Dynamics of the Mental Health Workforce: The Substitution of Physicians and Other Providers

Abstract

Objective: To employ production technology methods to evaluate how changes to mental health workforce levels, composition, and degree of labor substitution, may impact a typical mental health practice. Data Sources: FY 2007 Department of Veterans Affairs (VA) detailed data, assembled from multiple sources, on 134 mental health practices. Study Design: A generalized Leontief production function was employed to estimate the q-complementarity and q-substitutability of different mental health factors of production. We employ a formulation similar to Thurston and Libby (2002) who analyze a healthcare production function proposed by Diewert (1971). We do not impose parameter non-negativity in estimation. However, we test the negative semi-definiteness of the resulting production function Hessian. The marginal patient care output contribution was estimated for each labor type as well as the degree to which physicians and other mental health workers may be Hicks substitutes or complements. Principal Findings: Seven of the eight labor and capital factor inputs have positive marginal products while most factor inputs exhibit diminishing marginal productivity. Of 28 unique labor-capital factor input pairs, 17 are q-complements and 11 are q-substitutes. The q-complementarity of several labor types provides evidence of a mental health care team approach.Most other mental health workers are q-complements to physicians. Social workers and psychologists, who have related scopes of practice, are shown to complement each other’s services. Clinical support, including nursing staff, complements the services of psychiatrists, residents, psychologists, and social workers, documenting a difference in patient care responsibilities between the direct service providers and support staff for those service providers. Clerical and administrative workers complement the services of residents, APPs, social workers and psychologists, but substitute for physician labor. Four of eight principal minors, derived from the Hessian and evaluated at the factor input bootstrap means, exhibit signs opposite to that implied by concavity assumptions. We postulate incentive compatibility issues and interaction with the process of teaching residents as possible explanations of why these efficiency assumptions are not satisfied. Conclusions: These results represent observed tradeoffs among these mental health practices. They may serve to better inform healthcare providers regarding appropriate mental health workforce composition both in and outside of VA as dramatic increases are made in the volume of services over time.

Keywords: Hicks elasticity, mental health services, mental health workers, Production function, q-complementarity