Health Economics BHT CH 1-5

The flatter/more horizontal the curve, the price is more

**Elastic** & sensitive

In a **non-randomized** setting, the study will

**Overestimate** healthcare demand at low prices and **underestimate** demand at high prices

The increase in healthcare is due to

the new scientific discoveries and technological improvements

Healthcare market differs from other markets for one major reason

uncertainty

Non-random selections distort estimated demand curve in what way?

the groups facing each price level differ in important ways

What results in no **selection bias**?

A large sample size + randomization

Marginal efficiency of capital curve shows how efficient each unity of health capital is in increasing

lifetime utility

Physicians may not be perfect agents of healthcare for patients due to all of the following **EXCEPT**

Medical arms race

What is selection bias from comparing observed outcomes of the treated and non-treated groups?

Difference in group means = selection bias + TOT

Randomized Experiement

study that assigns treatments randomly to different groups of study participants

\
Includes treatment group and control group (also randomly chosen and receives no treatment)

What are the potential **problems** associated with this **non-randomized** approach?

not generalizable nor representative of the area

\
Selection bias

RAND Health Insurance Experiment (HIE)

First large-scale experiment with random assignment of insurance status

\
After publishing, the demand curve for health care is not vertical but downward-sloping

Copayment/Coinsurance rate

fraction of the medical bill which the **patient** is responsible for

Oregon Medicaid Experiment (ME)

Involved two groups of **low-income** adults in Oregon

Cost-sharing plan

one with positive copayment rate, so that costs are shared between the insured and the insurer

Claim: both the RAND Health Insurance Experiment and the Oregon Medicaid Experiment found compelling evidence for the effect of the price of healthcare on mortality

False

\
found much impact on utilization

not much compelling evidence on mortality/morbidity

\
negative correlation between price and demand

Outpatient Care

any interaction with a doctor or other medical care professional that **does not involve** an **overnight stay**

Inpatient Care

any interaction with a doctor or other medical professional that **involves** an **overnight stay** at a hospital

Emergency Room (ER) Care

any interaction with a doctor or other medical care professional that **involves the emergency room**

RAND HIE and Oregon Medicaid Experiment gathered use and spending data for

mental health care, dental care, and prescription drug use

Claim: in the Grossman model, the marginal efficiency of investment in healthcare increases as health improves

False

\
once health is optimal, there is little improvement if invested

Key assumption for differences-in-differences

in absence of treatment, the trends should move in parallel

When should you use **regression discontinuity**?

the graph has **only one** line with a discontinuous jump

\
There are “cutoff” rules that determine actions individuals may take and seem arbitrary

When should you use **differences-in-differences**?

the graph has **two** lines

Arc Elasticity of Demand

Treats price **increases** and price **decreases** *symmetrically*

Price-Elastic

|e| >= 1

Price-Inelastic

|e| < 1

Unit Elastic

|e| = 1

Perfectly Inelastic

|e| = 0

Claim: health is something that both happens to you and something you choose

True

Grossman Model

Health is a consumption good, input into production, and a form of capital

\
ties together the health decisions that people make on a daily basis

Single-Period Utility

In a given period, an individual’s utility depends on her health and other non-health goods

\
Ut = U(Ht, Zt)

What is Ht in the Grossman Model’s single-period utility

level of health

What is Zt in the Grossman Model’s single-period utility

composite good that includes everything else

\
**home good**/all other goods

How does healthcare affect utility through Ht?

Health level (Ht) **increases**

Utility **increases**

Time Constraint within a Single Period

Θ = Tw + TH + TZ + Ts

What time constraint is **wasted time**?

Ts (time spent being sick)

What time constraint **improves BOTH H and Z**?

Tw (time spent working)

\
earns money/income to improve

What time constraint **improves Z**?

TZ (time spent playing)

What time constraint **improves H**?

TH (time spent improving health)

Productive Time in Grossman Model

Tp = Total Time(Θ) - Ts

Average Treatment Effect (ATE)

avg(Y_treated) - avg(Y_untreated)

In the **Grossman model**, neither Ht nor Zt can be purchased directly in the market

Must be **combined** with personal **time** to produce inputs into **utility**

Difference between Ht and Zt

Zt is **assumed to be created and consumed each period**

\
Ht is a stock that accumulates/deteriorates over time

\
(Yesterday’s *health* affects today’s *health* ***BUT*** Yesterday’s *time* does NOT affect today’s *time*

What is Mt in the production function Ht = H(Ht-1, TH, Mt)?

market inputs like vaccines and treadmills into health H during period t

What is Jt in the production function Zt = Z(Tz, Jt)?

market inputs like video games and opera tickets into the home good Z

Market Budget Constraint

cannot spend more than you earn assuming there is no option to save

The relationship between Ht and Ts is

negative

The relationship between Ht and Tp is

positive

How is health a **consumption good** according to the Grossman Model?

It contributes directly to the individual’s utility function each period. Being healthy is valuable in and of itself.

How is health a **input into production** according to the Grossman Model?

It generates productive time Tp which is useful for producing more H and Z

How is health a **form of capital** according to the Grossman Model?

Unlike the home good, it endures from period to period.

It can accumulate (or depreciate) over time, so improvements in health today can lead to better health tomorrow.

Production Possibility Frontier (PPF)

Graphical model that illustrates central economic concepts and the core issues of production and allocative efficiency

What does the PPF do?

maximizes output combinations of Ht and Zt, given an individual’s budget and time constraints

How is the standard PPF different from the Grossman PPF?

