Open Access

On the differential impact of the recent economic downturn on work safety by nativity: the Spanish experience

IZA Journal of Migration20132:4

DOI: 10.1186/2193-9039-2-4

Received: 5 October 2012

Accepted: 28 February 2013

Published: 18 April 2013

Abstract

Abstract

This paper explores differences in work injury and fatality rates between immigrants and natives and how they may have been impacted by the recent economic downturn. Our focus is on Spain over the 2001–2010 decade -a period of time during which Spain received one of the largest immigrant inflows of any developed economy and subsequently experienced a recession that has raised national unemployment rates above 20 percent. We find that immigrants worked in riskier jobs than natives during this high immigration period. Furthermore, the recession appears to have exclusively reduced job injury rates, but not fatality rates, among the average immigrant -hinting on their misreporting due to fear of dismissal as the primary cause for the observed decline. Overall, the figures are suggestive of work safety inequalities that may be important to address.

JEL codes

J61, J81

Keywords

Working Safety Injuries Fatalities Immigration Great Recession Spain

1. Introduction

Migration across national borders generates vigorous political and policy debates that are unlikely to diminish as the share of the world’s population residing in a country other than their country of birth rises. For instance, it is frequently argued that immigrants take jobs that natives do not want, such as more dangerous jobs. This popular belief is consistent with stylized facts for many countries, as well as for Spain, where the immigrant work injury rate, 5.03 percent, is above the 4.20 percent rate of natives (see Additional file 1: Table S1 for details). Yet, it remains unclear how the most recent recession may have impacted differences in work injury and fatality rates by nativity when present. The latter could have widened if immigrants have endured, overall, worse job prospects than natives and/or accepted riskier jobs. Alternatively, the aforementioned gaps could have narrowed if: (a) fear of dismissal and unemployment have reduced immigrants’ reporting behavior to a larger extent than that of natives, (b) workload reductions have been more prominent in sectors with a higher concentration of immigrants, or (c) selection of less accident-prone workers into employment has been more acute among immigrants than natives.

We examine work injury and work fatal accident rates among immigrants and natives in Spain over the 2001–2010 decade. We start by investigating whether, after accounting for a variety of factors potentially correlated to immigrants’ higher work injury rates –such as educational attainment, language proficiency, assimilation to the host country or industry and occupation of employment, we observe significant differences in work injury and fatality rates by nativity and region of origin. Unlike previous studies, we include data on legal as well as undocumented immigrants during an interesting decade that encompasses the immigration boom experienced from 2000 through 2008, as well as the most recent recession. Subsequently, we examine how the economic downturn has impacted work injury and fatality rates by nativity and region of origin, paying close attention at its likely determinants.

The focus on Spain is of special relevance given the purpose of the study and the time period being examined. During the 2001–2010 decade, Spain displayed one of the largest rates of immigration in the world –three to four times as large as the average immigration rate in the United States between 2000 and 2008. Just between 2003 and 2008, the foreign-born population four-folded and, by 2010, twelve percent of Spanish residents had a foreign nationality and 14 percent were foreign-born (Vasileva 2011).1 The large and rapid inflow of immigrants may have resulted in significant disparities in work safety by nativity for numerous reasons –such as immigrants’ lack of awareness of job risks, their need to get a job upon their immediate arrival, or their greater willingness to take on a riskier job (relative to natives) in exchange for a higher pay. Furthermore, Spain is one of the recent immigrant-receiving economies most hard hit by the latest recession. Unemployment rates have climbed to double-digits and currently hovered around 25 percent. Some risk prone industries with a higher concentration of immigrant workers, such as construction, have particularly suffered. Workload reductions, workforce composition biases and the pressure felt by more vulnerable and uninformed workers to misreport work injuries in order to avoid dismissal could have impacted work injury rates differently by nativity. Thus, Spain offers the ideal scenario to examine work injury and fatality gaps by nativity, as well as their evolution following the recent economic crisis.

This article is structured as follows. In the next section, we review the literature on work injuries and fatalities, focusing our attention on studies investigating differences by nativity or over the economic cycle. In section 3, we provide some background information on immigration to Spain and on the regulation of work injury and fatality rates in the country. Section 4 discusses the data and provides some interesting descriptive statistics on the evolution of work injury and fatality rates by nativity in Spain over the past decade. We then describe the methodology and discuss our findings in sections 5 and 6, respectively. Finally, section 7 concludes the study with a summary of our findings and some closing remarks.

2. Work injuries and fatalities by nativity and over the economic cycle

The literature on differences in work injury and fatality rates by nativity is quite extensive. It is often argued that immigrants may hold riskier jobs than natives for a variety of reasons ( Orrenius and Zavodny e.g. 2009). First, immigrants may have fewer job alternatives than natives. As a result, they may be more likely than natives to hold temporary jobs, which have worse work safety records than those of open-ended work assignments in Spain (Amuedo-Dorantes 2002). In that regard, Loh and Richardson (2004) argue that poor language ability and low educational attainment may limit many immigrants’ employment options. Alternatively, immigrants might have different knowledge and perceptions of job risks than natives due to their educational attainment, language proficiency or social capital (Marvasti 2010). For instance, immigrants may perceive work-related risks differently than natives if working conditions are generally better in the host country and do not perceive the job as particularly dangerous. In both cases, personal and human capital characteristics may account for differences in working conditions between immigrants and natives.

Second, from a compensating wage differentials framework in which riskier jobs pay more, immigrants might still occupy riskier jobs than natives because of differences in wealth or risk preferences. Immigrants may be more willing to take risky jobs if safety is considered a normal good and immigrants have lower incomes and wealth than natives. Alternatively, immigrants may be less risk averse than natives, as evidenced by the fact that they were willing to take on the risk of migrating (Berger and Gabriel 1991).2 These two facts may imply that immigrants are more willing than natives to trade off work safety in exchange for a higher pay as it is assumed in the hedonic equilibrium framework (Rosen 1986, Viscusi 1993).

A third possibility might have to do with the “healthy immigrant effect”. Immigrants tend to be healthier upon arrival than natives (Antecol and Bedard 2006). Therefore, it is conceivable that immigrants, particularly recent ones, might choose more physically strenuous jobs than natives.

Finally, immigrants and natives might have different safety-related productivities and abilities to benefit from safety training. Hersch and Viscusi (2010) suggest that immigrant workers as a group may impose higher safety-related costs because of language or cultural barriers. As a result, firms employing immigrants may have fewer incentives to invest in injury prevention (. Bauer et al 1999).

