Measuring Welfare and Consumption

Question 1

Which of the following are used in surveys to measure welfare at the household level? (Select all that apply)

GDP
GNP
Consumption
Subjective well-being
Multi-dimensional approaches

Answer 1

Consumption
Subjective well-being
Multi-dimensional approaches

Macro-level indicators like GDP and GNP cannot be used to understand individual level welfare. Household survey data on consumption, subjective well-being and other multidimensional approaches are used to measure welfare at the individual or household level.

Question 2

Why does variance in consumption matter beyond just the level of consumption?

Answer 2

An average level of consumption with large variance can create anxiety, especially during low-consumption periods

Large variances in consumption are an indicator of vulnerability. A house that has a low average level of consumption with a lower variance of consumption is less vulnerable than a household with a higher average level of consumption but huge variability in consumption across different months or seasons

Question 3

For which consumption goods do we need to think about depreciation? (Select all that apply)

Food purchased in the market place
Home-produced foods
Food earned or received as a gift
Fuel, utilities
Transit/Transportation expenditure
Education
Durables
Housing

Answer 3

Both Durables and Housing are consumption items that are purchased once but ‘consumed’ over a period of time. So, when thinking about consumption at a given point in time, it is useful to incorporate the depreciation rate in the estimation.

Question 4

If we observe in the data that one household is consuming one thousand times more rice than another, it’s most likely a problem with:

Answer 4

The unit of measure they use for rice

As Dean mentions, the unit of measurement can also cause substantial measurement error when estimating consumption. Here, one needs to be mindful of two aspects – (1) In what unit is consumption of a particular item referred to in a local context and (2) How can this unit of consumption be converted into a more standard unit for comparison across studies.

Question 5

For consumption questions, the evidence suggests that if we use a longer recall period

Answer 5

We tend to underestimate consumption

Question 6

A subset module asks about a sample of food items. A collapsed module asks about categories of food items. What does the evidence from Beegle et al. (2011) say about these?

Answer 6

Subset modules (when reweighted) provide closer estimates to the total consumption module than collapsed modules

The Beegle et al. (2011) study in Tanzania, that Dean Karlan refers to, finds that the subset module when scaled based on reference data performs very close to long module surveys. The paper also cites that subset modules perform better than collapsed modules tested in the study.

Question 7

Which of the following methods asks respondents to measure their well-being relative to some reference point(s)? (Select all that apply)

Economic Ladder Questions (ELQ)
Satisfaction With Life questions (SWL)
Minimum Income Question (MIQ) followed by  
    Consumption Adequacy Question (CAQ)
Vignettes

Answer 7

Economic Ladder Questions (ELQ)
Minimum Income Question (MIQ) followed by Consumption Adequacy Question (CAQ)
Vignettes

The SWL question measures satisfaction in general or satisfaction with respect to specific attributes of life as a standalone measure. All other measures listed above are a comparison. For instance, ELQ measures where a household stands on a relative ‘ladder’. MIQ and CAQ measures a household minimum income/consumption with reference to the average income/consumption in the respective neighborhood or village. A vignette provides examples of different household circumstances and choices and requires respondents to rank themselves relative to these households.

Question 8

What might cause someone’s answer to a subjective well-being question to change over a relatively short time period? (select all that apply)

A certain event may have become more salient over that period
The weather may have changed
It may be a different day of the week
When asked the second time, the question may have come at a different point in the survey

Answer 8

All of the above

Several context and environment related factors affect responses to questions. As described by Dean Karlan, there is some interesting evidence of how responses to subjective well-being questions can vary even within the same week, or how they may be affected by weather patterns, order of questions, recency bias or other contextual factors.

Question 9

When measuring vulnerability, consumption variance captures:

Answer 9

A combination of a household’s income shock and their coping mechanism

Measuring variability in consumption helps identify if a household has access to coping mechanisms or not and is thus a better measure of vulnerability.