P would not be feasible for the Grossman Model because the person would be dead (no health = dead)

Where is the Optimal Ht and Zt located on the Grossman PPF?

Between C and E

need to introduce preferences or indifference curves to give optimal point

What is the key prediction of the Grossman Model?

optimal level of health is **less** than the maximum

\
individuals makes **trade offs** leading to less than maximal health

What can be said about this individual? (Grossman PPF w/ vertical indifference utility)

Individual only cares about **health**

\
Indifference curves are **vertical** because home good does **NOT** have effects on utility

What can be said about this individual? (Grossman PPF w/ horizontal indifference utility)

Individual only cares about **home good**

\
Indifference curves are **horizontal** because health does **NOT** have effects on utility

Marginal Efficiency of Capital (MEC) Curve

shows how efficient each unit of health capital is in **increasing** lifetime utility

Costs of Health investment

alternative market investments that pay an interest rate *r* and depreciation due to aging γ

Why must health pay a return of at least r + γ?

Other options will be more attractive if health pays a return of less than r + γ

When health H is low

the rate of return is higher than r + γ

increased γ

more depreciation as time goes on

\
increases with age

Costs of health increase with age

 individuals are less willing to spend productive time and money on health

Empirical Health Phenomena

Grossman model provides economic explanations for the phenomena of

\
better health among the educated & declining health among the aging

Claim: Grossman model is a very useful tool for understanding the roles and tradeoffs of health

True

The more-educated enjoy better health than the less-educated. Why?

the more-educated are more efficient in producing health

\
Better educated individuals are just better at producing health

Claim: We can use Grossman model to understand empirical findings about the relationship between socioeconomic status and health

True

Claim: the relationship between health, socioeconomic status (SES), and other variables is **causal**

True

Claim: lower SES leads to higher health

False

How is health disparities observed?

by education, income, race, and social standing

Claim: College graduates have **higher** survival rates or life expectancy compared to high school dropouts

true

Claim: there is strong evidence of health disparities in the US (largely private healthcare system)

true

Claim: A universal healthcare system may help reduce health differences between rich and poor. There are still health disparities, just not as much as America

true

Causes of health disparities

* Adherence to medical advice
* Early life events
* Income levels
* Stress of being poor
* Work capacity
* Impatience

Efficient Producer Hypothesis

health disparities exist because better-educated individuals are more efficient producers of health than less well-educated individuals

\
Education makes better producers of health

Thrifty Phenotype Hypothesis

resources deprivation in utero and during early childhood can lead to activation of “thrifty” genes optimized for sparse conditions

\
**There may be genetic reasons why individuals differ in health production efficiency**

Natural Experiment

a study that uses an environmental shock that creates a treatment group and a control group naturally

\
ex: famines, earthquakes, snowfalls

Direct Income Hypothesis

looks at absolute level of income; can buy many things to improve health

\
health disparities arise because the rich have more resources available to invest in health

\
**More financial resources can result in higher optimal levels of health**

Allostatic Load Hypothesis

health disparities exist due to cumulative physiologic toll exacted on the body over time by efforts to adapt to life experiences

\
This prolonged and repeated stress, also known as allostatic load, increases the rate of depreciation of health capital ; stress makes us age faster

\
**Higher depreciation rates of health capital mean that investing in health becomes more expensive**

Income Inequality Hypothesis

doesn't care about absolute level of income but the relative level of income or distribution of income; if distribution of income is unequal in society, you will be stressed out bc you'll compare yourself -> poor health -> age faster

\
health disparities are caused by income inequality, which in and of itself is a source of allostatic load for the poor

\
**More equal societies are less stressful and therefore have better health**

Productive Time Hypothesis

SES differences are caused by disparities in health

\
**Individuals with poor health have low productive time to work and produce income**

Fuchs Hypothesis

Individuals with a lower rate of time discounting (who are therefore more patient) invest more in both health and education

\
**Observed links between health and SES may be caused by an unobserved third factor: patience**

What does standard errors tell us?

the amount that the estimates differ from the true values

What does a 95% confidence interval mean?

there is a 95% chance that the true values of the parameters fall into these confidence intervals

What are the typical p-value cutoffs for rejecting H0?

5% and 1%

*Ceteris paribus*

Latin phrase that means "all other things held constant"

\
every variable is constant except for variable that you are testing

Is ceteris paribus possible for individuals or groups?

Groups of people (focusing on characteristics of the average)

TOT > ATE

Treatment **effects** are **higher** among **treated individuals**

TOT < ATE

Treatment effects are lower among treated individuals

selection bias

Y_treated,before - Y_untreated,before

Average Effect of Treatment on the Treated (TOT)

avg (treatment effect & treated)

Difference in group means

avg (observed & treated) - avg (observed & untreated)

Claim: the TOT and ATE are **not equal**, this would supply a selection bias

false

What is the banking system built on?

trust

Claim: Banking system works as long as everyone withdraws their money at the same time

false

Which is better for financial stability?

\
* Availability of cheap loans may help good banks to survive
* Availability of cheap loans may encourage bad banks to survive

Availability of cheap loans may help good banks to survive

What would violate the key assumption of DD?

in the absence of treatment, the trends are not parallel to one another

i.e. economic growth was different and more rapid in one part than another

How would you test the assumption made in a DD?

look at data and see if they were moving in parallel. If not, stop, assumption is not satisfied, and cannot use DDestimate. If yes, assumption is okay but need to make sure that something else didn't happen during the years of pre&post period.