Independent variables: ENVIRONMENT = changes in the ratio of the value of investment in pollution-control equipment to total costs in industry; five variables for changes in the intensity of OHS regulations that served as proxies for firm expenditures for compliance, INSPECTION for inspection rate, INFRACTION for rate of penalties imposed, REFUSAL for rate of interventions for refusal to work, PREVENT for percent of firms having a prevention program, and PROTECT for rate of protective reassignment; control variables were CYCLE, change in capacity utilization index (to control for cyclic fluctuations in the presence of quasi-fixed costs), SCALE, change in level of output (to control for effect of changing economies of scale), and ENERSHARE, change in the cost share of energy (to control for differing rate of productivity growth across industries in the presence of oil price changes, arising form their different energy-intensities). The data were obtained from various published and unpublished reports form sources such as Statistics Canada, the Federal Government, Bank of Canada, and CSST. Dummy variables for manufacturing sectors and time were also included as independent variables.
Barriers Statistical Tests:
Estimates were performed using generalized least squares procedures based on the cross- sectionally and time-wise autoregressive model presented by Kmenta (1986). Thus, the model accounted for possible serial correlation (annual industry data) and heteroskedasticity due to the cross sectional data from diverse sources. Additional statistical tests for exogeneity of variables were performed to see if productivity growth influenced the level of regulation (the exogeneity of the prevention program was rejected). Six models, based on 57 data points, were developed to investigate the effect of various groups of variables in the overall model (i.e., effect of removing ENVIRONMENT or ENERSHARE or SCALE). The final model for TFP contained nine variables.
Intermediate OHS Outcomes Final OHS Outcomes Cost of Intervention Economic Outcomes
Effect estimates: The prevention program variable (PREVENT) had a statistically significant (p<0.05) and a positive regression coefficient, meaning that it appears to have a positive effect on productivity growth. INFRACTION also had a significant and positive coefficient, and the implied contribution of both variables on productivity growth was 0.007 (at the sample mean). The authors suggest that these findings indicate that prevention programs and penalties have reduced the incidence of workplace injuries in the manufacturing sector, leading to a reduction of direct and indirect costs sufficient to have an enhancing effect on productivity growth. It should be noted, however, that the PROTECT variable was also significant and with its negative coefficient its implied contribution on productivity was -0.019. Thus the net effect of the OHS variables was -0.012 which is larger than the average impact found in a study of American OHS regulations (-0.003).
Institute for Work & Health