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
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