Question 10

The progress out of poverty index (PPI): (select all that apply)

Generates an estimate of total consumption within +/- 2 percent
Estimates the probability of being above or below the poverty line
Has been standardized to ask the same set of questions in each country
Could be useful for targeting resources for the poor
Could help an organization decide which households to treat

Answer 10

Estimates the probability of being above or below the poverty line
Could be useful for targeting resources for the poor
Could help an organization decide which households to treat

The PPI uses a consumption-based definition of poverty to estimate the probability of being above or below the poverty line. The questions on each scorecard are derived from national household expenditure or consumption surveys and can be different across countries. As Dean suggests, PPI is a useful tool when long surveys are not feasible and can be used for targeting resources for the poor or helping an organization determine treatment.

Question 11

Consumption

Answer 11

the most classic method that we use in economics, in that we think that a lot of the things that are basically proxies for how well-off you are and what your well-being

captured by observing the things that you actually consume now and in the future.

Question 12

subjective well-being

Answer 12

In a sense, this is just asking people, how well off are you? How comfortable are you? How much stress do you have over economic decisions and economic outcomes in your life?

Subjective questions on how healthy you feel tend to actually be very correlated with lots of components, suggesting that when you don’t have time for a full-blown survey, a question about subjective well-being on health can actually be a very good measure.

Question 13

Level of consumption vs variance of consumption

vulnerability

Answer 13

Large variances in consumption are an indicator of vulnerability. A house that has a low average level of consumption with a lower variance of consumption is less vulnerable than a household with a higher average level of consumption but huge variability in consumption across different months or seasons

Question 14

Why would we prefer to ask questions about consumption vs. income?

Answer 14

Consumption is actually easier to measure more precisely, particularly for people who live in an informal economy. If someone is in a formal setting, like the United States, then income actually might be even easier to measure, because you know where the wage comes from, the single source. But in most developing country settings, income is actually harder to measure than consumption.

Consumption is capturing, in a sense, accumulated past income and expectations of future income - it’s a more all-encompassing measure of how well-off someone is and what their overall well-being is

Consumption is considered less sensitive than income - so people are more willing to share information about it

Question 15

Consumption vs. expenditures?

Answer 15

In an informal economy, people are usually consuming things that they grow themselves. So if you only ask about expenditures, you will underestimate their consumption because you’re ignoring what was produced in the household.

Another issue with expenditures has to do with timing. If you buy things in bulk & then consume them over a long period of time - you might overestimate their consumption

Question 16

4 components of consumption

Answer 16

Food items
Non-food items
Consumer durables
Housing

Question 17

Food items

Answer 17

• Food purchased in the market place
• Food that is home-produced
• Food items received as gifts or remittances (i.e. dowry, festivals)
• Food received from employers as in-kind payments (lunch when working)

Question 18

Non-food items

Answer 18

Frequently consumed - ex: domestic fuel and power, toaccoo and alcohol, transport, etc.

Not frequently consumed - education, health, clothing, footwear, etc.

Question 19

Consumer durables

Answer 19

A durable is something that you buy once, and you use it for lots of time.

Important to think about the depreciation rate of a durable.

You buy a TV. That TV cost $100. You’re going to use that TV for five years. Then you think about that TV as being consumed at a rate of $20 a year, and that’s the consumption value of that television.

Question 20

Housing

Answer 20

So housing, if you’re renting, can be very simple. It’s rent that you pay or the rent’s equivalent

In a lot of contexts in developing country, there is no rent, and there is no actual ownership of a home. You’re in some– you’re maybe squatting or things of this nature. But there’s still expenses that go with having a house. You might have to pay for some repairs, pay for a wall, pay for roofing, flooring. These are all housing expenditures that are fairly similar to the durable goods issue, in the sense that you spend them once. And you use it over time.

Question 21

4 major sections of measurement error in consumption

Answer 21

• Method of data capture
• Reference period
• Level of detail
• Units of measurement

Question 22

recall survey

Answer 22

in which people just ask them, how much did you spend in the past week, or month, or some period of time on something?

Question 23

consumption diary & challenges

Answer 23

where you ask people to keep a record–to give them a little log book.

And the hope, of course, is that having people write down everything doesn’t change their actual behavior.

There are challenges to maintaining diaries. People can get exhausted. They might not have the literacy to appropriately record everything, so there might be mistakes from that. So having the surveyor there, guiding them through, answering questions, might actually help elicit more accurate questions. So there are tradeoffs here.

