INSTITUTIONAL CHANGE AND INDUSTRIAL ACCIDENTS: …



INSTITUTIONAL CHANGE AND INDUSTRIAL ACCIDENTS: SPAIN, 1900-1936(

Javier Silvestre

Universidad de Zaragoza

Correspondence address:

Facultad de Ciencias Económicas

C/ Doctor Cerrada 1-3

50.005, Zaragoza, Spain

e-mail:

javisil@unizar.es

Very preliminary draft (February 2004):

Please do not quote without author’s permission.

* Most of the previous work for this article was carried out during a visit as a Pre-Doctoral Marie Curie Fellow in the Department of Economics at University College Dublin. I would like to thank Cormac Ó Gráda for his advice, hospitality and encouragement. I am especially grateful to William Boal for his very helpful insights. Earlier versions of this work have also benefited from suggestions by Mª Isabel Ayuda, Sebastián Coll, Luis Germán, Alfonso Herranz, Vicente Pinilla, David Reher, Jaime Reis, Blanca Sánchez-Alonso and Daniel Tirado.

INTRODUCTION

Economic and social historians discuss the impact of the Industrial Revolution on living standards. This debate is based on the analysis of the evolution of a number of economic and noneconomic indexes such as wages, prices, consumption, rents, employment, human capital, mortality, life expectancy, height, health, housing and urban environment, working conditions, working hours and leisure time, political rights, etc. However, although the impact of urban conditions has received considerable attention, one of the least studied fields refers to the impact of changes in working environments on health. In this regard, one limitation of Occupational Safety and Health (OSH) studies in historical contexts is the heterogeneity and lack of quantitative data. Indeed, this is true to such an extent that Bartrip and Burman (1983, p. 7-14) are reluctant to state whether the British Industrial Revolution increased or decreased the level of work accidents.

In fact, studies by Bartrip and his collaborators, McIvor, and others, represent a "revisionist" point of view on the issue of the impact of the Industrial Revolution on workplace accidents. This new interpretation is far from the optimism based on technological improvements and changes in labour relations, on the one hand, and the Marxist-oriented pessimism based on the degrading process of capitalist work, on the other (McIvor, 2001, pp. 111-113). The new perspective on the study of OSH is more complex, less generalistic, and prone to include institutional factors. One of the areas of research deals with the establishment of public institutions whose aim was to reduce industrial accidents and to improve workers´ health. As pointed out by Bartrip y Burman, (1983, p. 14) "To an extent, however, the question of magnitude is of limited importance, for the significant point is that industrial injury came to be perceived as a serious issue". In this sense, if we adopt the reasoning of Fraser (2003, p. 11) with regard to child labour, work accidents did not begin with the industrialisation process, but "the Industrial Revolution concentrated and multiplied what had previously been diffuse and remote from the public gaze".

In any event, as has been shown by a number of studies devoted to both modern and historical times, the creation of OSH administrations might be not enough to improve working conditions. The gap between aims and results can be wide, and determined by several economic and institutional factors. Thus, OSH administrations can require a period of adaptation before reaching an acceptable level of performance of previously enacted laws and standards. The effectiveness of accident prevention systems can be also determined by the existence and extent of accident liability systems, in such a way that workers´ compensation might alter the system of incentives to reduce accidents on the part of both workers and employers. Moreover, state intervention is not the only institutional factor determining the level of safety. Abundant works coming from different social sciences, economics, industrial relations, history, sociology, psychology, etc., have shown that unions involvement and, in general, workers´ attitude to this issue is crucial. Workers are, in short, in contact with risk and, therefore, their collaboration with managers and technicians is very relevant when seeking to achieve a successful implementation of OSH policies.

This work offers a historical case study aimed at contrasting the efficiency of safety institutions. Spanish labour and social history distinguish between two periods in the implementation of OSH policies. During their first two decades of implementation, early Spanish OSH policies designed to reduce industrial accidents during the industrialisation of the economy has been said to be particularly inoperative. According to those studies, the 1922 reform of the legal framework provoked substantial changes in the effectiveness of Spanish OSH policies from the early 1920s to the outbreak of the Civil War (1936). This reform was based on the compliance of Spain with several International Labour Organization (ILO) agreements on OSH, the promotion of formative actions, and the reinforcement of economic incentives and punitive actions. However, accidents rose during the 1920s. This work shows that there was only a modest improvement of the effectiveness of OSH policies from the early twenties. In any event, this improvement was insufficient to counteract the strong procyclical relationship between the economic cycle and accidents. This work also shows that, for several reasons, OSH practices were not one of the main priorities in Spanish union demands during the industrialisation, with the result being a weak impact of unions on the reduction of accidents. The rest of the work is organised as follows. Section I reviews the debate on the effectiveness of OSH policies. Sections II and III offer a description of the structure, evolution, and institutional framework of industrial accidents in Spain. The specification, data and estimation of the model are considered in Sections IV and V. The paper closes with a review of the main conclusions in Section VI.

I. THE EVIDENCE ON THE EFFECTIVENESS OF OSH POLICIES

Empirical studies on workplace accidents are usually devoted to studying the impact of public regulation, the relationship between workers´ compensation and safety, and the role played by unions. First, empirical works on the effectiveness of workplace injury regulation have mainly concentrated on the US since the Occupational Safety and Health Administration (OSHA) was established in 1970. Econometric evidence tends to show that regulation has had little or no impact on workplace safety.[1] Although these studies do not usually focus on the causes of inefficiency, the shortage of inspectors, the reluctance to initiate actions to close down companies or impose high fines when serious violations are found, as well as the standard setting process, are the main motives proposed (Viscusi, 1992, pp. 198-199, 219-220; Dorman, 1996, pp. 191-197; Kniesner and Leeth, 2000).

Second, from the theoretical point of view, it is not clear whether higher workers´ compensation increases or decreases worker safety. On the one hand, higher benefits might induce workers to reduce their accident-preventing efforts. Additionally, higher benefits might lead workers to report more accidents. These moral hazard factors will tend to increase the number of registered accidents. On the other, an increase in benefits coverage will raise firms´ incentives to provide safer workplaces. Logically, higher firm costs will induce employers to improve safety and, finally, reduce accidents. Thus, as Fortin and Lanoie (1998, p. 9) remark, the net impact of workers´ compensation depends on the degree of complementary or substitutability between the safety efforts of each party, as well on the nature of the labour contract.[2]

Third, apart from enforcement of the law and wage incentives, OSH literature has also considered the way in which standards are implemented. OSH studies included in the meta-analysis by Smallman (2001, pp. 408-411) shows the relevance of partnership and compliance in improving safety at work.[3] These studies emphasize the importance of worker participation, independent or union-based, in safety institutions or committees composed of managers, authorities, stakeholders, etc. In short, this line of research proposes that safety legislation is better implemented when all parties are encouraged to participate in the safety process.