Despite all the aforementioned reasons for expecting higher work injury and fatality rates among immigrants, some studies provide conclusive evidence of natives having a higher work injury and/or fatality rate than immigrants, whereas others find no differences by nativity. For instance, in the United States –the country most widely studied (Ahonen and Benavides 2008), early studies found that immigrants generally endured lower work injury rates than natives (Berger and Gabriel 1991, Hamermesh 1998). However, more recent studies suggest that immigrants, especially Hispanics, endure higher work injury rates than natives (Loh and Richardson 2004, Leeth and Ruser 2006, Orrenius and Zavodny 2009. (). The evidence for other nations is rather scarce and varies widely from country to country. For instance, in Germany, Bauer et al 1999. () report no significant differences in the unconditional probability of enduring a less severe accident by nativity, but immigrants endure a higher probability of experiencing a severe work accident. In contrast, focusing on Spain and using data on legal immigrants who arrived to the country by 2001, Solé et al 2010. () conclude that, although legal immigrants are more likely to work in riskier jobs, they display a lower likelihood of becoming disabled. In that vein, Moral de Blas et al 2010) use data for 2005 and report that native workers in Spain have a higher rate of soft tissue injuries –a finding that they attribute to the false reporting of injuries. Focusing on Cataluña –a region in Northeast Spain, Diaz-Serrano (2010) also concludes that African immigrants work in riskier jobs than natives. In contrast, Ahonen and Benavides (2008) find that immigrants enjoy a lower risk of work injury and fatality using data from a sample of immigrants collected between September 2006 and May 2007 in 5 Spanish cities.

Aside from the mixed findings regarding differences in work injury and fatality rates by nativity, it remains unclear whether such differences would be exacerbated or narrowed during an economic downturn. Previous research examining workplace safety during economic cycles finds that work injuries are pro-cyclical. The rationale behind the pro-cyclical nature of work injury rates is that higher production requirements and work hours might increase stress and tiredness among workers, resulting in an increased work injury rate ( Kosssoris e.g. 1938, Shea 1990, Fairris 1998). Hence, during periods of high unemployment (or reduced economic activity), work injuries might decrease with the workload. Alternatively, injury and fatality rates may behave pro-cyclically due to changes in the composition of the workforce over the business cycle (Boone and van Ours 2006, Fahr and Frick 2007). Less accident-prone workers may be selected into the workforce when unemployment is high and, as such, fewer accidents may be recorded. The average workload may not have decreased, but the propensity to be involved in a work accident might have decreased. Finally, it is also possible for work injury rates to behave pro-cyclically and diminish during an economic downturn if workers are less likely to report injuries for fear of dismissal. As noted by Boone and van Ours (2006. () and Boone et al 2011), such a fear appears to be the main reason behind the observed pro-cyclicality after comparing the performance of work injuries as opposed to work fatality rates, which are not likely to be misreported.

Yet, to our knowledge, no previous study has explored the differential effect that the economic downturn may have had on the job risks faced by immigrants and natives –particularly in countries characterized by a large and recent immigration inflow and severely hit by the recession. Did the economic downturn reduce injury and fatality rates among both immigrants and natives as would be expected from an even reduction in workload across all sectors? Or did it only lower injury rates solely among immigrants suggesting other potential causes, such as misreporting, uneven workload reductions or, in particular, workforce composition biases?

In what follows, we merge industry and occupation work injury and fatality rates with individual level data from the Spanish labor force survey (i.e. Encuesta de Población Activa. (or EPA) for the 2001–2010 decade to learn about immigrant job segregation into riskier or safer jobs, and about changes in such nativity segregation during the past recession. The first part of this study is close in spirit to previous work by Solé et al 2010), who study differences in the probability of becoming disabled between immigrants and natives in Spain. The authors use cross-sectional data on severe work injuries and illnesses from the 2006 Muestra Continua de Vidas Laborales (MCVL) –a Social Security database that collects data on natives and legal immigrants. They also focus their attention on working age individuals who have contributed at least five years to the social security system –the minimum required to be eligible for a non-accident disability pension. Consequently, their study is informative of differences in permanent disability rates between natives and legal immigrants who arrived to the country prior to 2001.

We focus, instead, on work injury and work fatality rates, as permanent disability rates exclude deaths and their recording is likely conditioned on the legal status of the migrant. Additionally, since the immigration boom in Spain took place between 2000 and 2008, we look at the entire 2001–2010 decade. To include both legal and undocumented immigrants –a non-trivial share of the immigrant population allegedly more likely to endure worse working conditions than their legal counterparts,3 we use data from the Spanish labor force survey. The Spanish labor force survey is updated using the local population registers (Padrón Municipal). As noted by previous researchers ( Gonzalez and Ortega e.g. 2011), because registration in the Padrón allows for free educational and medical services, undocumented immigrants have an incentive to register. Finally, we examine for the first time how the recent economic downturn has impacted any differences in work safety by nativity and the likely explanations for such an effect.

3. Institutional framework

3.1. Background on immigration to Spain

Before proceeding any further, it is important to provide an overview of immigration to Spain and, in particular, its history and recent features. Until quite recently, Spain was a country of emigrants. However, the arrival of democracy in 1975, the entry of Spain in the European Union in the 1980s, the long-standing decline in Africa and the economic crises in several Latin American countries during the 1990s marked a sudden change. As noted in the Introduction, Spain has displayed one of the largest rates of immigration in the world since the year 2000. In 2001, the foreign-born population amounted to less than 1.4 million (Instituto Nacional de Estadística 2004). By 2008, it had five-folded, reaching 5.5 million (Instituto Nacional de Estadística 2012). In less than one decade, the foreign-born population had increased from 3.3 to approximately 14 percent of the population (Instituto Nacional de Estadística 2004, Vasileva 2011).

According to the official Spanish Statistical Institute, up to 52 percent of immigrants are male. On average, immigrants are younger than natives and have higher labor force participation rates (in the order of 73 percent compared to 57 percent in the case of natives) (e.g.. Reher et al 2008, Instituto Nacional de Estadística 2009). Fifty-nine percent of immigrants have secondary schooling and only 17 percent has tertiary education or a university degree. Available evidence also suggests that for 44.9 percent of immigrants Spanish is their native tongue, and 58.3 percent of those with a different native tongue consider themselves fluent in Spanish (Instituto Nacional de Estadística 2009). Only 14.5 percent of immigrants indicate not being able to speak the language.This new immigrant population is heavily concentrated in Madrid, the Mediterranean arc (i.e. Cataluña, Valencia, Murcia, and Andalucía), and the Balearic and Canary islands, and their origins are quite diverse. The vast majority of immigrants come from Latin America (39 percent), Europe (38 percent), and North Africa (17 percent) (Instituto Nacional de Estadística 2009). The most common countries of origin for immigrants are: Morocco, Romania, Ecuador, Colombia, the United Kingdom, and Colombia. Most Moroccans reside in Cataluña and Andalucía, Ecuadorians concentrate in Madrid, Cataluña and Murcia. People from the United Kingdom primarily reside in Alicante (Mediterranean arc) and Málaga (Andalucía), and half of Romanians reside in Madrid and Castellón (Mediterranean arc).

