Le présent article poursuit l'analyse, effectuée dans Lanoie (1992a), de l'effet conjoint des différentes mesures préventives et des interventions de la CSST sur les risques d'accident du travail au Québec, en élargissant la période étudiée (1983-90 au lieu de 1983-87), en ajoutant une autre catégorie d'accident — les incapacités permanentes — comme variable dépendante, et en réévaluant ces relations dans les secteurs à risque uniquement. Comme dans cette étude de Lanoie (1992a), nos résultats indiquent que les politiques adoptées au Québec ont, au mieux, engendré une diminution mineure de la fréquence des accidents durant cette période. Les faits marquants de cette étude sont l'impact constant, par rapport à 1987, des mesures d'intervention de la CSST sur la fréquence et la gravité des accidents dans l'ensemble des industries, l'effet plus prononcé des inspections sur la fréquence des accidents dans les secteurs à risque, et la relative inefficacité des mesures adoptées par la CSST à prévenir les accidents avec incapacité permanente.
In Québec, policies regarding health and safety in the workplace, administered by the CSST (Commission de la santé et de la sécurité du travail), are among the most progressive in North America. As in other Canadian provinces, the govemment acts as an insurer, paying compensation to injured workers and receiving insurance fees from firms, adjusted for past accidents by an experience rating System. In addition, five innovative prevention programs have been created in Quebec in order to protect workers against unsafe working conditions.Since the adoption of workplace safety and health policies, many econometric studies have tried to evaluate their impact on accident risk in the workplace. Economists have usually used the frequency or duration of accidents to approximate the risk of accident as a dependent variable, and have used proxies of the application of some policies (usually the inspections of workplaces for conformity) as independent variables. In North America, studies using aggregate data have shown few, if any, significant impact of safety programs on the frequency or duration of accidents ; however, studies using data at the firm level have led to more positive results, detecting effects of inspections on inspected firms.In Canada, this type of study has only been conducted once, by Lanoie (1992a), with Quebec data at the industry level for 28 different sectors on a sample for the period 1983-1987. It explained the frequency and severity of accidents with variables measuring each of the policies adopted by the CSST in 1980, and with control variables including socioeconomic characteristics of workers for each industry. It was the first study to take account of all the government health and safety measures at the same time (compensation, experience rating and prevention). But the only significant result obtained was a minor impact of inspections on frequency and severity of workplace accidents.The present paper extends Lanoie's (1992a) study in three ways. First, extending the previous sample of 1983-87 to 1983-90, enables us to evaluate if these relatively new policies have been more effective in recent years, showing an adaptation of firms and workers' behaviour to the new regulations (1980). Second, by adding permanent disabilities as a new dependent variable, we can evaluate the specifie impact of each safety policy on the frequency of the most expensive injuries for society. And, finally, by reestimating the same relations in the 15 most risky industries only, we can analyze the specifie impact of these policies in most risky industries in which safety efforts of the CSST have been concentrated. To estimate these relations, we drew upon on a theoretical model inspired by Viscusi (1986) in which the risk of accident is influenced by workers and firms. An intensification of prevention policies or interventions may reduce the risk of accident because of the increased opportunity cost of accidents for firms. But the risk may be unaffected if firms are already taking care of workplace safety or if, at the same time, the workers' environment appears so safe to them that they reduce their own protection efforts. Three particular aspects of our results must be stressed. First, by extending the sample to 1983-90, we did not detect more important effects of preventive measures on accident risk. However, the experience rating mechanism, and the change of generosity in compensation now have a significant impact on frequencies of accidents. Second, when we compare the effects of the CSST policies on all industries to their effects on the most risky industries, we find that inspections have more impact on accident risk in risky industries. That result is consistent with those obtained by Gray and Scholz (1990). Finally, the study of permanent disabilities shows that neither preventive measures nor other interventions have any major effect on the frequency of this type of accident. In fact, accidents of this category are usually fortuitous and may be hard to prevent altogether. In terms of social policy, these results imply, as in the previous paper, that inspections appear to be the most effective preventive policy adopted by the CSST, while the other more innovative ones do not seem to have any significative impact on accident risk. Moreover, in order to reduce the incidence of workplace accidents, our results show that emphasizing the experience rating System, as well as targeting inspections in risky industries, may be helpful.Finally, the generosity of the compensation paid by the CSST to workers seems to induce them to report more injuries. For the first time, our results show that this behaviour is more prevalent in less risky industries. We therefore believe that the CSST could reevaluate the level of its compensation and its control of workers' access to the compensation System.