Empirical analyses in historical contexts are very few. If we first consider the impact of regulation on injury rates, the conclusions also tend to be pessimistic. In the case of Britain, Bartrip and Fenn (1988) find no significant effect of expected penalties on the level of reported factory fatalities in the period 1878-1913, due to the limited number of inspectors and the low level of fines.[4] In the case of the US, Fishback (1986) finds insignificant effect of legislation on mine safety during the Progressive Era (1912-1923), also due to weak enforcement.[5]

The review by Fishback and Kantor (2000, pp. 77-82), on studies devoted to the impact on accident rates after increasing worker compensation in the US during the first decades of the twentieth century, shows that the results vary considerably from industry to industry. Thus, employers´ reaction improving prevention dominated employees relax and moral hazard behaviour in manufacturing and railroads, where the costs of prevention were relatively low. By contrast, in the case of the coal industry, where workers played a more active role in prevention, accidents increased. For the case of Britain, Bartrip and Fenn (1988) have suggested that the increase in fatality rates after replacing the employer negligence system with a no-fault system was due to an increased number of deaths reported by relatives.

The relevance of unions in reducing accidents has been studied for early twentieth-century U.S. coal mining. The recent work by Boal (2003) shows that the effect of unionism on accidents was far more significant than the impact of safety laws and technological innovations.[6] In this sense, Fairris (1998) suggests that the rise in U.S. manufacturing injury rates since the late 1950s was due to changes in shopfloor governance that reduced the power of workers to influence shopfloor conditions. Finally, several safety regressions for U.S. railroad, mining and manufacturing industries in Aldrich´s (1997) Safety First book show positive impacts of different proxies of safety regulation on reducing accidents during the first two decades of the twentieth century. According to Aldrich, this improvement in safety was due to an alliance of reformers, unionists and workers, and business executives in response to higher levels of accidents in the U.S., in contrast to less capital-intensive European countries such as Britain, Germany and France.

On the other hand, qualitative evidence also demonstrates workers´ reluctance to collaborate with safety regulations when its implementation implies changes in the way that work is organised. This could have been especially the case during the nineteenth and early twentieth centuries, when changes in safety organization were often imposed by authorities influenced by social reformers and technicians, and changed traditions or forced workers to adopt new procedures (Fishback, 1986, p. 292; Ramm, 1986, p. 110; Aldrich, 1997, pp. 85-86, 126-143, 157-160, 191-193, 210, 237-240, 276; Aries, 1998, p. 54-55; Rodgers, 1998, p. 247). Additionally, it should not go unremarket how both the General Secretary of the Red Cross (Sand, 1923) and the ILO (1924) pointed out how the predominance of wage and economic demands in European unions priorities could have affected a better implementation of OSH programs.

II. INDUSTRIALISATION AND INDUSTRIAL ACCIDENTS IN SPAIN

Spanish statistics for fatal and nonfatal accidents (Estadística(s) de los Accidentes del Trabajo) before the Civil War (1936-1939) began in 1904 (more than sixty years after the first U.K. official figures) and extended to 1934. The early twentieth century represents a remarkable period in recent Spanish history, with the economy and society changing dramatically during those years. The country underwent an intense structural change, led by

the contraction of the agricultural labour force (stable at around 72 per cent until 1910, it reached 51 per cent in 1930); and the diversification of the industrial sector, with the appearance or rapid growth of intermediate products and investment goods industries. Spain reduced its difference in per capita GDP with the European industrial core, and on the eve of the Civil War the country was far more urbanized than at the turn of the century.

Until 1924 the statistics on accidents were compiled by the Instituto de Reformas Sociales (Institute for Social Reforms), the public institution devoted to labour affairs. In that year the Instituto was absorbed by the Spanish Ministry of Labour. The Instituto admits to some problems in data collection for the first years, but the data are said to be collected on a consistent spatial and industrial basis from 1909 onwards. The statistics include data at regional level on the kind of injury caused by the accident, that is to say, slight, pending evolution, serious, partial disability, total disability, and death. On the other hand, the statistics also include the number of accidents at industry level, although unfortunately without distinguishing between categories such as fatal and nonfatal.

One figure coherence problem may arise from the fact that the 1922 reform of the Labour Accidents Law (1900) improved the reporting of accidents. Thus, the new system required both employers and workers to report any kind of accident, in an attempt to overcome the shortage of accident reports provided by some companies.[7] Figure 1 shows the number of fatal and nonfatal accidents reported for the whole period. The change in the tendency from 1922 onwards is clear. In any event, it is also true that the 1920s was a period of intense economic growth in Spain (Prados de la Escosura, 2003). It is well know that, caeteris paribus, the number of accidents is strongly influenced by economic cycles, rising with upsurges and declining during recessions (Robinson, 1988; Fabiano et al, 1995; Adnett and Dawson, 1998; Fairris, 1998; Ruhm, 2000). Figure 2 shows total accidents and GDP from 1904 to 1934. Following Ruhm (2000), both variables are detrended, using a linear trend, and normalized by substracting the mean of the detrended variable and dividing by its standard deviation. The Figure illustrates a clear procyclical relationship between GDP and accidents. In order to confirm this relationship, in Table 1, I run regressions of the log of accidents on the log of GDP or the log of the estimated unemployment (and a constant) for 1909-1930, the period under further econometric study in Section IV. Estimations yields elasticities between 2.3 and 3.7 for the case of GDP and around -0.3 for the case of estimated unemployment.[8]

Since the level of risk is not equally distributed, labour accidents vary considerably among industries. For the Spanish data, there is a lack of correspondence between the decennial population censuses industry classification and the industries included in the accident statistics. This problem does not allow to calculate industry accident rates, that is to say, relative indexes of risk, for a considerable number of industries.[9] Table 2 offers accident rates for eleven industries. In any event, these industries are relevant from the point of view of both output and accidents. A second problem when making time comparisons at industry level is that the number of registered accidents decreases around 1920, with the Estadística(s) de los Accidente(s) del Trabajo not making any reference to this fall. For instance, in the case of Mining, the registered number of accidents fell from 4,107 and 4,361 in 1918 and 1919 to 1,323 and 1,253 in 1920 and 1921; in Metallurgy, accidents fell from 2,126 and 1,744 in 1918 and 1919 to 863 in 1920; and, in the case of Chemistry, accidents fell from 677 and 758 in 1917 and 1918 to 383, 438 and 57 in 1919, 1920 and 1921 respectively.[10]

In any event, according to the weight of each industry in the total of accidents shown in Table 3, it is also true that the distribution of accidents among industries remained very stable along the whole period. In Table 3 we have divided the first third of the twentieth century in two sub-periods, delimited by the legal change affecting the registration system introduced in 1922. The correlation between the two periods is very high, 0.91. In conclusion, and in accordance with Tables 2 and 3, we find that Building, Metals (in fact, Iron; see Table 3 and note to Table 2), Mining, Chemicals and Transport, tended to be the industries with the highest level of accidents.

International comparisons on industrial accidents are also very complicated. The meaning of an accident is not always the same, varying according to, for instance, its severity, the "scope" of the workplace (including, or not, accidents in itinere), and, in the case of fatalities, the period passing between the accident and death.[11] In spite of these (and other) problems, Table 4 suggests that, in general, Spanish accident rates did not differ considerably from those of other countries.

