Bayesian Geweke Summary - BAYESIAN INFERENCE FOR HOSPITAL...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon
B AYESIAN I NFERENCE FOR H OSPITAL Q UALITY IN A S ELECTION M ODEL Paper by J. Geweke, G. Gowrisankaran and R. Town: Econometrica , 2003 – Vol. 71(4), pp. 1215-38 Hendrik Wolff April, 2006 Example for Bayesian Analysis in nonlinear IV context ARE213 Spring Semester 2006 UC – Berkeley
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
1 Considering the high costs of the health system in modern society, the efficiency of the health system should be of major interest to taxpayers, insurance companies and policy makers. As an illustrative example, Geweke et al. state that if changes in health-policy cause hospitals to reduce their pneumonia mortality rates by one percentage point, this would translate to over 6000 lives saved annually in the US. The principal empirical questions are: i) Quality - size and ownership status of hospitals ii) Quality Rankings of individual hospitals In order to address these questions the authors concentrate on one the mentioned illness pneumonia: This illness seems appropriate a) it is common: the dataset of the study uses discharge information of 75,000 patients in the Los Angeles County of 114 hospitals. b) about 10% of these patients die c) the mortality rates of the hospitals are commonly seen as an excellent indicator of hospital quality in the literature: it is unambiguously defined (compared to disease specific variables) and its link with quality is almost “tautological”. This suggests the following relation:
Background image of page 2
2 M ORTALITY E QUATION (1) m i * = c i β + x i γ + ε i i = 1. . 75,000 Mortality indicator m i : m i = 1 if m i * > 0 m i = 0 if m i * 0 Assignment indicator c i : c i is a J × 1 vector: c ij = 1 if patient i is admitted to hospital j , c ij = 0 otherwise k Covariates: x i age, race, sex, disease stage (observable!), income (cip-code average income) Noise: iid ε i ~ N(0, σ ²) Interpretation: If patient i were randomly assigned to hospital j , Prob( m i = 1) = Φ (( c i + x i γ )/ σ )
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
3 However, random assignment assumption not plausible: Even though we control for the disease stage in x, there is an unobserved severity of illness (USI) which might significantly influence the decision of the patient (or the patient’s agent) which of the J hospitals to choose. Because high-quality hospitals attract sicker patients, those hospitals may not look very “good” after estimating their adjusted mortality rates.
Background image of page 4
Image of page 5
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 08/01/2008 for the course ARE 213 taught by Professor Imbens during the Spring '06 term at University of California, Berkeley.

Page1 / 16

Bayesian Geweke Summary - BAYESIAN INFERENCE FOR HOSPITAL...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online