Week 8: Health

Question 1

Health is a vital component in human capital and as an economic asset itself.

So 3/8 Millenium Development goals were based on health. What were they

Answer 1

Child mortality - reduce by 66%
Maternal health - reduce mortality by 75%
Halt/prevent spread of STIs

Question 2

Did they meet the Millenium Development goals?

Answer 2

Child mortality reduced by 52%
Maternal health - mortality reduced by 50%
Prevent spread of STIS - failed

Question 3

3 Culprits of the failed health goals

Answer 3

Geography e.g extreme temperatures
Poor health care e.g lack of medicines, hospitals etc
Poor private health behaviour

Question 4

What is meant by poor private health behaviour, use example

Answer 4

2/3 of under-5 deaths could’ve been averted if parents used simple cheap technologies e.g bednets

Question 5

3 difficulties in measuring health

Answer 5

Multi-dimensional - no single measure

Proper clinical evaluations are too expensive

Measurement errors (non random/systematic)

Question 6

Pros and cons of using self-reported health measures e.g questionnaires - “How would you judge your overall health”

Answer 6

Pro
Easy

Con
Not comparable between individuals
Negative correlation with access to care (Systematic measurement error !)

Question 7

What is the systematic measurement error

Answer 7

Richer peoples generally report being sicker.

(rich can afford access to care so more aware of health issues!)

Question 8

3 Problems of using measure of illnesses or “normal activities” missed as a self reported measures

“Have you had malaria in last 12 months”
“did you miss any work days due to illness”

Answer 8

Different opportunity costs e.g farmers less willing to miss days despite illness since harvests are seasonal so will only skip if extremely sick!!!

Need to actually know they’re ill

Strategic misreporting - may report ill more if think programs will happen that benefit them

Question 9

So don’t ask “how would you judge overall health” , or use “normal activities missed” or measure illnesses,

What about self reported physical functioning e.g “how far can you walk comfortably”, or “do you have difficulty walking”

Pro and con

Answer 9

Measurement errors less likely (esp systematic)- no technical knowledge required e.g know illness/symptoms

Difficulty is subjective

Question 10

So what is the best to use out of all the (self reported) survey reports?

Answer 10

Self reported physical functioning

Question 11

We can also use nutritional intake.

How is it often measured, and flaws (2)

Answer 11

Food expenditure is imperfectly correlated with calories

Food bought ≉ food eaten - can go to guests. (this would overestimate their nutritional intake since it doesn’t all go to them)

Question 12

What assumptions do we have to make using food expenditure as a measure of nutritional intake approach

Answer 12

No wasted food.

Question 13

So we could directly measure calorie intake instead to avoid the flaws of using food expenditure. E.g “what did you eat this week”

But what are issues with this (2)

Answer 13

Expensive to measure

Noisy recall - can remember consumption over last 24hrs at best (so inaccurate results perhaps)

Question 14

With noisy recall, how do non-random measurement errors exist?

Answer 14

poor eat the same so can remember, richer have more diversity so might not remember as well. OR better-fed (richer) have better cognitive skills so can remember more.

Question 15

We can also use anthropometrics e.g height weight BMI

PROS (2)

Answer 15

Cheap info to collect

ME would only be random (uncorrelated with income) - e.g no systematic errors, since just measuring e.g height, weight etc.

Question 16

Random measurement errors in Y and X vs non-random/systematic measurement erros

Example of non-random/systematic vs random ME’S
Non random - poor eat the same food, so more likely to remember, while rich have more diversity so more likely to forget, so measurements are inaccurate.

Random - using anthroprometrics - e.g just weighing someone incorrectly

Answer 16

Random ME’s in Y: no bias, just larger standard errors

Random ME’s in X: health coefficents are biased towards 0 (Attenuation bias!)

If systematic ME’s: we get spurious correlation (not causal but correlation) e.g we see as income increases, measured health falls, (rich notice illnesses, poor report heathier)

Question 17

How can we reduce effect of random ME in X:

Answer 17

Repeated measures

Question 18

3 recent developments in measurement of health overtime? (3)

Answer 18

Development of cheap clinical tests e.g blood pressure

Direct measure of functional limitations e.g ask them to lift something (an upgrade to self-reported functioning!)

Randomly allocate health inputs e.g bednets, see improved effect on other outcomes e.g attendance, which indirectly shows poor health reduced attendance and health has improved!!)