III. OCCUPATIONAL SAFETY AND HEALTH POLICIES IN SPAIN

With the remarkable exception of two nineteenth-century laws on female and child labour, most labour legislation in Spain was enacted between 1900 and 1936. From a comparative perspective, the delay of Spain in adopting social reform during the 19th century is clear according to the recent two indexes for seventeen European countries elaborated by Huberman and Lewchuck (2003). During the first third of the twentieth century a number of reforms were introduced, with the Instituto (1903-1924) and the Ministry of Labour (from 1920) being the main public institutions that coordinated state intervention in labour markets.[12]

The question of industrial accidents featured in Parliamentary and extra-Parliamentary debates from the late 1880s, due to the situation described by oral informants -social reformers, workers, unionists, employers and doctors.- included in the report conducted by the Comisión de Reformas Sociales (Commission for Social Reforms), the body which preceded the Instituto, with the aim of studying working class conditions. However, it was not until 1900 when the first accident-related legislation was enacted. The Labour Accidents Law (1900) included the setting of safety standards, the replacement of employer negligence liability theory with employer objective (no-fault) liability theory, the introduction of wage (postaccident) compensation, and also the introduction of voluntary (and state-subsidized) affiliation to an insurance company (private or employers´ mutual).[13]

Workers´ compensation for fatal accidents before 1933 varied between seven months and two years, depending on the number of family members. Compensation for nonfatal accidents, far more abundant than fatal accidents, depended on the kind of disability. Permanent disabilities were separated into three categories: total (i.e. full paralysis), professional (the inability to do the usual job), and partial (i.e. loss of a leg), which were remunerated with two, one and a half, and one years´ wage respectively. Temporary disability compensation (i.e. a broken arm) was 50 per cent of the wage before 1922, and 75 per cent of the wage between 1922 and 1933. [14] Despite the 1922 reform of the law (see below), it was not until 1933 when workers´ compensation system was seriously transformed, following Spain’s adhesion to the ILO Geneva Treaty (1925). This adhesion implied the implementation of the mandatory affiliation insurance system, the introduction of life annuities, and the increase in insurance premiums.

One of the main problems of the Labour Accidents Law could have been its implementation, in particular before the 1922 reform. In 1906 the Labour Inspection Service was created with the aim of enforcing labour laws. However, the compliance with labour laws in general, and accident law in particular, was subordinated to the opposition of different social groups and a lack of funding. Labour inspectors’ complaints about the implementation of labour regulations were serious in tone, as included in their annual reports (Memoria(s) General(es) de la Inspección del Trabajo, 1908-1930). First, they noted the lack of personnel and the difficulties of covering all establishments (with these sometimes being very dispersed and inaccessible). Second, they demanded more coordination between central and local public administration levels and warned of local employers’ influence on local delegations of the Instituto. Third, the imposition of the fine process was slow, often lasting years, and the instructions given to inspectors put persuasion before penalization. Moreover, remissions were not unusual, and on other occasions employers could easily avoid payment by declaring themselves bankrupt.[15]

The result was, first, a weak implementation of OSH standards and, second, the existence of inefficiencies in the workers compensation system. Labour inspectors themselves admitted great difficulties in implementation of the new, and sometimes expensive, safety standards in a economic context mainly constituted by hardly capitalized small and medium firms.[16] Labour inspectors also reported a considerable volume of delays and failures in processing compensations. On other occasions, workers refused to claim compensations, due to ignorance of the regulations and fear of reprisals, or came to agreements with employers in order to hide partial disabilities (Soto, 1989, pp. 675-677). Furthermore, it seems that a number of accidents were not finally considered as industrial accidents before 1922, due to the fact that the law was very imprecise (García-Ormaechea, 1935, pp. 13-18; Soto, 1989, pp. 636-637; Martínez-Pérez, 1992, pp. 354-356).

This situation appears to be different from 1922 onwards because of, first, the compliance of Spain with several ILO agreements on OSH. Thus, the recording of accidents was improved and the inspection system was reinforced. According to data provided by the Memorias, inspections rose from the 7,277 firms visited in 1908, to 35,505 in 1922, before reaching 77,297 in 1930. Moreover, the Código de Trabajo (Labour Code) of 1926, a specific law on employment, explicitly refers to the role played by employers, making compulsory the implementation of OSH steps and the use of specialized and appropriate machinery and personnel for risky tasks.[17] Second, and following other European countries during the 1920s, Spanish OSH policy partially shifted its orientation from the traditional European focus on the compliance to legal safety standards, to the adoption of American OSH policies focusing on educational actions related to prevention steps and the reinforcement of economic incentives.[18] In particular, the system of payments related to workers´ compensation was reinforced (Soto, 1989, pp. 636-638, 675-683).

In any event, some problems remained. First, it is true that the 1922 reform of the Accidents Law reinforced labour inspection and gave inspectors more capacity to impose fines (Soto, 1989, pp 281-286). However, educational actions and persuassion related to OSH steps were still the main priorities, and the imposition of fines when serious violations to labour laws were found continued being very low.[19] Second, in spite of the changes introduced in the law in 1922 and 1926, the confusion over the definition of industrial accident persisted. Court cases increased and workers received, at best, smaller premiums (García-Ormaechea, 1935, pp. 18-32; Soto, 1989, pp. 637-638; Martínez-Pérez, 1992, pp. 356). Workers´ compensation system presented another serious inefficiency. In 1922, the state created a reserve fund with the aim to assure workers´ payments in case of employer´s bankrupt. However, because of the lack of funding, this fund did not came into operation until 1933 (Jordana de Pozas, 1933, pp. 3, 19).

The role of Spanish unions in the prevention of accidents does not seem to be particularly relevant in the period of industrialisation. Spanish workers mainly affiliated to or supported two unions, the Socialist Unión General de Trabajadores (UGT) and the Anarchist Confederación Nacional del Trabajo (CNT). The former, founded in 1879, spread in services and industrial subsectors like mining, metallurgy or transport, and was more prone to collaborate with employers and public labour and social reform institutions. The latter, founded in 1910, strengthened in agriculture, textiles and building, and held a more revolutionary and anti-capitalist attitude. The exhaustive study carried out by Soto (1989) on the history of industrial labour includes some examples of union concern about accidents at local level. However, the analysis of the motives for strikes and other qualitative labour history studies suggest that the demand for better workplaces was not a priority for Spanish unions during their early years.

Rather, according to the Estadística de las Huelgas (Strike Statistics) Spanish unions seemed to focus on wage demands, since around 45 per cent of the strikes between 1905 (the first year available) and 1935 were motivated or included wage claims. During the period under study, union participation in the design of labour policy in Spain was very limited. Not only was the case that Anarchist unionism rejected (not without internal disagreement) integration in the new labour institutions, but there was also a lack of opportunities granted by successive governments.[20] Thus, although employers were afraid of an increase in workers´ control caused by the appearance or strengthening of labour institutions set up to manage conflict, such as the Instituto de Reformas Sociales, the Instituto Nacional de Previsión or, finally, the Ministry of Labour, in fact, wage levelling and hours reduction (before achieving the eight-hour day in 1919) were the main union claims in a context of limited power.[21]

IV. MODEL AND DATA FOR SAFETY EQUATIONS

With the aim of testing whether labour market institutions and institutional change had a significant impact on the level of industrial accidents, we have adapted to our data the standard model used in OSH literature. Previous studies usually focus on only one or two institutions, namely OSH inspection intensity, workers´ compensation, or the impact of unions. Following econometric estimations by Fishback (1986, 1987), we consider the impact of the three main labour market institutions affecting the level of accidents by using, in this case, data for industries. We estimate panels of industry averages for two periods, from 1909 to 1920, and from 1923 to 1930.[22] The first corresponds to the period in which OSH policies have been thought to be inoperative; the second corresponds to the period in which OSH policies have been thought to be reinforced. The sources and descriptive statistics of all the variables are included in Tables 5 and 6. The equation for both periods takes the following form:

ACCIDENTSit = (0 + (1 INSPECTIONit-1 + (2 COMPENSATIONit + (3 UNIONSit

+ (4 ACCIDENTSit-1 + ( INDUSTRY + ( TIME +(it

With regard to the dependent variable, researchers model both accident rates and accident counts. Accident rates, usually accidents per thousand workers, can be difficult to construct, especially using historical sources. In our case, data on the active population at national level and for some industries are only available for census dates (1910, 1920, 1930). Therefore, we are forced to use accident counts instead of accident rates. In any event, other things being equal, factors determining a count will also tend to determine the rate.

