑧 − 𝑥D
𝑃𝐺 = C 𝐼 𝑥D ≤ 𝑧 = C
𝑁 𝑧 𝑁 𝑧
z D<: D<:

§ The poverty gap index (PG)

accounts for the depth of poverty: it
0 .5 1
tells how poor the poor are.
Quantile
Training

30

The Poverty Gap index

Disassembling the PG index

§ Use simple algebra to rewrite PG as follows:

𝜇L
𝑃𝐺 = 𝐻×𝐼 where 𝐼 = 1 −
𝑧

§ The term I is the ‘income-gap ratio’, where µz is the average income among the
poor.
§ Neither H nor I are – individually taken – ‘good’ poverty indicators, but are useful
building blocks...
§ PG combines incidence of poverty (H) with depth (I).

Training

31

The Poverty Gap index

Interpretations

§ Suppose PG = 0.20
§ Interpretation 1
“On average, the poor have an expenditure shortfall of 20 percent of
the poverty line”
§ Now suppose z = $1,000 per month (poverty line).
§ Interpretation 2
The per capita cost of eliminating poverty is equal to PG × z. In our
example: $200 ( = 0.20 x 1,000) per month.

Training 32

32

10
 Why do we need to go beyond the PG index?

a b g d H PG PG2
A 1 2 3 9 0.75 0.375 0.219
B 2 2 2 9 0.75 0.375 0.188

poverty line = 4

PG is insensitive to distribution of income among the poor

Training
33

33

The Poverty Gap Squared

Definition

§ The squared poverty gap index attributes more weight to the poorest
among the poor:
N 2 q 2
1 æ x ö 1 æ xi ö
PG 2 =
N
å çè1 - z ÷ø I ( x
i =1
i
i £ z) =
N
å çè1 - z ÷ø
i =1

§ The contribution of the i-th individual to PG2 is larger the poorer she is, that is, the
larger is her poverty gap (z- xi)/z:

weight gap
q
1 æ xi ö æ xi ö
PG 2 =
N
å çè1 - z ÷ø ´ çè1 - z ÷ø
i =1
Training

34

A highly influential article

The headcount ratio, the PG and PG2 all

belong to the Foster-Greer-Thorbecke
(FGT) class of poverty measures.

Training 35

35

11
 FGT (1984)
Definition

The FGT class of poverty measures:

9
1 𝑧 − 𝑥- M
𝑃M = C 𝐼 𝑥- ≤ 𝑧 , 𝛼≥0
𝑁 𝑧
-<:
a Pa Index
0 P0 = H HEADCOUNT RATIO
1 P1 = PG POVERTY GAP INDEX
2 P2 = PG2 POVERTY GAP SQUARED
…
¥ P¥ weights the poorest person

Training

36

Lessons learned

1) We argued in favour of objective, absolute, CBN poverty lines.

2) Regarding poverty measures:

§ The headcount ratio is a crude and ‘theoretically inferior’ poverty index. H is useful,
but should not be used exclusively.

§ The Poverty Gap Index and the Squared Poverty Gap Index are complements to H;
poverty analysis should combine the three measures. We recommend FGT (1984).

§ The axiomatic approach does not succeed in identifying the “best” poverty measure.
Yet, it is useful, as it reveals the principles underlying the poverty measures.

Training 37

37

References
Required readings Haughton and Khandker (2009). Handbook on Poverty
Ravallion M. (2016). The Economics of Poverty: and Inequality, Chapters 2-4.
History, Measurement, and Policy Oxford University Lipton, M., Ravallion, M. (1993). Poverty and policy. Policy,
Press. (Chapters 3, 4, 5.1-5.5). Research working papers ; no. WPS 1130. Poverty and
human resources. Washington, DC: World Bank.
Suggested readings
Ravallion M. (1994). Poverty Comparisons.
Atkinson T., WBG (2017). Monitoring Global Poverty, Report
of the Commission on Global Poverty Ravallion M. (2008) Poverty Lines. In: Durlauf S.N., Blume
L.E. (eds) The New Palgrave Dictionary of Economics.
Deaton, A. (1997). The Analysis of Household Surveys: A Palgrave Macmillan, London
Microeconomic Approach to Development Policy. Ravallion, M. and B. Bidani (1994). How Robust is a Poverty
Washington, D.C.: World Bank. Sections 3.1 and 3.2
Profile?, World Bank Economic Review, 8: 75-102.
Deaton, A. and S. Zaidi (2002). Guidelines for Constructing
Sen, A. (1976). Poverty: An Ordinal Approach to
Consumption Aggregates for Welfare Analysis. Living
Standards Measurement Study Working Paper N. 135. The Measurement”, Econometrica, 44(2): 219-31.
World Bank, Washington, DC. Zheng, B. (1997). Aggregate Poverty Measures, Journal of
Economic Surveys, 11(2): 123-62.
Foster, J., J. Greer, and E. Thorbecke (1984). A Class of
Decomposable Poverty Measures, Econometrica, 52, 3: 761–
65.
Training

38

12
 Thank you for your attention

Training 39

39

Homework

Training 40

40

Exercise 1 – Engaging with the literature

§ What does Zheng (1997) show

regarding the Watts index?

Training 41

41

13
 Exercise 2 – ADePT
http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EXTPROGRAMS/EXTADEPT/0,,menuPK:7108381~pagePK:64168176~piPK:64168140~theSitePK:7108360,00.html

§ Take any expenditure survey

dataset of you interest

§ Download ADePT Poverty

§ Generate selected poverty

measures through the software

Training 42

42

Exercise 3 – DASP
http://dasp.ecn.ulaval.ca

§ Take any expenditure survey

dataset of you interest

§ Install DASP for Stata

§ Generate selected poverty

measures through the package

Training 43

43

14