3.2. The Spanish national system of health and safety at work

The Law of Prevention of Labor Risks (November 31, 1995) regulates the Spanish National System of Health and Safety at Work, whose organizational structure is summarized in Additional file 1: Figure S1 (. Sessé et al 2002). The Labor Administration develops norms and legislation; trains and informs about risks; watches over the application of Spanish Safety Laws; applies sanctions; and processes official statistics on occupational accidents and diseases. The Health Administration focuses on the design of tools and systems that pursuit health at work, and trains sanitary personnel in health and safety in close collaboration with the Labor Administration. Finally, insurance organizations provide the mandatory work accident insurance. Some of the most popular insurance organizations include the Spanish Social Insurance Institute, which generally covers diseases, and the Mutual of Work Accidents, which covers temporary disability.

Companies are obliged to have in place preventive services, which can range from internal services provided by designated workers, prevention delegates or a health and safety committee, to external services, depending on firm size. It is compulsory to declare occupational accidents and disease cases, and inspection agents examine all accidents in order to establish the causes and consequences, and to initiate prosecution in the case of criminal negligence. The system is harmonized at the European level and, as shown by Table 1, the statistical data on the relative performance of the Spanish system vis-à-vis those of other European countries indicate a clear recent amelioration.
Table 1

Standardized incidence rate of fatal accidents at work by member state

Country

2000

2001

2002

2003

2004

2005

2006

2007

2008

EU15

2.8

2.7

2.5

2.5

2.4

2.3

2.4

2.1

:

Euro Area

3.2

3.1

2.9

2.9

2.7

2.5

2.8

0.5

:

Belgium

3.1

3.8

2.6

2.4

2.9

2.6

2.6

2.5

3.6

Denmark

1.9

1.7

2

1.8

1.1

2.2

2.7

2

1.7

Germany

2.1

2

2.5

2.3

2.2

1.8

2.1

1.8

1.6

Ireland

2.3

2.6

2.6

3.2

2.2

3.1

2.1

1.7

2.4

Greece

2.7

2.9

3.8

3

2.5

1.6

3.8

:

:

Spain

4.7

4.4

4.3

3.7

3.2

3.5

3.5

2.3

3.3

France

3.4

3.2

2.6

2.8

2.7

2

3.4

2.2

1.5

Italy

3.3

3.1

2.1

2.8

2.5

2.6

2.9

2.5

3.4

Luxembourg

6.8

1.7

2.4

3.2

:

2.6

1.7

:

2.8

Netherlands

2.3

1.7

1.9

2

1.8

1.6

1.7

:

1.6

Austria

5.1

4.8

5.1

4.8

5.4

4.8

4.2

3.8

4.2

Portugal

8

9

7.6

6.7

6.3

6.5

5.2

6.3

5.3

Finland

2.1

2.4

2

1.9

2.5

2

1.5

1.3

1.4

Sweden

1.1

1.4

1.2

1.2

1.1

1.7

1.5

1.4

1.5

Great Britain

1.7

1.5

1.4

1.1

1.4

1.4

1.3

1.3

0.6

Notes: Number of accidents which leads to the death of a victim within one year of the accident relative to the number of persons in employment in the reference population times 100,000.

Source: Eurostat.

4. Data

We use two sources of data in our analysis: (1) individual level data from the Spanish labor force survey (Encuesta de Población Activa, EPA) spanning from 2001 through 2010, and (2) data on work injury and fatality rates from the Workplace Accidents Statistics (Estadística de Accidentes de Trabajo, EAT) published by the Spanish Ministry of Labor and Immigration (Ministerio de Trabajo e Inmigración, 2001–2010).

The EPA provides the most representative and frequent data on the Spanish workforce. As noted by Gonzalez and Ortega (2011), relative to other Spanish surveys with employment information, the EPA more accurately captures the demographics of the foreign-born population, including undocumented immigrants. This is because Spain keeps a continuously updated population registry at the local level, which plays an important role in the sampling design of the EPA. All residents, regardless of legal status, are required to register and simultaneously have a strong incentive to do so since it grants them access to health and educational services and provides them with an official proof of residency in the country –a document later on needed to apply for legalization.4 As a result, the data on the foreign-born population in the EPA can be considered to be reasonably accurate and up-to-date. Our sample includes data from the second quarter of each year starting in 2001 and ending in 2010.5 Of particular interest to us is the information on the occupation and industry of employment, given at the three-digit level, as well as individual level characteristics, such as place of birth and the length of time they have resided in Spain. We define immigrants as foreign-born individuals. The EPA does not provide information on language proficiency. Nevertheless, we use information on the country of origin to identify immigrants for whom Spanish is their native tongue. Finally, the EPA asks foreign-born individuals about the year they arrived to Spain. We use that information to construct a measure of the duration of the migration spell.

Table 2 displays a few characteristics of individuals in our sample. We focus on working immigrants and natives in the labor force survey. Immigrants account for approximately 13 percent of the sample, even though their rates vary from 5 percent in 2001 to over 16 percent in 2010 as the immigrant population rose during the decade. Relative to natives, employed immigrants are more likely to be female, young, and non-married. They are also less likely than natives to have a university degree and more likely to have less than a primary education. Immigrants also seem more likely to hold a temporary contract and overall display shorter tenures than natives. On average, they came to Spain eight years ago and Spanish is the native tongue of approximately half of them. Finally, the vast majority of immigrants in our sample originate from Latin America, followed by other European countries, Africa and, lastly, Asia.
Table 2

Immigrant and native characteristics in the sample

Characteristics

Natives

Foreign-born

 

Mean

S.D.

Mean

S.D.

Male

0.60

0.49

0.56

0.50

Age

39.59

11.60

35.99

9.83

Married

0.60

0.49

0.55

0.50

Less than primary

0.03

0.17

0.05

0.23

Primary

0.17

0.38

0.17

0.38

Secondary

0.47

0.50

0.52

0.50

University degree

0.33

0.47

0.26

0.44

Temporary contract

0.22

0.42

0.43

0.50

Job tenure

121.54

124.21

42.50

62.74

Years in Spain

0.00

0.00

7.79

7.98

Spanish as native tongue

1.00

0.00

0.46

0.50

Africa

0.00

0.00

0.14

0.35

Asia

0.00

0.00

0.03

0.18

Europe

0.00

0.00

0.34

0.47

Latin America

0.00

0.00

0.48

0.50

Other origin

0.00

0.00

0.01

0.07

Notes: Observations are weighted using the individual weights in the EPA. The sample includes only individuals aged 16 and older who are employed in the private or public sector, except for military personnel, and not self-employed.