As several researchers have pointed out, ordinary least squares can present statistical shortcomings when estimating accident equations since the dependent variable can take only non-negative integer values and many (in this case) industries can report no accidents in one or more years (Ruser, 1991; Gray and Jones, 1991a; Boal, 2003). The common solution is to use Poisson and negative binomial models suitable for analysing count data. However, the Poisson model assumes that the conditional mean of the dependent variable equals its conditional variance. Since our data present overdisperion (variance greater than mean), we use the negative binomial, which allows the mean and variance to differ.[23] In any event, we also include a log-linear OLS model in order to compare. Following Gray and Jones (1991a, 1991b), in the first panel we use log (accidents + 1) instead of log (accidents), since the dependent variable can be zero.

The first of our dependent variables (INSPECTION) measures the impact of OSH regulation through a proxy of the intensity of labour inspection, i.e. the number of infractions to the Labour Accidents Law in each industry. We take into account that the labour inspection did not equally cover all the industries, and therefore we divided the number of infractions per the number of inspections in each industry. Since there is no information on fines imposed to firms or industries, this variable is the only available proxy to the intensity of prevention policies.[24] We introduced the one-year lagged variable because it is plausible to assume that the impact of the enforcement of the law on accidents occurs with a lag involved in making capital and labour investments decisions related to safety standards (Viscusi, 1992; Lanoie, 1992).[25] The predicted sign of the coefficient of this variable should be negative. However, according to qualitative evidence provided by labour inspectors, social reformers, and historians, the impact of this variable on the reduction of accidents prior to the 1922 Accident Law reform could be very weak.

As we know for previous studies, the extension of workers´ compensation (COMPENSATION) can affect the level of accidents. The sign of the coefficient of this variable would depend on what kind of prevention incentives dominates. As Fishback and Kantor (2000, p. 78) summarize, accidents tend to fall in industries where employer prevention increases more than worker prevention declines, and vice versa. Researches insist that the net effect of a change in benefits on accidents is ambiguous and can only be resolved empirically (for instance, Ruser, 1991; Lanoie, 1992; Fishback and Kantor, 2000). In our case, the prediction is, in fact, difficult. Before the 1922 reform, we find a lack of incentives to alter the number of accidents by both parties. On the one hand, labour inspectors reported an insufficient safety prevention effort on the part of employers. On the other, according to the law, employees played a minor role in the process of accident registration. The situation could have changed after 1922, but the predicted sign of the coefficient remains ambiguous. On the one hand, the reform increased prevention policies (although penalization remained very low). On the other, reform forced both parties to report any kind of accident (although inefficiencies in the system of workers´ compensation payments continued).

In order to approximate the generosity of compensation policies, in this case and taking into account the lack of better data, we follow a procedure based on average industry wages and a classification of industries according to the level of risk. This classification was elaborated by the Spanish Ministry of Labour because of the implementation of mandatory affiliation insurance system and the introduction of life annuities in 1933. Official premiums, i. e. percentages of the average industry wage, were established to be paid by employers for class of industries based on the level of risk. Secondarily, in the case of the non-existence of proper estimates of risk, premiums were based on firms´ adoption of safety steps such as the existence of specialized personnel, the maintenance of machinery, the size, ventilation and lighting of the establishment, the existence of medical services and the availability of a factory safety regulations manual (Instituto Nacional de Previsión, 1940, pp. 3-5). Although imperfect and launched in 1933, at the end of the period under study, this classification offers us a detailed description of the risk map, that we use to classify industries during the years 1909-1930. Therefore, we have multiplied these percentages by industry wages to obtain an approximation to the level of workers´ compensation.

The union variable (UNIONS) reflects the role of unions in preventing accidents. According to social and political historians, Spanish unions did not focus on OSH demands, since their participation in labour policy was limited both by their reluctance and by successive governments, and their demands were concentrated in wage levelling and hours reduction. That is the reason the relationship between union power and the reduction of accidents should be very weak.[26] The lack of continuous and disaggregated union density data forces us to use the other traditional proxy for labour pressure, namely strikes. In any event, given the importance of non-affiliated anarchists in the Spanish labour movement, strikes might be a better indicator of labour pressure. The three usual shape indicators of strike activity defined by Shorter and Tilly (1974, pp. 306-309) are frequency (strikers / active population), size (strikers / strikes), and duration (lost-workday / strikers). Because of the lack of annual data for active population at industry level, we have used a variant of strike frequency compounded by the number of strikers per workers at establishment. The result using strike size was very similar. Duration is not available at industrial level.

Following previous works, we also add a lagged dependent variable as a proxy for the safety conditions inheredited from the previous period (Bartrip and Fenn, 1988; Viscusi, 1992; Lanoie, 1992). In any event, results without this variable were practically the same. Finally, we include fixed effects to control for unobserved differences across industries. These variables are necessary since there is no systematic information on industry characteristics susceptible to affect the level of accidents, such as women density, average age, human capital, hours of work, productivity, or establishment size.[27] Year fixed effects capture the impact of differences across time that are common to all industries.

V. EMPIRICAL RESULTS

Tables 7 and 8 show the main results. We add institutional variables one by one, controlling for the lagged dependent variable and industry and time dummies in all the cases. The signs and significance of institutional variables when these were added in different orders were the same. Table 7 refers to the pre-reform situation. The inspection rate is, as expected, negative, but it is only significant in NB regressions. The coefficient of the compensation variable is not significant, suggesting that workers´ compensation policies did not have a substantial influence on the level of accidents. This result is, in fact, not surprising, since the level of postaccident compensation and the efficiency of the system of payments appears to be very low in this period. On the other hand, although the t and z statistics are small, the positive impact of the coefficient suggests that workers incentives to reduce prevention tended to dominate over employers incentives to increase safety. The union variable is not significant, as expected to quantitative and qualitative evidence. Finally, the coefficient of the lagged dependent variable is positive, significant, and powerful. This result coincides with previous analyses and, as according to Viscusi (1992, p. 13) and Lanoie (1992, p. 648), the interpretation is that, because of the capital-intensive nature of safety investments, firms that are risky are likely to remain so.

The regressions in Table 8 refer to the post-reform situation. In this case, the inspection rate is significant in OLS and NB specifications. In spite of the fact that the t and z statistics are bigger, the compensation variable is still not significant, but its sign has changed. Here, it is suggested that, because of the enforcement of the system of payments and the adoption of safety steps, employers prevention incentives tended to dominate over workers incentives to reduce safety. Unions remained inactive in reducing accidents. In any event, if we only consider the sign of the coefficients, in this case the positive relationship, although extremely weak, might reflect a greater propensity on the part of the unions to report accidents in order to receive worker compensation. That is to say, and since in this period the system of payments was reinforced, to counteract employers incentives to reduce accidents. Finally, the lagged dependent variable reduces its impact considerably during this period. In this sense, we have to take into account that new safety programs promoted constant changes in firms´ investment strategies on the safety issue from the early 1920s and throughout the whole decade.