Question 24

recall period

Answer 24

do you ask about a week, a month, a past year?

Question 25

reference period

Answer 25

do you ask about the last week, or do you ask about a typical week, and a good week, or a bad week?

Question 26

Tradeoff with a very short recall period?

Answer 26

So when you have a very short recall period, if it’s something that people only do infrequently, you have a decent chance of getting a zero. And that might be less than ideal, if you want to be able to say something more concrete about what people are typically spending on something, when they do spend on it.

Also, risk of telescoping bias

Question 27

Telescoping bias

Answer 27

basically a bias where you’re not really sure about the timing of it

even though it happened a month ago, and you’re just asked about what did you spend in the past two weeks, you go back and you remember that thing from a month ago, and you report it.

What does this mean? It means you’re going to overreport how much you spent on these goods, because you’re taking things that are further from the back, and you’re bringing them forward, and reporting it as more recently.

Question 28

Tradeoff with a recall period that is longer?

Answer 28

Well, the first is you just might forget. Very long recall, you’re taking an educated guess. But you’re just going to add measurement error to that. But then the measurement error might not just be random. It might be biased.

if it’s too long of a period, the evidence suggests, for various psychological reasons and psychological biases, that we tend to underestimate the total consumption.

Question 29

Issues with longer modules

Answer 29

• Traditionally, detailed ‘long’ modules preferred
• Issues
– Respondent fatigue
– Rounding errors
– Increase survey length -> Takes much longer to execute

Question 30

Short modules: ‘Subset’ vs. ‘Collapsed’

Answer 30

do you take a subset of questions or do you try to collapse broad categories into fewer questions?

(Name 3 meats you eat (subset) vs. total meat consumption (collapsed))

Question 31

Food Items

List the suggested: (1) Mode of data capture, (2) reference period/recall, (3) level of detail

Answer 31

Mode of data capture: Diary

Reference period/recall: 1 week - 1 month

Level of detail: Subset module

Question 32

Non-food Items

List the suggested: (1) Mode of data capture, (2) reference period/recall, (3) level of detail

Answer 32

Mode of data capture: Diary/Recall

Reference period/recall: 1-2 weeks, a month (frequent items); 6-12 months (infrequent items)

Level of detail: Subset module

Question 33

Consumer durables

List the suggested: (1) Mode of data capture, (2) reference period/recall, (3) level of detail

Answer 33

Mode of data capture: Recall

Reference period/recall: Can be annual

Level of detail: Long or subset modules

Question 34

Housing

List the suggested: (1) Mode of data capture, (2) reference period/recall, (3) level of detail

Answer 34

Mode of data capture: Recall

Reference period/recall: Monthly actual or depreciation

Level of detail: Long modules

Question 35

Pros to measuring subjective well-being?

Answer 35

• Aggregate way of assessing overall well-being
– Understanding what matters to an individual when it comes to well-being

• Mental health issues
– Not consumption but how people ‘feel’ about it

Question 36

Economic Ladder Question (ELQ)

Answer 36

Imagine six steps, where on the bottom, the first step, stand the poorest people, and on the highest step, the sixth, stand the rich. On which step are you today?

One thing to note about this question is by the way it’s phrased is this is not asking how well you are doing, it’s asking how well you are doing compared to others in your community.

Question 37

Satisfaction with life (SWL)

Answer 37

This is a question which is usually asked with a very open-ended, vague set of options to answer. How satisfied in general are you with life?

What’s useful here is comparing across people and thinking about things in standard deviation shifts - if their satisfaction increased after the program, ex.

Question 38

minimum income question and a consumption adequacy question

Answer 38

“What is the minimum consumption you need in your community in order to lead a stable satisfying life?”

And then the second question, consumption adequacy, asks them, OK, “Are you consuming more, the same, or less than that minimum number that you just gave me a moment ago?”

Question 39

vignette method

Answer 39

So a vignette method, basically, gives a long description of an individual. And the opportunities they have that they’re able to seize and the challenges that they face and the shocks that they’re vulnerable to. And gives a description of their level of consumption.