Source: EPA (2001–2010).

Aggregate statistics on the number of work injuries and fatalities according to different classifications are published by the Spanish Ministry of Labor and Immigration. We use four different data series from the EAT: (1) the number of work injuries at the industry level, (2) the number of work injuries at the occupation level, (3) the number of fatalities at the industry level, and (4) the number of fatalities at the occupation level. Work injuries include trivial as well as severe accidents leading to at least one day of work absence and exclude commuting accidents. We merge the aggregate work injury and work fatality rate time series to the individual labor force survey data by industry and occupation. Industry is coded in both data sources using the Spanish version of the NACE (Rev. 1 and Rev. 2),6 whereas occupation is coded using the Spanish version of ISCO-88 (COM).7 The data are merged at the two-digit level industry and occupation level –the most detailed level at which data on work injury and fatality figures are made available.8 We then calculate industry and occupation injury, as well as fatality, rates.

Table 3 summarizes the aforementioned rates. By industry, work injury rates are higher in mining, extractive industries and utilities, followed by construction, manufacturing and machinery/transportation equipment. Work fatality rates, however, are the highest in transportation and warehousing, followed by construction, mining and extractive industries, and agriculture. Of those industries, construction, followed by agriculture, are the ones exhibiting a higher concentration of immigrants.
Table 3

Injury and fatality rates by major industry and occupation groups

By Industry/Occupation categories

Injury rate

Fatality rate

Share of Foreign-born

 

(per 10,000)

(per 100,000)

 
 

Mean

Std. Dev.

Mean

Std. Dev.

(%)

Major industry categories

 

Agriculture, forestry, fishing, and hunting

361.29

112.80

7.85

9.64

15.12

Nondurable goods and wood products manufacturing

590.85

267.72

3.96

2.09

10.64

Mining and oil and gas extraction, utilities, and metal products

954.11

549.99

9.35

6.87

8.50

Machinery, electronic products, and electrical and transportation manufacturing

548.08

185.59

3.53

2.64

7.59

Construction

959.74

218.45

11.72

2.30

19.36

Wholesale and retail trade, accommodation and food services

390.98

120.24

2.31

2.24

15.42

Transportation and warehousing and Information

429.11

187.79

12.27

9.79

10.47

Financial activities, professional and business services

249.71

188.86

2.45

1.64

9.58

Education and health services and public administration

232.82

132.89

1.66

1.49

6.23

Other services

193.84

260.96

1.83

3.10

29.21

All industries

454.54

340.47

4.81

5.79

13.13

Major occupation categories

 

Managers

18.23

29.54

1.02

1.49

8.71

Professionals

48.67

35.18

0.98

1.16

6.67

Technicians and associate professionals

97.61

78.81

2.40

2.13

6.78

Clerical support workers

148.38

97.91

1.44

1.36

6.97

Service workers

396.70

89.70

1.82

1.95

19.59

Sales workers

407.86

81.81

1.35

0.46

10.59

Skilled agricultural, forestry and fishery workers

301.60

179.40

7.59

10.40

7.15

Skilled construction workers

931.43

369.79

10.66

4.31

17.68

Skilled extraction and manufacturing workers

988.84

554.24

8.30

5.09

10.00

Craft and related trades workers

920.72

161.05

4.18

1.77

12.12

Plant and machine operators, and assemblers

694.77

377.93

13.96

9.97

9.13

Unskilled service workers except transportation

472.17

361.80

2.89

4.24

29.53

Unskilled non-service and transportation workers

1423.98

640.53

11.19

5.75

26.99

All occupations

453.39

496.33

4.81

5.79

13.13

Source: Estadística de Accidentes de Trabajo (EAT) (2001–2010) and EPA (2001–2010).

A closer look by occupation reveals that work injury rates are the highest among unskilled non-service and transportation workers, as well as among skilled extraction and manufacturing workers. Work fatality rates are the largest among plant and machine operators and assemblers, followed by unskilled non-service transportation workers and by skilled construction workers. Of the aforementioned occupations, unskilled non-service and skilled construction jobs display a larger share of immigrant workers.9

Do work injury and fatality rates then significantly differ according to nativity? Sample means in Table 4 do not disclose a clear pattern. Immigrants appear to work in riskier occupations than natives, but in less risky industries. Because the figures in Table 4 only inform on the average work injury or fatality rate for the entire decade, we take a closer look at how those rates may have varied over time for both immigrants and natives.
Table 4

Immigrant and native injury and fatality rates by occupation and industry

Work injury and fatality rates

Natives

Foreign-born

 

Mean

S.D.

Mean

S.D.

Occupation injury rate per 10,000 workers

438.26

492.41

553.46

510.32

Industry injury rate per 10,000 workers

457.78

340.89

433.11

336.95

Occupation fatality rate per 100,000 workers

4.79

6.51

4.89

5.93

Industry fatality rate per 100,000 workers

4.86

5.87

4.43

5.18

Observations

618501

44990

Note: Observations are weighted using the individual weights in the EPA. The sample includes only individuals aged 16 and older who are employed in the private or public sector, except for military personnel, and not self-employed.

Source: Estadística de Accidentes de Trabajo (EAT) (2001–2010) and EPA (2001–2010).

Figure 1A and B display industry and occupation work injury rates for immigrants and natives over the 2001–2010 period under examination. A couple of things are worth noting. First, work injury rates have been declining consistently throughout the entire time period, except between 2005–2007, when they stabilized among immigrants coinciding with increased immigration inflows and the housing boom. Later on, they dropped with the onset of the economic recession. Second, there are no significant differences in the work injury rate of immigrants and natives by industry; yet, immigrants appear to endure a consistently higher work injury rate than natives by occupation.
https://static-content.springer.com/image/art%3A10.1186%2F2193-9039-2-4/MediaObjects/40176_2012_Article_18_Fig1_HTML.jpg
Figure 1

Figures 1A and B: work injury rates by industry and occupation.