VI. CONCLUSIONS

Industrialisation processes during the nineteenth and early twentieth centuries generated a demand for social reform. One of the areas of state intervention was Occupational Safety and Health (OSH). At the same time, European unions expansion from the last quarter of the nineteenth century, provoked an intensification in demands, such as wage increases, hours reductions, work methods or working conditions. In this study, we have considered the case of Spanish industrialisation in order to analyse the impact, according to safety literature, of the three main institutional factors affecting the level of accidents at work, namely safety legislation, economic incentives affecting both employers and workers, and the role played by unions.

In table 9, according to NB regressions, we have estimated the impact of safety legislation on the reduction of accidents. Ours results show that the effect of the 1922 reform of safety legislation was modest. Evaluated at mean, the estimated impact of increasing inspection on the reduction of accidents rose from 1 to 4 per cent. Evaluated at standard deviations (one and two), the reduction rose from around 5 and 10 to 7 and 13 per cent. In any event, the reinforcement of the law was incapable of counteracting the economic upturn during the 1920s. A combination of motives has been identified. First, the lack of funds and personnel; second, the weakness of labour inspection to enforce the law and impose fines.[28] In fact, as was remarked by labour inspectors themselves, underlying these problems was the great difficulty in implementing a new and wide-ranging set of early labour laws. Thus, if at the time of the First World War Spain was still one of the countries with the lowest levels of legislation in terms of social and labour intervention, at the eve of the Civil war (1936) more than twenty laws had been enacted, putting Spain on the European average. Furthermore, the Spanish industrial sector was at that time mainly constituted by low capitalised small and medium-sized companies, with establishment size being one of the main factors in determining the success of safety legislation -as remarked by a number of safety studies.[29]

Our results also show the limited impact of economic incentives on reducing accidents. Workers´ compensation for accidents was established in 1900. However, social reformers pointed out that compensation was low, and labour inspectors reported agreements between employers and workers in order to avoid the registration of accidents, as well as abundant delays and failures in processing compensation claims. The 1922 reform of the law affected the system of payments and workers were better post-compensated, but problems persisted and the impact of employers´ reaction on safety was weaker than the impact of legislation.

Finally, we cannot find any notable impact of unions in reducing accidents. Spanish unions were obviously aware of working conditions. However, Spanish unions had little participation, either forced or voluntary, in the design of labour policy during those years. In a period in which unions were struggling to gain political and social recognition, their focus was on “primary” demands like wage and hour demands, but not working conditions.

REFERENCES

Adnett, N. and Dawson, A. (1998): "The economic analysis of industrial accidents: a re-assessment", International Review of Applied Economics, 12, 2, pp. 241-249.

Aldrich, M. (1997): Safety First. Technology, Labor and Business in the Building of American Work Safety, 1870-1939, Baltimore, The John Hopkins University Press.

Aries, P. (1998): “Inspection du travail et inspection ouvrière dans le discours syndical, de la genèse de l´institution à l´entre-deux-guerres”, in J.L. Robert (dir.), Inspecteurs et inspection du travail sous la IIIe et le IVe République, Paris, Ministère de l´emploi et de la solidarité, pp. 52-59.

Bartrip, P. W. J. and Burman, S. B. (1983): The Wounded Soldiers of Industry. Industrial Compensation Policy, 1833-1897, Oxford, Clarendon Press.

Bartrip, P. W. J. and Fenn, P. T. (1988): "Factory Fatalities and Regulation in Britain, 1878-1913", Explorations in Economic History, 25, pp. 60-74.

Blank V. L. G.; Diderichsen F. and Andersson, R. (1996): "Technological Development and Occupational Accidents as a Conditional Relationship: A study of Over Eighty Years in the Swedish Mining Industry", Journal of Safety Research, 27, 3, pp. 137-146.

Boal, W. M. (2003): “The effect of Unionism on Accidents in Coal Mining 1897-1929”, Drake University, Working Paper ().

Byrne, J. (1992): "`Nuestro pan de cada día: accidentes de trabajo y respuesta de los albañiles de Madrid en el cambio del siglo", in R. Huertas and R. Campos (eds.), Medicina social y clase obrera en España (siglos XIX y XX), Vol. I, Madrid, FIM, pp. 21-48.

Calle, de la Mª D. (1992): “Instituto de Reformas Sociales: Higiene y seguridad en el trabajo”, in R. Huertas and R. Campos (eds.), Medicina social y clase obrera en España (siglos XIX y XX), Vol. I, Madrid, FIM, pp. 245-261.

Cohen, A. y Ferrer, A .(1992): “Accidentes y enfermedades profesionales de los mineros: realidad y derecho”, in R. Huertas and R. Campos (eds.), Medicina social y clase obrera en España (siglos XIX y XX), Vol. I, Madrid, FIM, pp. 215-244.

Cuesta, J. (1984): “1922: Reformismo imposible ante un régimen en quiebra”, Studia Historica, 22, 4, pp.

Dorman, P. (1996): Markets and mortality. Economics, dangerous work and the value of human life, Cambridge University Press, Cambridge.

Fabiano, B.; Parentini, I.; Ferraiolo, A. and Pastorino, R. (1995): "A century of accidents in the Italian industry: Relationship with the production cycle", Safety Science, 21, pp. 65-74.

Fairris, D. (1998): "Institutional Change in Shopfloor Governance and the Trayectory of Postwar Injury Rates in U.S. Manufacturing, 1946-1970", Industrial and Labour Relations Review, 51, 2, pp. 187-203.

Fishback, P. V. (1986): "Workplace Safety during the Progressive Era: Fatal Accidents in Bituminous Coal Mining, 1912-1923", Explorations in Economic History, 23, 3, pp. 269-298.

Fishback, P .V. (1987): “Liability Rules and Accident Prevention in the Workplace: Empirical Evidence from the Early Twentieth Century”, Journal of Legal Studies, XVI, pp. 305-328.

Fishback, P. V. and Kantor, S. E. (2000): A Prelude to the Welfare State. The Origins of Worker’s Compensation, Chicago, The University of Chicago Press.

Fortin, B. and Lanoie, P. (1998): “Effects of Worker´s Compensation: A Survey”, Université Laval, Cahiers de Recherche nº 9816.

Fraser, D. (2003): The Evolution of the British Welfare State. A History of Social Policy since the Industrial Revolution (third edition), Basingstoke, Palgrave.

García-Ormaechea, R. (1935): Jurisprudencia del Tribunal Supremo y de la Comisión Superior de Previsión sobre accidentes de trabajo, 1902-1934, Madrid, Sobrinos de la Sucesora de M. Minuesa de los Ríos.

González-Sánchez, J. J. (1997): Seguridad e Higiene en el Trabajo. Formación histórica y fundamentos, Madrid, Consejo Económico y Social.

Gray, W. B. and Jones, C. A. (1991a): "Are OSHA health inspections effective? A longitudinal study in the manufacturing sector", The Review of Economics and Statistics, LXXIII, 3, pp. 504-508.

