MeDEA

March 4, 2010

Metrics Data Envelopment Analysis (MeDEA)

Introduction to MeDEA
DEA is commonly used to evaluate the efficiency of a number of producers. A typical statistical approach for such application would be a central tendency approach which evaluates producers relative to an average producer. In contrast, DEA is an extreme point method and compares each producer with only the "best" producers. By the way, in the DEA literature, a producer is usually referred to as a decision making unit or DMU. In other word , DEA is a methodology directed to frontiers rather than central Tendencies. Instead of trying to fit a regression plane through the center of the data as in statistical regression, for example, one floats a piecewise linear surface to rest on top of the observations Because of this perspective, DEA proves particularly adept at uncovering relationships that remain hidden from other methodologies. For instance, consider what one wants to mean by efficiency, or more generally, what one wants to mean by saying that one DMU is more efficient than another DMU. This is accomplished in a straightforward manner by DEA without requiring explicitly formulated assumptions and variations with various types of models such as in linear and nonlinear regression models. Relative efficiency in DEA accords with the following definition, which has the advantage of avoiding the need for assigning a priori measures of relative importance to any input or output,

The heart of the analysis lies in finding the "best" virtual producer for each real producer. If the virtual producer is better than the real producer by either making more output with the same input or making the same output with less input then the original producer is inefficient. The procedure of finding the best virtual producer can be formulated as a linear program. Analyzing the efficiency of n producers is then a set of n linear programming problems.

Metric UK

Page 1

MeDEA

March 4, 2010

Simple Numerical Example

A simple numerical example might help show what is going on. Assume that there are four bank branches. For each branch we have Output measure: Number of personal transactions completed Input measure: number of staff The data we have is as follows: Branch Personal transactions ('000s) 125 44 80 23 Number of staff 18 16 17 11

Croydon Dorking Redhill Reigate

For example, for the Dorking branch in one year, there were 44,000 transactions relating to personal accounts and 16 staff were employed. How then can we compare these branches and measure their performance using this data? We view branches as taking inputs and converting them (with varying degrees of efficiency, as we shall see below) into outputs. Take some output measure and divide it by some input measure. In this case bank branches have a single input measure, the number of staff, and a single output measure, the number of personal transactions.

Metric UK

Page 2

MeDEA

March 4, 2010

Hence we have: Branch Personal transactions per staff member ('000s) 6.94 2.75 4.71 2.09

Croydon Dorking Redhill Reigate

As Croydon has the highest ratio of 6.94 we can compare all other branches to it and calculate their relative efficiency with respect to Croydon. Branch Croydon Dorking Redhill Reigate Relative efficiency 100(6.94/6.94) = 100% 100(2.75/6.94) = 40% 100(4.71/6.94) = 68% 100(2.09/6.94) = 30%

The other branches are relatively less efficient at using their given input resource (staff members) to produce output (number of personal transactions). We could, if we wish, use this comparison with Croydon to set targets for the other branches. For example we could set a target for Reigate of continuing to process the same level of output but with one less member of staff. This is an example of an input target as it deals with an input measure. An example of an output target would be for Reigate to increase the number of personal transactions by 10% (e.g. by obtaining new accounts). In practice, we can set a mix of input and output targets to achieve. Extending the example Typically we have more than one input and one output. For the bank branch example suppose now that we have two output measures (number of personal transactions completed and number of business transactions completed) and the same single input measure (number of staff) as before.

Metric UK

Page 3

MeDEA

March 4, 2010

The data we have is as follows: Branch Personal transactions ('000s) 125 44 80 23 Business transactions ('000s) 50 20 55 12 Number of Staff 18 16 17 11

Croydon Dorking Redhill Reigate

As before, a commonly used method is ratios, just as in the case considered before of a single output and a single input. Typically we take one of the output measures and divide it by one of the input measures. For our bank branch example the input measure is plainly the number of staff (as before) and the two output measures are number of personal transactions and number of business transactions. Hence we have the two ratios: Branch Personal transactions per staff member('000s) 6.94 2.75 4.71 2.09 Business transactions Per staff member('000s) 2.78 1.25 3.24 1.09

Croydon Dorking Redhill Reigate

Here we can see that Croydon has the highest ratio of personal transactions per staff member, whereas Redhill has the highest ratio of business transactions per staff member. Dorking and Reigate are relatively less efficient at using their given input resource (staff members) to produce outputs (personal and business transactions). One problem is how to combine the entire set of ratios into a single numeric judgment. This problem becomes even more complex given few extra branches

Metric UK

Page 4

MeDEA

March 4, 2010

Branch

Personal transactions per staff member('000s) 6.94 2.75 4.71 2.09

Business transactions Per staff member('000s) 2.78 1.25 3.24 1.09

Croydon Dorking Redhill Reigate

Graphical analysis
A simple graphical analysis can be used for problems involving just two outputs and a single input.

Metric UK

Page 5

MeDEA

March 4, 2010

The positions on the graph represented by Croydon and Redhill demonstrate a level of performance which is superior to all other branches. A horizontal line can be drawn, from the y-axis to Croydon, from Croydon to Redhill, and a vertical line from Redhill to the x-axis. This line is called the efficient frontier (sometimes also referred to as the efficiency frontier). Mathematically the efficient frontier is the convex hull of the data. The efficient frontier, derived from the examples of best practice contained in the data we have considered, represents a standard of performance that the branches not on the efficient frontier could try to achieve. We say that any branches on the efficient frontier are 100% efficient (have an efficiency of 100%). Hence, for our example, Croydon and Redhill have efficiencies of 100%. This is not to say that the performance of Croydon and/or Redhill could not be improved However we can say that, on the evidence (data) available, we have no idea of the extent to which their performance can be improved. It is important to note here that:

DEA only gives you relative efficiencies - efficiencies relative to the data considered. It does not, and cannot, give you absolute efficiencies.

We have used no new information here, merely taken data on inputs and outputs and presented it in a particular way.

Extending to more inputs/outputs

If we have just one input and two outputs then a simple graphical analysis is adequate. However, if we have more inputs or outputs then it is still possible to carry out EXACTLY the same analysis as above but using a Linear Programming algorithm developed by Metric called MeDEA.

Metric UK

Page 6

MeDEA

March 4, 2010

Applications
The simple Bank branch example described earlier may not convey the full view on the usefulness of MeDEA. It is most useful when a comparison is sought against "best practices" where the analyst doesn't want the frequency of poorly run operations to affect the analysis. DEA has been applied in many situations such as: o o o o o o o o health care (hospitals, doctors) education (schools, universities) banks manufacturing benchmarking management evaluation fast food restaurants retail stores

MeDEA can help you with;

identification of best practice identification of poor practice target setting resource allocation monitoring efficiency changes over time

Metric UK

Page 7

MeDEA

March 4, 2010

Strengths of MeDEA
As the earlier list of applications suggests, MeDEA can be a powerful tool when used wisely. A few of the characteristics that make it powerful are:

DEA can handle multiple input and multiple output models. It doesn't require an assumption of a functional form relating inputs to outputs. DMUs are directly compared against a peer or combination of peers. Inputs and outputs can have very different units. For example, X1 could be in units of lives saved and X2 could be in units of dollars without requiring an a priori tradeoff between the two.

Metric UK

Page 8