As shown by Figure 2A and B, work fatality rates by industry and by occupation also declined throughout the examined period for both immigrants and natives.10 Nevertheless, unlike work injuries, fatality rates by occupation were not that different by nativity. Additionally, while immigrants display lower work fatality rates at the beginning of the period, immigrant fatality rates catch up with native fatality rates as the decade progresses and immigration rises.
https://static-content.springer.com/image/art%3A10.1186%2F2193-9039-2-4/MediaObjects/40176_2012_Article_18_Fig2_HTML.jpg
Figure 2

Figures 2A and B: work fatality rates by industry and occupation.

In sum, on average, only work occupation injury rates appear to significantly differ by nativity. These differences could, however, be explained by dissimilarities in personal and job characteristics. Therefore, in what follows, we turn to a more rigorous regression-based analysis to address such differences.

5. Methodology

Our purpose is to learn about differences in the work injury and fatality rates experienced by workers according to nativity in Spain during the 2001–2010 decade, how the latter may have been impacted by the recession, and some of the potential explanations for the observed pattern. With that aim in mind, we estimate the following equation by ordinary least squares (OLS):
Rat e ikt = α + β 1 Migran t i + X i γ + η k + φ r + ϕ t + t + η k × t + ε ikt
(1)

where the dependent variable is the work injury or fatality rate in individual i’s industry or occupation k in year t. The variable Migrant is a dummy indicative of whether the respondent is foreign-born that, in alternative specifications, is substituted for a set of dummies indicative of the region of the world where the migrant is from. We also control for a variety of individual level personal and job characteristics included in the vector X known to be correlated to the likelihood of a work accident, such as age, gender, marital status, educational attainment, whether Spanish is their native tongue, time in Spain, job tenure, and contract type.11 Following Hamermesh (1998), we also include occupation fixed-effects when examining work injury and fatality rates at the industry level, and industry fixed-effects when assessing work injury and fatality rates at the occupation level. These are captured by η k . Additionally, denoting by r the region of employment of individual i, regional r and year ϕ t fixed-effects account for a variety of macroeconomic factors possibly correlated to the work injury and fatality rate, such as differences in the distribution of occupations and industries across Spanish regions or specific economic shocks. Likewise, a time trend captures the progressive improvement in work injury and fatality rates exhibited by Figure 1A through 2B; whereas the interaction term η k × t accounts for distinct trends in different industries (occupations). Finally, standard errors are clustered at either the occupation or industry level, depending on the level at which the dependent variable is aggregated at.

After gauging if there are any differences in work injury and fatality rates by nativity during this time period of intensive immigration, we look at how the recession may have impacted such differences by estimating a similar model by OLS. In addition to the previous regressors, the model includes a dummy indicative of the onset of the economic downturn in 2008 (i.e. Crisis), as well as an interaction term capturing any differential impact of the crisis on workers’ injury and fatality rates by nativity:12
Rat e ikt = α + β 1 Migran t i + β 2 Crisi s t + β 3 Migran t i × Crisi s t + X i γ + η k + φ r + t + η k × t + ε ikt
(2)

Equations (1) and (2) are estimated for all occupations and industries in our analysis.

6. Findings

6.1. Differences in work injury and fatality rates by nativity

Table 5 displays the results from estimating equation (1) for work injury rates computed at the occupation and industry levels. Columns (1) through (4) show the estimated gap in occupation injury rates by nativity. The gap drops from 156 injuries per 10,000 workers (specification 1) to 141 per 10,000 (specification 3) as we control for a variety of personal characteristics (such as educational attainment, having Spanish as the native tongue, and years lived in Spain, among other ones). It further drops to 123 per 10,000 workers (specification 4) once we account for a variety of job-related characteristics, such as contract type, job tenure and industry. These differences are all statistically different from zero at the 1 percent level as well as significant from an economic standpoint. For instance, the fact that immigrant workers endure 123 more injuries per 10,000 workers than native workers when the average occupation injury rate among natives is 438 per 10,000 implies that, relative to natives, immigrants tend to work in jobs with work injury rates that are approximately 28 percent higher. However, immigrants and natives do not seem to display different work injury rates by industry (see columns (7) through (10) in Table 5). Likewise, the figures in Table 6 suggest that immigrants and natives do not display statistically different from zero work fatality rates at the industry level. Nevertheless, immigrants do seem to endure 0.49 more deaths per 100,000 workers than natives by occupation once job-related characteristics are taken into account. As such, relative to natives, immigrants not only work in jobs with higher injury rates, but also with fatality rates nearly 10 percent higher than those endured by natives.
Table 5

Effects of immigrant status and region of origin on occupation and industry injury rates

Independent variables

Occupation injury rate

Industry injury rate

 

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

 

Spec.1

Spec.2

Spec.3

Spec.4

Spec.5

Spec.6

Spec.1

Spec.2

Spec.3

Spec.4

Spec.5

Spec.6

Foreign-born

156.70***

220.09***

140.61***

122.95***

  

10.11

−13.13

−32.46

−29.24

  
 

(58.597)

(90.106)

(57.171)

(31.897)

  

(60.539)

(98.861)

(67.930)

(25.146)

  

Years in Spain

 

−8.99***

−4.87***

−3.88***

−4.90***

−3.90***

 

−0.92

0.78

0.84

0.80

0.82

  

(3.186)

(1.831)

(0.904)

(1.844)

(0.921)

 

(2.954)

(2.019)

(0.745)

(1.989)

(0.731)

Spanish native tongue

 

−46.24**

−18.93

−2.54

   

−66.10**

−38.66**

−19.97**

  
  

(27.175)

(19.705)

(12.050)

   

(31.345)

(21.925)

(9.513)

  

Africa

    

201.33***

157.50***

    

11.06

−0.24

     

(76.882)

(46.989)

    

(41.200)

(14.489)

Asia

    

−40.67

27.11

    

−105.16***

−38.26**

     

(44.326)

(34.095)

    

(42.148)

(20.164)

Europe

    

171.41***

127.39***

    

19.86

−7.97

     

(53.472)

(26.405)

    

(56.162)

(19.114)

Latin America

    

137.54***

120.87***

    

−33.47

−29.33

     

(56.481)

(31.579)

    

(67.520)

(25.103)

Other origin

    

8.30

16.10

    

−64.58

−29.68

     

(49.361)

(33.961)

    

(40.910)

(21.904)

Male

  

228.05***

93.41**

227.66***

93.43***

  

220.81***

58.02***

220.85***

58.02***

   

(61.974)

(37.085)

(61.926)

(37.174)

  

(72.632)

(13.197)

(72.500)

(13.188)

Age

  

−8.98**

−3.31

−9.02**

−3.36

  

−2.98

−0.96

−2.98

−0.97

   

(4.455)

(2.690)

(4.469)

(2.700)

  

(2.306)

(1.131)

(2.308)

(1.130)

Age squared

  

