Supply Chain Uncertainty and Environmental Management
Methodology
A survey of Canadian manufacturing plants was conducted in spring of 2011. A sample of 1001 Canadian plants, located in the provinces of Quebec and Ontario, with more than 100 employees was randomly selected from the Canadian Scott's Directory. The Canadian Scott's Directory is a systematic and comprehensive dataset of Canadian manufacturing plant's executives contact information with data that are verified continuously to assure accuracy. The target respondent was the plant manager and a total of 251 responses were collected from which a total of 215 to 237 were usable for the different models tested. The effective response rate was 21.5%.
More specifically, the industries selected included those from the North American Industrial Classification Systems (NAICS) codes 315 to 337, mainly including discrete goods industries excluding process-based sectors such as paper, petroleum, and chemical products which are heavily controlled by command-and-control regulations. Also, the discrete good industries have more opportunities to perform product modifications than commodity-based industries leading to wider possibilities in terms of pollution prevention technologies.
Several nonresponse bias tests were conducted and revealed no indication of such a bias. To minimize key-informant bias, we contacted each plant by phone prior to sending the survey to identify the manager most knowledgeable about the environmental management at the plant.
Measurements
Supply chain uncertainty was measured using six items in the survey that were making inquiries about the degree of predictability of the supply base, the internal operations, and the demand as compared to the industry average. These items were developed specifically for this study and are presented in Table 1. They were inspired from the work in supply chain complexity. For example, Bozarth et al. suggest that complexity has a dynamic dimension, which is very close to the notion of supply chain variability and uncertainty. For example, they were explicit about demand variability and suppliers' unreliable deliveries in their description of dynamic complexity. Demand volatility and production schedule changes were metrics used in the literature to capture uncertainty in the supply chain. The different items were reverse coded to reflect supply chain uncertainty. Two items reflected the level of uncertainty from the supply base by determining the level of lots acceptance and the delivery reliability. The internal production system uncertainty was measured through the level of equipment reliability and the stability of the production schedule. Finally, two items aimed at capturing the uncertainty from the demand.
Table 1 Factor Analysis: Supply Chain Uncertaintya,b
| Items | Loadingsc | |
|---|---|---|
| Component 1 | Component 2 | |
| Demand stability | .875 | .129 |
| Demand forecasting accuracy | .889 | .099 |
| Level of supplier's delivery reliability | .273 | .762 |
| Level of supplier's lots acceptance | .049 | .873 |
| Reliability of the production equipment | .169 | .672 |
| Stability of the production scheduling | .624 | .357 |
| Eigenvalue | 2.788 | 1.209 |
| Cronbach's alpha (items in bold) | .764 | .698 |
- aThe leading question was: "Rate the following plant's characteristic against the industry average". The items were reverse coded to capture uncertainty
- bExploratory factor analysis using principle components with varimax rotation
- cComponent 1 = demand uncertainty; component 2 = supply uncertainty
An exploratory factor analysis was performed on these six items leading to a solution with two dimensions: (i) demand uncertainty (Cronbach's alpha = 0.764) and (ii) supply uncertainty (Cronbach's alpha = 0.698) (Table 1). The factor analysis indicated that the items considered as internal systems uncertainty were split between demand and supply uncertainty. The item pertaining to scheduling changes (arguably triggered by demand fluctuations) loaded on the demand uncertainty dimension. The item reflecting equipment reliability together with the two supply base related items can be viewed as uncertainty associated with the task of supplying goods to customers, hence, the label supply uncertainty.
A four-item scale was used to capture the degree of environmental practices implemented in a plant. This scale is a proxy for the level of resources for environmental management that are spent in a plant. These items asked the respondents to express the degree of resources invested in different environmental initiatives such as pollution prevention, recycling of materials, life cycle analysis, and waste reduction. This set of items was also used in previous studies linking lean management to environmental management. The factor analysis indicated that the four items were loading on the same component (Table 2) with a Cronbach's alpha of 0.794.
Table 2 Factor Analysis: Environmental Practicesa,b
| Items | Loadings |
|---|---|
| Pollution prevention | .788 |
| Recycling of materials | .800 |
| Life cycle analysis | .717 |
| Waste reduction | .842 |
| Eigenvalue | 2.483 |
| Cronbach's alpha | .794 |
- aThe leading question was: "Over the last 2 years, to what extent has your plant invested resources (money, time, and/or people) in programs in the following areas?"
- bExploratory factor analysis using principle components with varimax rotation
A second variable capturing the level of resources devoted to environment management in a plant was the proportion of the capital budget that was allocated to the environmental projects. The respondents were asked to indicate the percentage of the capital budget by selecting one of the seven choices ranging from less than 1 to 12% – the answers were coded on a scale from 1 to 7, accordingly.
The selection of different environmental technologies by plant managers (i.e., pollution prevention, pollution control, and management systems) was measured by a 'forced' allocation question in the survey. The respondents were asked to allocate 100 points to five different types of environmental expenditures: two were associated with pollution prevention, two were related to pollution control, and the fifth type was about management systems.
Plant and company size measured by taking the natural logarithmic transformation of the number of employees were both introduced as control variables: this is consistent with recent research in environmental management that has included organizational size in the analysis. The respondents were from the two largest provinces in Canada (Ontario and Quebec) with different regulatory context: a dummy was introduced ("province") in the analysis to capture such a difference. Finally, because a certified environmental management system such as ISO 14001 could have an impact on both the level and the type of environmental spending, a dummy variable was also introduce to capture plant's certification.