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