Gray, W. B. and Jones, C. A. (1991b): "Longitudinal Patterns of Compliance with Occupational Safety and Health Administration Health and Safety Regulation in the Manufacturing Sector”, The Journal of Human Resources, 26, No 4, pp. 623-653.

Heyde, L. ([1920] 1931): Compendio de Política Social, Barcelona, Editorial Labor.

Huberman, M. y Lewchuk, W. (2003) “European Economic Integration and the Labour Compact, 1850-1913”, European Review of Economic History, Vol. 7, No. 1, pp. 3-41.

Instituto Nacional de Previsión (1940): Tarifas de primas para el seguro de accidentes del trabajo (riesgos de incapacidad permanente y muerte), Madrid, Caja Nacional del Seguro de Accidentes del Trabajo.

International Labour Organization (1923): “Industrial Safety. Industrial Accidents in Sweden in 1918 and 1919”, International Labour Review, 8, 3, pp. 437-440.

International Labour Organization (1924): “Some Problems of Factory Inspection”, International Labour Review, 9, 3, pp. 372-386.

International Labour Organization (1928): “Industrial Accidents in Russia”, International Labour Review, 8, 5, pp. 719-721.

Jordana de Pozas, L. (1933): La Caja Nacional de Seguro de Accidentes del trabajo y sus primeros resultados, Madrid, Instituto Nacional de Previsión.

Kniesner, T. J. and Leeth, J. D. (2000): “Workplace Safety Policy: Past, Present and Future”, Center of Policy Research, Maxwell School of Citizenship and Public Affairs, Syracuse University, Policy Brief nº 19.

Lanoie, P. (1992): "The Impact of Occupational Safety and Health Regulation on the Risk of Workplace Accidents. Quebec, 1983-1987", The Journal of Human Resources, XXIII, 4, pp. 643-660.

Martínez-Pérez, J. (1992): "La salud laboral en la II República: la actitud de los médicos ante la Ley de Accidentes de Trabajo en la industria", in R. Huertas and R. Campos (eds.), Medicina social y clase obrera en España (siglos XIX y XX), Vol. I, Madrid, FIM, pp. 349-369.

Marvaud, A. ([1910] 1975): La cuestión social en España, Madrid, Ediciones de la Revista de Trabajo.

McIvor, A.J. (2001): A History of Work in Britain, 1880-1950, Hampshire, Palgrave.

Montero, F. (1981): “La polémica sobre el intervencionismo y la primera legislación obrera en España, 1890-1900. 2ª parte: el debate político-parlamentario”, Revista de Trabajo, 61-62, pp. 35-89.

O’Toole, M. F. (2001): “The relationship between employees’ perceptions of safety and organizational culture", Journal of Safety Research, 33, pp. 231-243.

Paris H. (1935): Estudio actuarial y social de los Seguros obligatorios vigentes en España, Madrid, Sobrinos de la Sucesora de M. Minuesa de los Ríos.

Prados de la Escosura, L. (2003): El progreso económico de España (1850-2000), Madrid, Fundación BBVA.

Ramm, T. (1986): "Laissez-faire and State Protection of Workers", in B. Hepple (ed.), The Making of Labour Law in Europe. A comparative Study of Nine Countries up to 1945, London, Mansell Publishing Limited, pp. 73-113.

Robinson, J. C. (1988): "The Rising Long-Term Trend in Occupational Injury Rates", American Journal of Public Health, 78, 3, pp. 276-281.

Rodgers, D. T. (1998): Atlantic Crossings. Social Politics in a Progressive Age, Cambridge, Massachusetts, Harvard University Press.

Ruhm, C. J. (2000): “Are recessions good for your health?”, The Quarterly Journal of Economics”, 115, 2, pp. 617-650.

Ruser, J. W. (1991): "Workers´ Compensation and Occupational Injuries and Illnesess", Journal of Labor Economics, 9, 4, pp. 325-351.

Samaniego, M. (1992): “La representación obrera en el Instituto de Previsión”, in R. Huertas and R. Campos (eds.), Medicina social y clase obrera en España (siglos XIX y XX), Vol. II, Madrid, FIM, pp. 493-525.

Sand, R. (1923): “Health and the Workers”, International Labour Review, 7, 4, pp. 507-514.

Shorter, E. and Tilly Ch. (1974): Strikes in France, 1830-1968. Cambridge, Cambridge University Press.

Silvestre, J. (2002): “Los determinantes de la protesta obrera en España, 1905-1935: ciclo económico, marco político y organización sindical, Revista de Historia Industrial, (forthcoming).

Silvestre, J. (2004): Wage Compensation for Workplace Disamenities during Industrialisation: The Case of Spain, 1909-1920. Unpublished paper.

Smallman, C. (2001): “The reality of ‘Revitalizing Health and Safety’”, Journal of Safety Research, 32, pp. 391-439.

Soto, A. (1989): El trabajo industrial en la España contemporánea (1874-1936), Barcelona, Anthropos.

Timmer, A. and Williamson, J. G. (1998): “Immigration Policy Prior to the 1930s: Labor Markets, Policy Interactions, and Globalization Backlash”, Population and Development Review, 24, 4, pp. 739-771.

Viscusi, W. K. (1992): Fatal Tradeoffs. Public and Private Responsabilities for Risk, Oxford University Press, Oxford.

FIGURES AND TABLES

TABLE 1. Accidents and economic variables, 1909-1930

Dependent variable Log Log Log Log Log Log

Fatal Fatal Nonfatal Nonfatal Total Total

Intercept -3.5384 5.0964** -3.5351* 10.5858** -3.5089* 10.5901**

(-1.45) (33.06) (-1.99) (76.46) (-1.98) (76.52)

Log GDP 2.2801** 3.6956** 3.6900**

(3.65) (8.15) (8.13)

Estimated

Unemployment -0.3295** -0.2634* -0.2636**

(-2.05) (-1.83) (-1.83)

R2 0.412 0.319 0.778 0.271 0.777 0.271

Notes: 1909-1930 is the period under further econometric study in Tables 7 and 8. * Significant for values of p < 0.10; ** Significant for values of p < 0.05. t-statistics between brackets. Dickey-Fuller test for Accidents and GDP was carried out. Both variables are integrated of order one. The Johansen Cointegration test was also computed accepting the hypothesis that the residuals are cointegrated. Following Timmer and Williamson (1998), Unemployment is estimated using the residuals with the changed sign of the regression of GDP on trend and trend squared.

Sources: Spanish Real GDP at factor cost (1958 = 100) is provided by Prados de la Escosura (2003), pp. 323-326. For Accidents at nation level, Estadística(s) de los Accidentes del Trabajo (Accidents Statistics).

TABLE 2. Accident rates (selected industries)

1910 1920 1930

Mining 70.9 25.9 107.3

Metals 88.9 18.9 114.9

Chemicals 64.9 11.6 97.5

Textiles 8.4 4.4 23.0

Building 17.6 17.2 143.3

Food 16.9 12.7 71.8

Graphic Arts 25.2

Clothing 0.1 0.9 3.2

Leather and Skin 35.9

Wood 31.8

Transport 51.9 22.9 87.7

Notes: Accidents (five year moving averages) per thousand workers at census dates. Metals = Metallurgy + Iron.

