Using DEA and VEA to Evaluate Quality of Life in the Mid-Atlantic States
Marshall, Elizabeth, Shortle, James, Agricultural and Resource Economics Review
In this study we use data envelopment analysis (DEA) and an extension of DEA called value efficiency analysis (VEA) to explore the "production" of quality of life within counties in the mid-Atlantic region and the extent to which production frontiers and efficiency differ between rural and urban counties. These methods allow us to identify counties that are inefficient in their quality of life production, and to rank (using DEA) those counties according to their distance from a performance standard established by other observed counties, or (using VEA) by a single unit designated as "most preferred."
Key Words: data envelopment analysis, value efficiency analysis, quality of life
Speculation regarding the relationship between the attributes of a community and the quality of life experienced by its residents has gained vigor in recent years. In contemporary discussion, the issue arises most visibly in debate about the dispersed, automobile-dependent development patterns known as urban sprawl, and in burgeoning interest in such concepts as smart growth and liveable or sustainable communities. Additionally, observed changes in migration patterns nationwide, together with increasing interest in the role that natural amenities play in residential choice behavior, highlight the importance of understanding how residents value the various attributes associated with different types of communities. Given the abstract, multi-dimensional nature of the underlying concept of quality of life, however, quantifying relationships between community attributes and quality of life poses significant challenges.
Efforts to measure quality of life confront two equally challenging tasks. One is to identify a set of indicators that represent appropriate dimensions for measuring quality of life. Such indicators are often selected to reflect economic, social, and environmental factors. The second task is to aggregate such indicators into a composite measure that can be used to differentiate communities along a quality of life spectrum. One of the most commonly used methods for evaluating quality of life has been the hedonic price method. This method is based on theoretical work by Rosen (1979) and Roback (1982) suggesting that, at a labor- and land-market equilibrium, the value of regional amenity and quality of life factors should be capitalized into regional wages and rents (Delier 2001). Differentials among regional wages and rents should therefore reflect differences in quality of life, and these differentials can be used to estimate the values attached to each amenity factor. Blomquist, Berger, and Hoehn (1988) used hedonic wage and rent models to estimate implicit amenity prices for a variety of regional climatic, environmental, and urban factors; these prices then served as weights in a quality of life index applied to urban counties. A more recent application of this technique is Gabriel, Mattey, and Wascher (2003), who extend the hedonic equation system to include the price of locally traded goods other than housing, and apply the analysis to pooled cross-section and time-series data on a large variety of amenity and quality of life variables. This allows them to not only estimate the implicit price of amenity factors but also construct a state-level quality of life index based on these amenity weights.
Hedonic amenity weight estimates, and the quality of life indices that arise from them, are sensitive to the specification of the functional form linking amenities with existing wage and income differentials. Non-parametric quality of life indices avoid this issue. In an effort to eliminate completely the need for "ad hoc" selection of quality of life indicators, Douglas and Wall (1993) appeal to the legacy of Tiebout (1956) in arguing that migration patterns should reflect quality of life differentials: mobile residents will "vote with their feet" for those communities with high quality of life. They therefore construct a quality of life index based on the presumption that the probability of a resident moving from location A to B will depend on the degree to which the quality of life in location B exceeds that of location A (Douglas and Wall 1993). …