Analysis

Core Variables

One of the aims of the MEPI project was to develop practical approaches to environmental performance indicators. Reducing the number of indicators required to describe business environmental performance - thereby simplifying the task of data collection and analysis - is one way of improving the practicability of performance measurement.

Statistical analysis carried out identified a more limited number of 'core indicators' that can give a good representation of the overall environmental performance of a firm. Principal Component Analysis (PCA) was used to identify those variables which explain most variability within the data sets and therefore represent a simplified account of the environmental performance of a company. For example, in the printing sector CO₂ and SO₂ emissions give a good representation of all air emissions. These core indicators are distinguished from generic indicators used in the MEPI project (generic indicators are those for which data was collected across all six sectors in the study).

The four tables below list core indicators for four industrial sectors (data for the fertilisers and computer sectors was not complete enough to conduct a PCA). These indicators can be used as a starting point in performance measurement in these industrial sectors. The analysis distinguished between organisational and environmental variables. The number in brackets indicates the number of cases for which data was available.

TABLE 1: Variables with most explanatory value in the book and magazine printing sector

Organisational Variables		Environmental Variables (all per functional unit)
Management variables	Business variables	Waste	Air emissions	Factor inputs	Water consumption	Energy input
EMAS registration (74) ISO registration (74) No of non-compliance events (21) Environmental investment reported (60)	Total sales (47) Profit (14) No of employees (62)	Total solid waste (39) Hazardous waste (25)	CO2 (14) SO2 (16)	Total ink (53) Isopropyl alcohol (26)	Total water input (45)	Total fuel input (41) Total energy input (46)

TABLE 2: Variables with most explanatory value in the electricity generation sector

Variability

The data revealed wide variability in the environmental performance of companies operating in the same sector. It also found that the pattern of variability was not consistent across different dimensions of performance. That is, greater variability was discovered across some performance indicators than others. This may be explained by technological factors (different production processes may be used to produce the same output but with very different environmental characteristics), the effect of regulation (regulatory pressure may be expected to produce greater convergence in environmental performance), and the effect of relative prices (different producers may choose to optimise their facilities differently depending on the price of inputs and pollution control).

The figures below give illustrative examples from the electricity and pulp and paper sectors of patterns of performance variability identified in the MEPI data set. The figures show firm-year data normalised by production output (electricity generated and paper produced) for variables organised by size of firm. Note that the x-axis in each case is logarithmic.

FIGURE 1: Carbon dioxide emissions in the electricity generation sector

FIGURE 2: Sulphur dioxide emissions in the electricity generation sector

FIGURE 3: Solid waste generation in the pulp and paper sector

FIGURE 4: Water input in the pulp and paper sector

Correlation between business, management and environmental variables

Another aim of the MEPI study was to understand better underlying patterns in business environmental performance. In particular, we were interested in understanding whether there are relationships between aspects of business and management performance and environmental performance (for instance, are more profitable firms higher environmental performers?). The core models applied are:

1. Business Performance = f ( management variables or environmental variables )

Can we predict business performance from management or environmental data?

2. Management performance = f ( business variables or environmental variables )

Can we predict environmental management performance from the business data or the environmental variables?

3. Environmental performance = f ( management variables or business variables )

Can we predict the environmental performance from the business variables or the environmental management variables?

Regression analyses were carried out, using the reduced core variable sets only. All regressions were carried out using environmental indicators normalised by 'functional unit (FU)'. Multiple linear regressions were carried out with stepwise entering of dependent variables. This means that the regression model is optimised for the proportion of variability explained by first entering the independent variable that explains the largest part of the variability for the dependent variable (i.e. has the highest correlation coefficient with the dependent variable). In the next step, the independent variable that explains the second largest part of the variability encountered for the dependent variable is entered until adding further variables does not improve the regression model. The main way to treat missing values in the regressions was to replace empty cells with the mean for that variable.

This analysis was carried out for all the core variables identified for five of the MEPI sectors (not including computer manufacture). A summary of all statistically significant results is shown in the Table 5. A "+" in brackets following the variable name means a positive correlation, a "-" a negative regression coefficient. 'No result' signifies that no significant correlations could be found between the variable identified in the left hand column and other variables.

The table summarises many results, only a few of which are highlighted:

Environmental management

For the data collected by the MEPI study, there is no statistically significant relationship between ISO certification or EMAS registration and any of the key environmental performance variables in four of the sectors (a significant effect was identified only in the fertiliser sector). Firms with an environmental management system certified to ISO 14001 tend to have lower emissions of nitrogen to water, but also produce higher emissions of NO_x to air. In the electricity sector it a positive correlation between separate disclosure of environmental investments and profits, as was a positive correlation between ISO certification and profits in fertiliser manufacture.

Business profitability

Analysis showed a mixed picture for the relationship of profits and environmental performance. For instance, more profitable paper sector firms tend to produce less waste, whereas they also tend to emit higher COD emissions to water. A negative correlation between waste generation and profitability was also found for the textiles and printing sectors. However, a positive relationship was found between water inputs and profitability in these same two sectors (more profitable firms use more water). Across all sectors, no clear link between high environmental performance and higher profitability was found. What is more, none of the environmental variables identified as having a significant relationship with profits were found to do so in more than one sector. The link between environmental performance and profitability seems thus to be not only relatively weak, but also highly sector-specific.

Business size

Using sales and number of employees as proxies for the size of businesses, an even more complex picture emerges. For instance, where COD emissions are significant indicators, the correlation tends to be positive (higher COD emissions are correlated with higher sales and number of employees in the pulp and paper and fertiliser sectors). A positive correlation is also found for CO₂ and total waste in the textile finishing sector (larger firms tend to produce more pollution and waste). However, negative relationships are found between energy and water inputs and size in the electricity and textile finishing industries. These results suggest that some of the assumptions about differences between the environmental performance of larger and smaller firms may have been overstated. However, caution is required in interpreting these results. Different variables are correlated in different sectors.