0.03

0.01

0.03

0.01

  

−0.01

0.01

−0.01

0.01

   

(0.042)

(0.026)

(0.042)

(0.026)

  

(0.023)

(0.013)

(0.023)

(0.013)

Married

  

−3.20

−15.15**

−2.99

−15.03**

  

28.79***

12.86***

28.92***

12.87***

   

(11.958)

(8.948)

(11.901)

(8.937)

  

(7.960)

(2.657)

(7.839)

(2.622)

Less than primary

  

229.99***

181.12***

226.92***

178.07***

  

45.36

6.61

46.27

5.91

   

(65.824)

(42.673)

(64.124)

(40.926)

  

(46.200)

(7.200)

(47.153)

(7.551)

Primary

  

163.08***

117.09***

162.72***

116.62***

  

51.07**

7.79**

51.45**

7.73**

   

(40.808)

(26.993)

(40.487)

(26.714)

  

(27.941)

(4.449)

(28.120)

(4.491)

University

  

−290.27***

−228.21***

−290.21***

−227.97***

  

−113.81***

−24.59***

−113.94***

−24.59***

   

(47.156)

(34.538)

(47.145)

(34.522)

  

(41.488)

(7.847)

(41.456)

(7.860)

Temporary contract

   

128.18***

 

127.63***

   

−4.68

 

−4.78

    

(35.132)

 

(34.908)

   

(6.034)

 

(6.028)

Job Tenure

   

−0.15**

 

−0.15**

   

−0.07

 

−0.07

    

(0.056)

 

(0.056)

   

(0.048)

 

(0.048)

Regional FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time Trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Industry/Occupation FE

No

No

No

Yes

No

Yes

No

No

No

Yes

No

Yes

Ind/Occ. Time Trend

No

No

No

Yes

No

Yes

No

No

No

Yes

No

Yes

Observations

663,490

663,490

663,490

663,490

663,491

663,490

663,490

663,490

663,490

663,490

663,490

663,490

R-squared

0.049

0.054

0.239

0.417

0.240

0.417

0.072

0.074

0.222

0.525

0.222

0.525

Notes: Injury rates calculated per 10,000 workers. The sample includes only individuals aged 16 and older who are employed in the private or public sector, except for military personnel, and not self-employed. Regressions also include a constant term. Observations are weighted using the individual weights in the EPA. Standard errors are clustered on industry or occupation. Standard errors in parentheses. * significant at 10% ** significant at 5%; *** significant at 1% in one-tailed or two-tailed tests.

Table 6

Effects of immigrant status and region of origin on occupation and industry fatality rates

Independent variables

Occupation fatality rate

Industry fatality rate

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

 

Spec.1

Spec.2

Spec.3

Spec.4

Spec.5

Spec.6

Spec.1

Spec.2

Spec.3

Spec.4

Spec.5

Spec.6

Foreign-born

0.64

0.69

0.39

0.49**

  

0.07

−0.24

−0.29

−0.10

  
 

(0.815)

(1.220)

(0.671)

(0.249)

  

(0.782)

(1.173)

(0.687)

(0.168)

  

Years in Spain

 

−0.05

−0.02

−0.02***

−0.02

−0.02***

 

−0.02

−0.00

−0.00

−0.00

−0.00

  

(0.039)

(0.021)

(0.007)

(0.022)

(0.008)

 

(0.036)

(0.020)

(0.006)

(0.020)

(0.006)

Spanish native tongue

 

−0.91***

−0.31*

−0.02

   

−1.02***

−0.54***

−0.25***

  
  

(0.335)

(0.193)

(0.109)

   

(0.318)

(0.202)

(0.105)

  

Africa

    

0.68

0.49

    

0.17

0.19

     

(1.003)

(0.460)

    

(0.677)

(0.156)

Asia

    

−2.37***

−0.88**

    

−2.08***

−0.40**

     

(0.826)

(0.456)

    

(0.769)

(0.190)

Europe

    

1.09**

0.68***

    

0.58

0.21

     

(0.630)

(0.230)

    

(0.618)

(0.162)

Latin America

    

0.35

0.47**

    

−0.32

−0.10

     

(0.668)

(0.247)

    

(0.685)

(0.168)

Other origin

    

−0.19

−0.09

    

−0.15

0.17

     

(0.524)

(0.388)

    

(0.555)

(0.344)

Male

  

4.58***

2.37***

4.59***

2.38***

  

3.62***

0.80***

3.63***

0.80***

   

(1.108)

(0.615)

(1.111)

(0.617)

  

(1.047)

(0.181)

(1.047)

(0.182)

Age

  

0.03

0.02

0.03

0.02

  

0.05

0.02

0.05

0.02

   

(0.064)

(0.022)

(0.064)

(0.022)

  

(0.050)

(0.016)

(0.050)

(0.016)

Age squared

  

−0.00

−0.00

−0.00

−0.00

  

−0.00*

−0.00

−0.00*

−0.00

   

(0.001)

(0.000)

(0.001)

(0.000)

  

(0.001)

(0.000)

(0.001)

(0.000)

Married

  

0.32**

0.08

0.33**

0.08

  

0.46***

0.19***

0.47***

0.19***

   

(0.167)

(0.079)

(0.167)

(0.080)

  

(0.152)

(0.043)

(0.151)

(0.043)

Less than Primary

  

2.17***

1.36***

2.22***

1.39***

  

1.10**

0.16**

1.14**

0.17**

   

(0.677)

(0.365)

(0.660)

(0.355)

  

(0.497)

(0.094)

(0.487)

(0.092)

Primary

  

1.59***

0.97***

1.61***

0.98***

  

0.91***

0.14**

0.92***

0.14**

   

(0.431)

(0.255)

(0.432)

(0.255)

  

(0.315)

(0.066)

(0.314)

(0.066)

University

  

−2.39***

−1.72***

−2.40***

−1.72***

  

−1.36**

−0.20**

−1.37**

−0.20**

   

(0.741)

(0.426)

(0.743)

(0.427)

  

(0.596)

(0.099)

(0.596)

(0.099)

Temporary contract

   

0.76***

 

0.76***

   

−0.00

 

−0.00

    

(0.264)

 

(0.262)

   

(0.077)

 

(0.077)

Job tenure

   

−0.00***

 

−0.00***

   

−0.00

 

−0.00

    

(0.001)

 

(0.001)

   

(0.001)

 

(0.001)

Regional FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time Trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Industry/Occupation FE

No

No

No

Yes

No

Yes

No

No

No

Yes

No

Yes

Ind/Occ. Time Trend

No

No

No

Yes

No

Yes

No

No

No

Yes

No

Yes

Observations

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

R-squared

0.041

0.042

0.229

0.507

0.230

0.507

0.047

0.048

0.171

0.526

0.172

0.526

Notes: Fatality rates are per 100,000 workers. The sample includes only individuals aged 16 and older who are employed in the private or public sector, except for military personnel, and not self-employed. Regressions also include a constant term. Observations are weighted using the individual weights in the EPA. Standard errors are clustered on industry or occupation. Standard errors in parentheses.* significant at 10% ** significant at 5%; *** significant at 1% in one-tailed or two-tailed tests.