How much do they eat? How much do they spend on various goods? And then asks people, “Which household is closer to you?” And that vignette approach helps, basically, put people into bins of poorest to less poor, and et cetera.

Question 40

Recency bias

Answer 40

what they remember might be the most recent thing that happened to them - so they are answering based on how they feel right then or most recently

Question 41

Challenge of subjective well-being questions

Answer 41

Biggest issue is with how questions vary over time, even a period of a couple weeks - people’s perception of how well they are doing fluctuate too much

Environment and context - i.e. better weather might change answers; getting interviewed on a Friday might put you in a better mood

Other context effects - question order, mode of data collection, etc.

Question 42

Vulnerability

Answer 42

This can be a tough concept just to define, nonetheless measure. So in a paper by Chambers in 1989, he defines it as “exposure to contingencies and stress, which is defenselessness, meaning a lack of means to cope without damaging loss.”

In a paper by Dercon, 2006, defines vulnerability as “the exposure to uninsured risk leading to a socially unacceptable level of well-being.”

Question 43

Two approaches to measuring vulnerability

Answer 43

Shocks and coping mechanisms

& Consumption variance

Question 44

Types of questions asked to understand shocks and coping mechanisms

Answer 44

• Experienced shock
• Timing of the shock
• Costs (type of loss)
• Costs (currency)
• Did others experience it?
• Severity
• Individual in the household who was affected
• Has the household recovered?
• Coping type

Question 45

Consumption Variance

Answer 45

• Looking past the shock and studying the underlying
‘consumption’
• Smooth consumption vs. Varying consumption

if a household’s consumption moves up and down a lot in a community, that’s an indication that they don’t have sufficient coping mechanisms to have smooth consumption throughout the year.

Question 46

Progress out of Poverty Index

Answer 46

The PPI is a statistically rigorous, easy to administer poverty measurement tool. It was designed with practitioners in mind, with a focus on simplicity and low cost.

It can also be used in a context for a researcher who doesn’t have the time for a big long module but does want a quick and dirty assessment with 10 questions that gets you at an ability to roughly understand where they’re going to fall in the rank of poverty within a community.

Question 47

So how does the PPI measure welfare?

Answer 47

The PPI uses a consumption-based definition of poverty. The questions on each score card are derived from national household expenditures or consumption surveys.

The PPI uses a subset, just 10 questions that are strongly correlated to results from the country’s national expenditure survey.

So a lot of econometric work goes into figuring out what are those 10 questions that best predict the aggregate number.

In general, the group poverty rate estimates are accurate within about plus or minus 2% at the 90% confidence interval.

Question 48

What is the PPI predicting?

Answer 48

The PPI is not actually predicting your total income or consumption, it’s just predicting whether you’re above or below the poverty line.

So it’s not a good tool, for instance, for identifying whether someone’s rich versus middle class. It’s just good for identifying and predicting whether they are above or below the poverty line.

Question 49

How can we use PPI data?

Answer 49

Can be used when long surveys are not feasible
Can be used for targeting participants for a program
Sometimes used to track HH over time (but PPI may not be the best tool here!)

Question 50

Consumption vs. income vs. expenditures

Answer 50

Consumption captures accumulated past income and expectations of future income.

Income can be used to measure welfare in developed countries such as the US.

Consumption expenditure does not include food produced and consumed at home.

Question 51

Consumption diaries can be used to measure which components of consumption?

Answer 51

Food items

Frequently consumed non-food items

Question 52

Broadly speaking, what are some of the reasons why subjective well-being is measured?

Answer 52

Subjective well-being measures how an individual factors in several environmental, social and cultural contexts in estimating his/her well-being. It measures not the level of consumption per se, but what an individual feels about how much s/he consumes. Most subjective well-being measures are relative i.e. an individual to provides his/her responses based on reference point(s).

Question 53

Which is false regarding the Progress out of Poverty Index?

Answer 53

PPI is NOT a useful tool to measure change in poverty and consumption over time

PPI is a ‘quick and dirty’ measure to assess poverty levels at a given point in time and though it is sometimes used to measure changes over time, PPI is not the best tool to do so.