+ All Categories
Home > Documents > Introduction Induced Demand - University of Notre Damewevans1/health_econ/induced_demand.pdf ·...

Introduction Induced Demand - University of Notre Damewevans1/health_econ/induced_demand.pdf ·...

Date post: 28-Mar-2019
Category:
Upload: ngodat
View: 214 times
Download: 0 times
Share this document with a friend
11
1 1 Induced Demand Lather. Rinse. Repeat 2 Introduction Two key concepts we’ve be stressing this semester collide in this section – The role of incentives – Asymmetric information Subject know as “induced demand” Lively area of research Lots of suggestive but few definitive results 3 • Idea: when providers are paid on a per unit basis, they have an incentive to order more procedures, whether needed or not • Reason: patients are in poor position to understand whether a procedure is needed, so they do not have the ability to monitor 4 Notion took root when people observed that areas with greater hospital beds had more higher hospitalization rates Why would many dismiss these results immediately as simply correlation and not causation? Tests have become more sophisticated over time
Transcript

1

1

Induced Demand

Lather. Rinse. Repeat

2

Introduction

• Two key concepts we’ve be stressing this semester collide in this section– The role of incentives

– Asymmetric information

• Subject know as “induced demand”

• Lively area of research

• Lots of suggestive but few definitive results

3

• Idea: when providers are paid on a per unit basis, they have an incentive to order more procedures, whether needed or not

• Reason: patients are in poor position to understand whether a procedure is needed, so they do not have the ability to monitor

4

• Notion took root when people observed that areas with greater hospital beds had more higher hospitalization rates

• Why would many dismiss these results immediately as simply correlation and not causation?

• Tests have become more sophisticated over time

2

5 6

7 8

Plot: Hospital Beds vs. Hospital Discharges, Health Reportng Region

100

150

200

250

300

350

1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5

Hospital Beds/1000 Residents

Me

dic

are

Ho

sp

ital

Dis

ch

arg

es/1

00

0 E

nro

lle

es

3

9 10

6

8

10

12

14

16

18

20

2500 4500 6500 8500

30 D

ay R

ead

mis

sio

n r

ate

Part A Spending Per Capita

Plot: Medicare Spending (Part A) Per Capita vs 30 Day Re-admission rates (2008)

11 12

4

Example: invasive heart attack treatments

13 14

15 16

5

17 18

19 20

Hillman et al, 1990 (NEJM)

• Doctors can send patients to diagnostic clinics

• Sometimes, the clinics are owned by physicians, sometimes they are not

• Compared rates that physicians referred patients to imaging centers based on certain conditions

6

21

Compared two groups

• Physicians that owned their own imaging centers

• Those that did not (and had to refer patients to radiologists)

• Find much higher referral rates for physicians that had a stake in the business

• Possible explanations?

22

Fraction of Visits With a Diagnostic Image

0%

10%

20%

30%

40%

50%

60%

70%

Low back pain Difficultyurinating

Upper Resp.symptions

Pregnancy

Per

cen

t

Self-referral Radiologist referal

23

Some evidence against

• Dranove et al., looked at the induced demand for childbirth

• Compared frequency of childbirth with concentration of OB/GYNs

• Found that OG/GYNs ‘induced’ childbirth

• Tongue in cheek paper, but it drives the point home

Simple model

• Physician utility– U(Income, Inducement) = U(Y,I)

• baseline demand for service=Q0

• Price of service = m

• Income without inducement is Y0=mQ0

• Muy>0 but inducment is a “bad” MUI<0– MUii=d2U/dI2>0

– Increasing disutility with inducement

24

7

MRS

• U(Y,I)

• Hold utility constant

• Uydy + Uidi = 0

• dy/di = (slope of indifference curve)

= -Ui/Uy= [dU/di]/[dU/dY]=dY/di

• How much y do you need to induce one more unit of induced demand

• Since Uii>0, need increasing compensation25 26

Income

Induced demand

I1 I2

U1

U2

Y1

Y2

Budget Constraint

• Q0 = baseline level of demand

• m = price per procedure

• Y intercept = mQ0

• Y = mQ0+mI

for ever unit you induce, receive M

slope of BC = m = dY/dI

27 28

Income

Induced demand

mQ0

mQ0+mI

m*Q0+m*I

mQa

mQa+mI

Case 1: increase mto m*

Case 2: reduce Q0

to Qa

m*Q0

8

29

Income

Induced demand

Q1

U1

Y1

mQ0

30

Income

Induced demand

I1

U1

Y1

mQ0

mQ*

I2

Baseline demandfalls to Q*

U2

31

Income

Induced demand

I1

U1

m*Q0

I2 I3

Price increases to m*

Substitution effectI1→I3

Income effectI3→I2

mQ0

U2

Summary

• Price increase for m– Increase incentive to induce demand

– Because induced demand is an inferior good, income effects suggests less inducement

– Income and substitution effects going in opposite directions

– Do definitive prediction

• Drop in baseline demand– Generates an income effect that encourages induced

demand 32

9

33

Gruber-Owings

• Excellent example of empirical analysis of SID

• Different type of identification strategy

• Most papers rely on either– Cross area variation in doctors

– Fee schedules that may induce demand

34

• Exploits the fact that– OB/GYNs paid more if deliver by c-section

– Between 1970-82, fertility declined 13.5%

– C-sections were at 5.5% in 1970, rose 240% over the next decade

• Question: did doc’s respond to the income ‘shock’ of reduced births by performing more c-sections

35

• Some key facts– Tremendous variation across areas in fertility rates.

Use this fact in model

– During this time period, physicians made $500 more delivering by c-section (1989$)

• C-section, $2100

• Vaginal, $1600

36

10

37

21% reduction in Fertility from 1970-1976

38

39 40

11

41

Results in Gruber

• Fall in fertility can explain 16% of the rise in c-sections over the 1970-1982 period


Recommended