Columns (5)-(6) and (11)-(12) in Tables 5 and 6 further document work safety differences according to immigrants’ origin. Africans, followed by Europeans and, finally, Latin Americans, are all more likely to work in occupations with higher injury rates than natives. Likewise, Europeans are slightly more likely than their native counterparts to work in jobs with higher fatality rates. A quick look at Additional file 1: Table S2 further reveals how Africans are primarily concentrated in the agriculture sector and often occupy unskilled non-service and transportation jobs, whereas Latin Americans primarily concentrate in service jobs. Europeans are relatively concentrated in skilled construction, as well as in plant and machine operating jobs. In contrast, Asians are generally less likely to work in industries with higher injury or fatality rates, as well as in jobs with higher fatality rates, than natives. Specifically, as shown in Additional file 1: Table S2, they primarily work in wholesale and retail trade as managers, service workers, and sales workers.

Also worth discussing is the importance of the duration of the migration spell. Although not always statistically different from zero, the length of the migration spell is generally inversely related to work injury and fatality rates. As noted by Orrenius and Zavodny (2009), this coefficient could be capturing both assimilation and cohort effects. Nonetheless, given that we are focusing on one decade, we are more likely to be capturing the former. Additionally, immigrants for whom Spanish is their native tongue appear less likely to work in risky jobs and, in particular, less risky industries.

The rest of the results in Tables 5 and 6 are the expected. For instance, women tend to work in safer occupations and industries, while older workers tend to work in occupations with lower injury rates –even though the difference disappears once we control for job characteristics such as contract type, job tenure, and industry. Marital status also matters, although not uniformly. Married employees appear more likely to work in occupations that exhibit lower injury rates. Yet, their industries of employment appear to be riskier than those of their single counterparts, and their occupations also display higher work fatality rates. Educational attainment exhibits its expected inverse relationship with injury and fatality rates, with the most highly educated working in safer jobs. Finally, as we would anticipate, employees with temporary contracts or with shorter tenures are more likely to work in riskier occupations than their counterparts with permanent work contracts or longer job tenures.

6.2. The economic downturn and its impact on work safety by nativity

Did the crisis reduce work injury rates? And, if immigrants generally endure worse employment conditions than natives, did the economic downturn further raise their work injury and fatality rates and widen the work safety gap by nativity by pressuring immigrants into accepting riskier jobs? Or did the crisis actually reduce immigrant work injury and fatality rates relative to those experienced by natives due to a greater reduction in immigrant employment and/or an enhanced fear of misreporting?

Table 7 addresses these questions. Overall, the crisis appears to have been inversely related to industry and occupation injury and fatality rates but, for the most part, the effect is not statistically different from zero.13 This could be the case if, for example, lower injury and fatality rates resulting from workload reductions are offset by higher injury and fatality rates byproduct of reduced investments in training and work safety on the part of firms during the recession. In any event, the economic downturn seems to have impacted immigrant and native work injury rates differently. According to the figures in Table 7, the economic downturn particularly lowered work injury rates among immigrants by 32 accidents per 10,000 workers –a 7 percent reduction in the work injury rate. Their fatality rates remained, nonetheless, unaltered.
Table 7

The economic downturn and work injury and fatality rates by nativity

Independent variables

Injury rates

Fatality rates

Occupation rate

Industry rate

Occupation rate

Industry rate

 

Spec.1

Spec.2

Spec.1

Spec.2

Spec.1

Spec.2

Spec.1

Spec.2

Migration variables:

        

Foreign-born

134.45***

 

−26.44

 

0.55*

 

−0.09

 
 

(36.165)

 

(28.125)

 

(0.280)

 

(0.193)

 

Years in Spain

−3.43***

−3.49***

1.00

1.00

−0.02**

−0.02**

−0.00

−0.00

 

(0.783)

(0.822)

(0.678)

(0.670)

(0.007)

(0.007)

(0.005)

(0.005)

Spanish native tongue

−1.18

 

−19.48*

 

−0.01

 

−0.25**

 
 

(12.365)

 

(9.887)

 

(0.108)

 

(0.106)

 

Africa

 

179.03***

 

7.35

 

0.65*

 

0.22*

  

(49.373)

 

(14.669)

 

(0.483)

 

(0.166)

Asia

 

49.00

 

−40.61*

 

−0.75*

 

−0.49**

  

(40.851)

 

(26.664)

 

(0.529)

 

(0.258)

Europe

 

127.31***

 

−3.63

 

0.63**

 

0.24*

  

(27.918)

 

(20.877)

 

(0.255)

 

(0.184)

Latin America

 

133.76***

 

−30.19

 

0.56**

 

−0.12

  

(37.207)

 

(29.579)

 

(0.283)

 

(0.207)

Other origin

 

−11.22

 

−35.76*

 

−0.16

 

0.31

  

(40.368)

 

(23.923)

 

(0.515)

 

(0.488)

Crisis effect:

        

Post-crisis

−17.52

−17.53

−34.24*

−34.28*

0.28

0.28

0.18

0.18

 

(22.926)

(22.927)

(21.630)

(21.620)

(0.253)

(0.253)

(0.215)

(0.215)

Interaction terms:

        

Post-crisis* Foreign-born

−32.48**

 

−7.16

 

−0.17

 

−0.02

 
 

(15.448)

 

(9.719)

 

(0.127)

 

(0.092)

 

Post-crisis*Africa

 

−67.46***

 

−22.87**

 

−0.48***

 

−0.08

  

(19.792)

 

(10.359)

 

(0.176)

 

(0.098)

Post-crisis*Asia

 

−52.38

 

3.00

 

−0.31

 

0.17

  

(41.956)

 

(18.951)

 

(0.320)

 

(0.297)

Post-crisis*Europe

 

−6.10

 

−12.65*

 

0.09

 

−0.08

  

(11.540)

 

(7.883)

 

(0.145)

 

(0.089)

Post-crisis*Latin America

 

−35.55**

 

1.46

 

−0.22*

 

0.04

  

(18.825)

 

(13.369)

 

(0.134)

 

(0.131)

Post-crisis*Other origin

 

71.83**

 

15.16

 

0.21

 

−0.38

  

(37.846)

 

(22.410)

 

(0.566)

 

(0.567)

Personal/Job characteristics

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Regional FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Ind./Occ. FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Ind./Occ. Time trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Observations

663,490

663,490

663,490

663,490

663,490

663,490

663,490

663,490

R-squared

0.416

0.416

0.523

0.523

0.506

0.506

0.526

0.526

Notes: Injury rates are per 10,000 workers and fatality rates are per 100,000 workers. The sample includes only individuals aged 16 and older who are employed in the private or public sector, except for military personnel, and not self-employed. Regressions include a constant term and the personal and job characteristics shown in Tables 5 and 6. Observations are weighted using the individual weights in the EPA. Standard errors (in parentheses) are clustered on industry or occupation. * significant at 10% ** significant at 5%; *** significant at 1% in one-tailed or two-tailed tests.