Sources: Estadística(s) de los Accidentes del Trabajo and Soto (1989), pp. 74, 91, 106, and 116.

TABLE 3. Accidents, percentage of the total. 1909-1934

1909-1922 1923-1934 1909-1934

Government 2.74 0.98 1.57

Public companies 0.17 0.15 0.15

Mining 15.67 9.65 11.67

Metallurgy 1.05 4.55 3.37

Iron 16.07 14.46 15.00

Chemicals 2.62 3.09 2.93

Tobacco 0.39 0.62 0.54

Textiles 4.00 3.65 3.77

Agricultural industries 1.25 2.55 2.11

Building 15.29 24.07 21.12

Electricity 1.56 2.08 1.90

Food 6.34 6.39 6.37

Graphic Arts 0.38 0.49 0.46

Paper 1.05 0.78 0.87

TABLE 3 (cont.)

Clothing 0.30 0.57 0.48

Leather and Skin 0.52 0.56 0.55

Wood 4.67 4.44 4.52

Transport 24.76 19.27 21.11

Furniture 0.69 0.95 0.86

Ornamentation 0.50 0.72 0.64

Total 100 100 100

Note: There is no information about 1925.

Sources: Estadística(s) de los Accidentes del Trabajo.

TABLE 4. Accident rates in various countries (selected industries)

Spain G. B. Russia Spain Sweden Russia

Moscow Leningrad Moscow Leningrad

1910 1912 1912 1912 1930 1919 1926 1926

Mining 70.9 159.6 107.3 165

Metals 88.9 61.3 100.0 75.2 114.9 118 193.5 95.0

Chemicals 97.5 80

Textiles 8.4 16.5 23.9 37.5 23.0 15 48.1 33.5

Building 143.3 66

Food 71.8 43

Graphic Arts 25.2 74

Leather and Skin 35.9 27

Wood 31.8 91

Transport 51.9 46.2 87.7 72

Notes: Accidents per thousand workers. For Spain, Metals = Metallurgy + Iron. Five year moving averages in the cases of Spain and Great Britain.

Sources: For Spain, Estadística(s) de los Accidentes del Trabajo and Soto (1989), pp. 74, 91, 106, and 116; for Great Britain, McIvor (2001), p. 120; for Russia, ILO (1928), pp. 719-420; for Sweden, ILO (1923), pp. 438-439.

TABLE 5. Definitions and sources of variables included in regressions

Accidents: Accidents at industry level. Obtained from Estadística(s) de los Accidentes del Trabajo.

Inspection: Infractions to the Labour Accidents Law per inspection at industry. Own calculation from the Labour Inspection General Reports (Memoria(s) General(es) de la Inspección del Trabajo, MGIT).

Compensation: Industry premium based on the level of risk, i. e. percentage of the industry wage, multiplied per average industry wage. Premiums established by the Ministry of Labour and obtained from Paris (1935), pp. 78-92, and Instituto Nacional de Previsión (1940). Annual wages for 1909-1920 are provided by the Memorias. For the period 1923-1930, there are only wages for 1925 and 1930 provided by the Estadística de los Salarios y las Jornadas del Trabajo (Wages and Hours of Work Statistics). We construct one period-wage compound by the average of wages in 1925 and 1930. Results using the 1925 and 1930 wages did not vary substantially.

Unions: A variant of frequency of strikes compounded by the number of strikers per workers at establishment. Obtained from the Estadística(s) de las Huelgas (Strike Statistics).

Industry

Dummies: Mining, Metallurgy, Iron, Chemicals, Textiles, Building (omitted), Electricity, Food, Graphic Arts, Paper, Clothing, Leather and skin, Wood, Transport, and Furniture.

Time

Dummies: 1909 and 1923 are the omitted variables for the periods of 1909-1920 and 1923-1930 respectively.

TABLE 6. Descriptive statistics for variables (not dummies) included in regressions

1909-1920 1923-1930

Mean S. D. Min. Max. Mean S. D. Min. Max.

Accidents 1712.0 2316.3 0 11352 8356.7 9669.6 479 47717

Inspection 0.37 1.95 0.02 25.50 0.20 0.34 0.004 2.14

Compensation 14.19 8.53 3.23 58.82 35.69 12.94 15.71 55.83

Unions 69.51 32.39 0 100 75.17 27.59 0 100

Notes: For 1909-1920, 180 observations (15 industries ( 12 years). For 1923-1930, 94 observations. There is no information about 1925. There is also no information about 11 cases between 1923 and 1930 (94 = 15 industries ( 7 years - 11).

TABLE 7. Determinants of industrial accidents, 1909-1920

Dep. variable: Log Accs. Log Accs. Log Accs. Log Accs. Accs. Accs. Accs. Accs.

OLS OLS OLS OLS Neg.Bin. Neg.Bin. Neg.Bin. Neg.Bin.

(1) (2) (3) (4) (5) (6) (7) (8)

Intercept 5.591** 5.593** 5.239** 5.292** 7.723** 7.737** 7.641** 7.652**

(5.83) (5.83) (4.67) (4.76) (40.04) (40.49) (17.85) (18.21)

Inspection -1 -0.004 -0.001 -0.002 -0.026** -0.025** -0.025**

(-0.17) (-0.06) (-0.08) (-2.11) (-2.04) (-2.04)

Compensation 0.027 0.027 0.006 0.006

(0.85) (0.84) (0.27) (0.27)

Unions -0.001 -0.0002

(-0.35) (-0.13)

Log Accs. -1 0.290** 0.290** 0.283** 0.284**

(2.17) (2.17) (2.17) (2.17)

Accs -1 0.0001** 0.0001** 0.0001** 0.0001**

(2.53) (2.51) (2.47) (2.47)

Industry

fixed effects yes yes yes yes yes yes yes yes

Time

fixed effects yes yes yes yes yes yes yes yes

Adjusted R2 0.778 0.777 0.777 0.775

Log-likehood -1313.99 -1313.62 -1313.57 -1313.57

N= 180

** Significant for values of p < 0.05

Notes: Accidents = fatal + nonfatal. White standard errors. For OLS regressions, t-statistics between brackets. For Negative Binomial regressions, z-statistics between brackets. 15 Industries: Mining, Metallurgy, Iron, Chemicals, Textiles, Building (omitted), Electricity, Food, Graphic Arts, Paper, Clothing, Leather and Skin, Wood, Transport, and Furniture.

TABLE 8. Determinants of industrial accidents, 1923-1930

Dep. variable: Log Accs. Log Accs. Log Accs. Log Accs. Accs. Accs. Accs. Accs.

OLS OLS OLS OLS Neg.Bin. Neg.Bin. Neg.Bin. Neg.Bin.

