An evaluation of technical efficiency and managerial correlates of solid waste management by Welsh SMEs using parametric and non-parametric techniques

Abstract

Strong economic growth and environmental regulation stimulus make Welsh small and medium enterprises' (SMEs) sustainability performance merit investigation in the context of European Union (EU) sustainability initiatives. This is due in part to strong economic growth and the stimulus provided by environmental regulation. We use stochastic frontier analysis, a parametric econometric technique to generate estimates of the technical efficiency of solid waste management by 299 Welsh SMEs in 2003. We demonstrate that the ranking and efficiency scores of the Welsh SMEs studied correlate significantly with non-parametric data envelopment analysis efficiency measures and are related to the use of environmental auditing practices and the use of local business support groups, but not to monitoring of waste expenditures and publication of environmental policies.

Appendices

Appendix A

A.1. Sampling and methodology

The survey instrument was organized into three parts, part I required general information from the company including confirming SIC code and full time employment equivalent (FTE). Part II of the questionnaire noted the waste details, particularly:

• Types of waste (EWC Code);

• Quantity (either by weight or container type, size and quantity;

• Whether waste was special;

• The physical form of the waste;

• The waste management option;

• The waste contractor;

• Initial and final destination of the waste; and

• The cost of or income from the disposal.

Part III of the questionnaire asked the company questions about its general approach to management of environmental issues.

After full agreement by facilities to be involved in this study, data were acquired by a team of 13 surveyors who went through an intensive two-week training programme. The training programme provided guidance into the survey, the data within the survey, and standardization of the data collection process, and the data integrity process. This standardization reduced the errors and bias that might occur with multiple surveyors gathering data. The surveys were completed by the surveyors with observations (site inspections) made to confirm waste stream data.

Data integrity was also included in the follow-up to the actual data acquisition from the facilities where members of the Wales Environment Agency and the research team inspected the data and requested additional information where discrepancies in data occurred.

A stratified sampling procedure was designed drawing a random sample provided by the British Office of National Statistics (ONS) interdepartmental business register database. The data for this study were collected using broad-based site visits throughout Wales, using a short survey questionnaire. To ensure consistency in data collection the data collected was based on a series of predetermined categories:

• Standard Industrial Classification (SIC95) codes were used to identify business sectors;

• European Waste Catalogue codes (EWC) were used to identify waste type;

• A standard set of conversion factors were developed by the EAW to calculate weight from volume data; and

• Standard lists were developed to define—

  • waste management options;

  • skip and container types;

  • waste form (solid, liquid, sludge, etc);

  • hazardous/non-hazardous;

  • waste contractor and waste carrier names; and

  • initial and end destination local authority names.

The waste survey sample was designed to include 2210 companies. In this project it was decided that sectors with a high economic impact on the Welsh economy would be highlighted. The survey methodology used a stratified random sampling approach based on SIC codes and number of employees (size) of a company. Each company within a survey cell (SIC Code/size sub-sample) was randomly chosen. The ONS were provided with the sample categorizations and were requested to provide the necessary randomly generated populations, in this instance a total of 2210 companies. The sample was also proportioned to each region within Wales. This breakdown allowed the sample to be representative of Wales as it reflected the Welsh economy.

To ensure no bias appeared in the survey, the randomness of company selection had to be maintained throughout the life of the survey. As a result, all companies had to be approached in the order provided by the ONS. With a target of 2210 companies to be surveyed, a sample of 7237 companies were randomly allocated by the ONS. Only the companies on the first list of live 2210 companies, identified by appropriate coding could be approached first. The remaining companies could only be approached in the necessary order determined by the ‘Replacement company’ process. This was either to ‘Abandoned’, ‘Refused’ or ‘Uncontactable’. An explanation of these terms is as follows:

• 1 Abandoned companies were classified as those companies that had ceased to exist.

• 2 Duplicate companies were those companies who appeared more than once.

• 3 Uncontactable companies were not accessible by phone or postal.

• 4 Refusals—either due to unavailability of people or time frame of study or other reasons.

In the overall study a total 2122 surveys were actually completed from site visitations to SMEs in a broad variety of industries. There were 1198 who did not participate in the survey with 2122 respondents, a response rate of 64% (2122/3320). Of these 1198 non-participants, 82 were abandoned because the companies did not exist, 76 were duplicates of other companies already included in the sample, 368 were uncontactable for various reasons and 672 companies refused to participate. If we remove the abandoned, duplicate, and uncontactable companies from our sample size, our response rate only including those that either responded or refused to respond for the total sample is 85%. That is, the overall initial sample from this included a broad variety of companies and sizes, including non-manufacturing and service oriented organizations.

Since we wished to study manufacturing oriented companies we narrowed down our SIC coded respondent companies (from the total survey respondents of 2122) to those organizations that had between 10 and 250 employees. Using the EU designation companies from 10 to 50 employees were defined as small, while those that ranged from 51 to 250 employees were medium-sized companies. We also eliminated companies that had reported zero costs or wastes from our analysis. This reduced the total number of SMEs in our sample to 299 companies.

Appendix B

B.1. DEA models and extensions

B.1.1. Basic ratio-based technical and scale efficiency DEA models

DEA productivity models for a given DMU can use ratios based on the amount of outputs per given set of inputs. The definition of a DMU can vary greatly, from individuals to countries, as long as the unit can be modelled with input and output values. The definition in our case is a facility or company surveyed in our study. DEA allows for the simultaneous analysis of multiple inputs to multiple outputs, a multi-factor productivity approach. The general efficiency measure used by DEA is best summarized by Equation B.1.

where (E _ks ) is the efficiency or productivity measure of DMU s, using the weights of test DMU k; (O _sy ) is the value of output y for DMU s; (I _sx ) is the value for input x of DMU s; (v _ky ) is the weight assigned to DMU k for output y; and (u _kx ) is the weight assigned to DMU k for input x.

In the basic DEA ratio model developed by Charnes et al (1978) (CCR), the objective is to maximize the efficiency value of a test DMU k from among a reference set of DMUs s, by selecting the optimal weights associated with the input and output measures. The maximum efficiencies are constrained to 1. The formulation is represented in expression (B2).

maximize

subject to:

This non-linear programming formulation (B2) is equivalent to formulation (B3) (see Charnes et al (1978) for a complete transformation explanation):

maximize

subject to:

The transformation is completed by constraining the efficiency ratio denominator from (B2) to a value of 1, represented by the constraint ∑ _x I _kx u _kx =1.

The result of formulation (B3) (the CCR formulation) is an optimal simple or technical efficiency value (E _kk ^*) that is at most equal to 1 (this formulation has also been defined as the constant returns to scale formulation). If E _kk ^*=1, then no other DMU is more efficient than DMU k for its selected weights. That is, E _kk ^*=1 has DMU k on the optimal frontier and is not dominated by any other DMU. If E _kk ^*< 1, then DMU k does not lie on the optimal frontier and there is at least one other DMU that is more efficient for the optimal set of weights determined by (B3). The formulation (B3) is executed s times, once for each DMU.

The dual of the CCR formulation (also defined as the envelopment side) is represented by model (B4):

minimize θ

subject to:

The CCR model has an assumption of constant returns to scale for the inputs and outputs.