Efficient and Inefficient Composites in the U.S. Domestic Property-Casualty Insurance Industry
Ellis, Peter M., Southern Business Review
This work has the goal of establishing the comparative financial efficiency of the several operating composites of the U.S. property-casualty insurance industry. This industry is highly important to national private and commercial needs. Consumers need such coverage as auto and homeowners insurance. Industry needs a wide range of insurance coverage, including such lines as credit, financial guarantee, commercial, professional liability, and workers compensation. The goal of financial efficiency is important for several reasons. Solvency is vitally important to insureds and to stockholders. Cash flows need to be secure so that losses can be covered quickly. Net premiums must be sufficient to cover losses and expenses. Thus, regulators, management, stockholders and insureds all have an interest in preserving financial efficiency. This work will establish a measurement of relative financial efficiency for all of the major sectors of the U.S. domestic property-casualty industry. For the composites that are identified as relatively inefficient the nature of the inefficiency can be understood through the related slack and surplus variables of the several input and output measures.
Entities, whether governmental, private or commercial, can be thought of as having a set of inputs, some processing activities and a set of outputs. There is a sense that the entity is efficient if it obtains a great amount of output while expending few inputs. Data Envelopment Analysis (DEA) is typically used to compare the relative efficiency of each of a set of operating units. These operating units are usually called decisionmaking units (DMUs). The technique was pioneered by Chames, Cooper and Rhodes (1978) and extended by Banker, Chames and Cooper (1984). Cooper, Seiford and Tone published a text on the use of DEA (1999). Data Envelopment Analysis is steadily replacing multiple regression analysis as a tool in efficiency studies because it can simultaneously incorporate several output variables, whereas multiple regression studies permit just one dependent variable at a time.
Data Envelopment Analysis Applications
The cross-sectional application of the DEA technique has been applied in many environments. In the public sector McCarty and Yaisawarang (1993) did a DEA analysis of the several school districts in New Jersey. Vanden Eeckaut, Tulkens, and Jamar (1993) used the method to compare efficiencies of a group of municipal governments. An excellent application from the financial sector was the use of DEA to compare operational efficiency of the several branches of a regional bank (Lovell & Pastor (1997). Asimilar study was carried out by Barr, Seiford and Siems (1993). Siems and Bañil 998) extended the work by benchmarking efficiency throughout the United States. The groundbreaking work by AIy et al. (1990) showed how to extend DEA to establish the nature of returns to scale, and then applied the method to the U.S. banking industry.
An early application in the life insurance industry was that of Cummins and Zi (1998). They used the ability of DEA to provide an efficient frontier to compare the various firms. This was followed by the Cummins, Weiss, and Zi (1999) article that used DEA to compare and contrast stock and mutual property-casualty insurers. Ellis (2006) used DEA to show the consistent level of operational efficiency over time for a particular auto insurer. That work presented an innovation that created a single efficient DMU from which all existing entities can be compared. Doing so avoids the possibility that any relatively inefficient entities would be compared to differing efficient subsets. Ellis (2006-2007) also employed DEA with the single efficient DMU to track banking industry efficiency over time.
U.S. Property-Casualty Insurance Data
A DEA study requires comparing competing entities based upon the levels of a set of inputs that are used for the purpose of generating a set of outputs. The data was collected from Best's Aggregates and Averages (2006). …