(1) (2) (3) (4) (5) (6) (7) (8)

Intercept 9.877** 9.858** 10.189** 10.189** 9.838** 9.953** 10.252** 10.223**

(28.58) (35.61) (22.04) (22.11) (71.74) (95.20) (40.47) (40.46)

Inspection -1 -0.187** -0.186** -0.186** -0.205** -0.205** -0.201**

(-2.71) (-2.67) (-2.57) (-3.41) (-3.41) (-3.23)

Compensation -0.008 -0.008 -0.008 -0.008

(-1.06) (-1.04) (-1.22) (-1.24)

Unions 0.000001 0.0003

(0.001) (0.36)

Log Accs. -1 -0.002 0.016 0.014 0.014

(-0.07) (0.50) (0.44) (0.43)

Accs -1 0.000005 0.000004 0.000004 0.000004 (0.54) (1.45) (1.41) (1.44)

Industry

fixed effects yes yes yes yes yes yes yes yes

Time

fixed effects yes yes yes yes yes yes yes yes

Adjusted R2 0.976 0.978 0.977 0.977

Log-likehood -743.88 -739.81 -739.72 -739.66

N= 94

** Significant for values of p < 0.05

Notes: Accidents = fatal + nonfatal. White standard errors. For OLS regressions, t-statistics between brackets. For Negative Binomial regressions, z-statistics between brackets. There is no information about 1925. There is also no information about 11 cases between 1923 and 1930 (94 = 15 industries ( 7 years - 11). 15 Industries: Mining, Metallurgy, Iron, Chemicals, Textiles, Building (omitted), Electricity, Food, Graphic Arts, Paper, Clothing, Leather and Skin, Wood, Transport, and Furniture.

TABLE 9. Estimated impact of increasing Inspection on the reduction of Accidents (percentage)

Partial derivatives; Regressions (6) 1909-1920 1923-1930

Evaluated at mean -1.0 -4.0

Evaluated at standard deviation -4.9 -6.7

Evaluated at two standard deviations -9.6 -13.0

Notes: Partial derivatives are calculated as [exp((1Inspection) - 1], where (1 is the coefficient obtained in regressions 6 for both periods (Tables 7 and 8), and Inspection takes different values (mean, standard deviations). Impact are expressed in percentage changes.

-----------------------

[1] This literature, including some relevant exceptions, is exhaustively commented on by Viscusi (1992), pp. 206-222. The study by Lanoie (1992) of the Quebec system of safety regulation also reports a little impact on the reduction of accidents.

[2] The empirical work on modern times reviewed by these authors shows different results depending on the inclusion of other factors that could affect workplace safety, such as experience rating (i.e., the establishment of premiums paid by employers according not only to industry classes, but also to firm specific accident experience), safety regulation and the abusive uses of the system by reporting false accidents or accidents occurring off the job.

[3] Also see O´Toole (1999) and the works cited in Boal (2003), pp. 1-2.

[4] Also see qualitative evidence provided by Bartrip and Burman (1983), pp. 51-53, and McIvor (2001), pp. 127-130.

[5] Although the main aim of the work of Blank et al (1996) on the Swedish mining industry (1911-1990) is to analyse the relationship between technological change and accidents, the authors find a significant impact of changes in safety laws in reducing accidents during the 1950s.

[6] This author refutes previous findings by Fishback (1986), who found a positive relationship between unionism and accidents.

[7] Bartrip and Fenn (1988), pp. 62-63, also note a change in the reporting requirements which may have influenced homogeneity in British figures.

[8] Regressions were also estimated on the GDP, or the estimated unemployment, and time and time-squared trends; coefficientes were between 1.7 and 2.9 in the case of GDP and between -0.5 and -0.6 in the case of unemployment. Estimations using one-year lagged economic variables yield slight differences in elasticities.

[9] Only for Mining, Iron and Metallurgy, active population can be also found in the Estadística Minera y Metalúrgica (Mining and Metallurgy Statistics).

[10] There are two potential motives for the decrease in accidents around 1920. First, the short-run impact of winning the eight-hour day in 1919. Second, and probably more relevant, the dramatic increase in lost working-days. Lost working-days rose from around 3,000,000 per year in 1916-1918 to 8,887,779 and 18,154,405 in 1919 and 1920; and fell to 4,486,550, 3,040,463, and 3,373,483 in 1920, 1921, and 1923 (Silvestre, 2002). A regression of the log of total accidents on the log of lost working-days (controlling for the log of GDP and a constant), yields an elasticity of -0.11 being the coefficient significant at the 10 per cent.

[11] Thus, annual harmonised statistics for the European Union countries have been produced since as recently as 1994. Another potential factor distorting comparisons is the system of monetary incentives and disincentives for having a work injury officially recorded as such.

[12] See Silvestre (2004) for the main interventions.

[13] Following the early British system, the Spanish system before 1933 led firms free to choose between insurance companies and their own reserve funds. See Rodgers (1998), pp. 245-249, for the differences between British (voluntary insurance) and German (compulsory insurance) systems. See Montero (1981), pp. 37-42, for the Spanish case.

[14] A number of social reformers, independent or legal and medical state counsellors, pointed out that premiums were insufficient (Silvestre, 2004).

[15] Abundant social, political, and labour history studies have stressed the failure of early Spanish state intervention in the labour market. See Silvestre (2004) for a review of this literature.

[16] Also see Soto (1989), pp. 643-646, Cuesta (1992), pp. 247-250, the comments (p. 180) in the Spanish edition of the Heyde´s Social Policy, translated into Spanish from the German edition in 1931, and Cohen and Ferrer (1992), pp. 217-219, for mining industries. See Gónzalez-Sánchez (1997) pp. 91-94, for a description of the new OSH rules.

[17] See Soto (1989), p. 660, and González-Sánchez (1997), pp. 101-105, for a detailed description of these changes.

[18] See Aldrich (1997), p. 132, for this change in Europe. See McIvor (2001), pp. 132-145, for a detailed description of the British case.

[19] The Memorias give the number of fines imposed because of violations to the entire set of labour laws. Between 1908 and 1917, only 0.15 per cent of violations to three laws, Women and Children Work Law, Sunday Rest Law, and Accidents Law, ended in a fine. Between 1918 and 1930 other violations were incorporated, the most relevant being the failure to observe the eight-hour day winning in 1919. For this period, the rate of infractions ended in a fine was 1.63 per cent.

[20] See Silvestre (2002, 2004) and abundant studies cited therein.

[21] The study by Byrne (1992) on building workers at the turn of the century points out how these workers internalised and accepted high levels of risks due to the lack of confidence in the law, fatalism, and virility culture. Marvaud (1910), Cuesta (1984), p. 86, De la Calle (1992), p. 247, and Samaniego (1992), pp. 502, 505-506, refer to relatively weak unions and workers involvement in OSH issues.

[22] Recent literature for modern times also can use firm data, but this approach is not possible in our study.

[23] We follow the test provided by EViews in order to check if the data presents overdispersion. Also see Ruser (1991), p. 336.

[24] See similar variables in Viscusi (1992), pp. 181-222, (based on previous works published in 1979 and 1986), Gray and Jones (1991a, 1991b), and Lanoie (1992).

[25] For both periods, 1909-1920 and 1923-1930, the impact of the non-lagged variable was not significant, except in the case of Negative Binomial regressions for 1909-1920, where the variable was still significant, but its size lower. Longer lag variables were not significant.

[26] On the contrary, the impact of unions on wages was positive and highly significant (Silvestre, 2004).

[27] Moreover, we consider that industry dummies incorporate employment size (Gray and Jones, 1991a, p. 505; 1991b, pp. 629-630).

[28] See Bartrip and Fenn (1998) for similar reasons in the British case (1878-1913).

[29] We should also consider employers´ and workers´ opposition: the former made strong complaints based on costs, whilst the latter expressed distrust and even reluctance to change traditional work methods.

-----------------------

[pic]

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download