Association for Public Policy Analysis & Management

Psychiatric care provision has been of recent interest to policymakers. While advocates for psychiatric bed expansion have long speculated that an increase in psychiatric bed capacity may decrease the number of mentally ill incarcerated and decrease the psychiatric care burden for medical hospitals, estimating the magnitude of this effect is challenging. For example, technological and practice changes in psychiatry may drive provider decisions regarding long-run bed supply; thus bed numbers cannot be treated as exogenous.

In this paper, I address these problems by exploiting a feature of hospital cost structures to identify discrete discontinuities in hospital-referral region psychiatric bed supply using data from the American Hospital Association Survey (1972-2014) and Medicare Provider of Services (1991-2014). Subsequently, I examine outcomes using the National Inpatient Sample (1988-2014), Truven MarketScan (2003-2014), and the Annual Survey of Jails (1985-2014).

I find an asymmetric effect of psychiatric bed reductions and additions. While a local area reduction in psychiatric bed supply by 1 bed per 100,000 population corresponds to an observed increase of .02 standard deviations in the number of psychiatric admissions at nearby hospitals, local area additions to psychiatric bed supply do not substantially impact short-term psychiatric admissions rates at nearby hospitals. In particular, patients with more severe mental illness were affected by bed reductions: a local area bed reduction of 1 bed per 100,000 was associated with an average 200 excess psychosis admissions at nearby hospitals.

Local area reductions in psychiatric bed supply were associated with statistically insignificant increase in the admission rates of local area jail inmates.