A closer look by immigrant origin reveals some interesting differences. African immigrants –usually displaying worse employment outcomes and a slower assimilation rate (Amuedo-Dorantes and de la Rica 2010), experienced significant reductions in both job injury and fatality rates following the onset of the economic downturn. Similarly, Latin American immigrants experience a reduction in work injuries and, although only marginally statistically significant, also in work fatalities after 2008. As such, workload reductions and workforce composition biases may be largely responsible for the observed declines in injury and fatality rates among both immigrant groups.

In contrast, just as we find for the average migrant in the sample, Europeans and Asians only experienced reductions in their injury rates (either at the occupation or industry level) after 2008.14The fact that their fatality rates stayed the same lends support to the hypothesis that most of such work injury reductions may be due to misreporting (. Boone et al 2011).

6.2.1. Disentangling the reasons for the reduction in immigrant work injury rates

As noted above, the fact that, for the average immigrant, only work injuries, but not fatalities, decrease among immigrants during the recession supports the notion that most of the reduction in accidents taking place following the onset of the recession might be due to misreporting on the part of immigrants. In other words, immigrants may fear to a greater extent than natives that, in the midst of the economic downturn, the firm might choose to lay off accident-prone workers first. Perhaps, immigrants are less aware of their rights than natives, are more likely to be employed in the informal sector, or, in some instances, they may fear deportation (Orrenius and Zavodny 2012). Consistent with that hypothesis, we find that, for the average migrant in the sample, only work injury rates decrease with the onset of the economic crisis in 2008. Since fatal accidents are more difficult to misreport, there is not an observed decline in fatality rates.

As noted in the previous section, there are, nonetheless, some interesting differences according to the migrant’s origin. In particular, while misreporting appears to be the main cause for the exclusive reduction in work injuries observed for the average migrant in the sample, as well as for European, Asians and immigrants from other origins; African and Latin American immigrants display significant declines in both work injuries and fatalities after 2008. The fact that fatalities –not subject to misreporting– also decrease suggests that misreporting is not likely to be the main driver of such declines in work injuries and fatalities among those two immigrant groups. The literature notes alternative explanations (Boone and van Ours 2006). Specifically, workload reductions and changes in the composition of the labor force over the business cycle could explain the pro-cyclical behavior of accidents. In a boom period, there is a greater workload and also more new hires –often less experienced and more accident prone. During a recession, there are workload reductions that could result in fewer work injuries. Furthermore, firms may dismiss the least productive and more accident-prone workers first. If immigrants are perceived to be less knowledgeable or proficient than natives by employers, they may be dismissed first and workforce composition biases might explain the pro-cyclical pattern of immigrants’ accident rates. The fact that the unemployment rate gap by nativity grew from 4 percentage points to approximately 12 percentages points between 2007 and 2010 suggests that, indeed, workforce composition biases may be one of the explanations for the pro-cyclical pattern (Instituto Nacional de Estadística 2012). Alternatively, difficulties in finding work may induce return migration –a move that could also result in workforce composition biases.

To further sort out these alternative explanations, we re-estimate the model in equation (2) using, instead, a sample of both working and non-working individuals from the EPA. For non-working individuals, we use information on their last occupation/industry of employment. This robustness check allows us to decipher if employment composition biases originating from hiring/dismissal practices are the ones driving the observed reductions in work injuries and fatality rates among African and Latin American immigrants or, rather, workload reductions. If selection of less accident-prone and more capable workers into employment is the main cause for the observed reductions in work injuries and fatalities among African and Latin American immigrant workers during the recession, we should no longer find such a pattern once non-working immigrants are included in the analysis. However, if we still detect significant drops in work injury and fatality rates after 2008 after including non-working individuals in the sample, workload reductions are more likely to be responsible for the observed declines among Africans and Latin Americans.

Table 8 displays the results from the aforementioned analysis.15 The robustness check is intended to help us decipher whether workforce composition biases are the main driver of the observed reductions in work injury and fatality rates among African and Latin American immigrants or, rather, workload reductions. It is worth noting that there are still significant reductions in work injury and fatality rates among African immigrants after including non-working individuals in the analysis. Therefore, workforce composition biases are not likely to be the main explanation for such drops; rather, workload reductions are more likely driving the experienced declines in work injury and fatality rates of African immigrants.
Table 8

The economic downturn and work injury/fatality rates by nativity: working and non-working individuals

Independent variables

Injury rates

Fatality rates

Occupation rate

Industry rate

Occupation rate

Industry rate

 

Spec. 1

Spec. 2

Spec. 1

Spec. 2

Spec. 1

Spec. 2

Spec. 1

Spec. 2

Migration variables:

        

Foreign-born

163.23***

 

−21.87

 

0.82**

 

0.03

 
 

(43.679)

 

(30.586)

 

(0.327)

 

(0.221)

 

Years in Spain

−4.22***

−4.28***

0.98

0.99

−0.02***

−0.02**

−0.01

−0.01

 

(0.985)

(1.025)

(0.673)

(0.664)

(0.008)

(0.009)

(0.005)

(0.005)

Spanish native tongue

1.29

 

−18.42**

 

0.05

 

−0.26**

 
 

(12.879)

 

(10.251)

 

(0.122)

 

(0.114)

 

Africa

 

210.03***

 

9.27

 

0.84*

 

0.42*

  

(57.394)

 

(17.406)

 

(0.526)

 

(0.215)

Asia

 

74.54*

 

−31.04

 

−0.55

 

−0.40*

  

(53.243)

 

(26.108)

 

(0.497)

 

(0.256)

Europe

 

148.41***

 

0.54

 

0.83***

 

0.35*

  

(32.231)

 

(22.170)

 

(0.272)

 

(0.204)

Latin America

 

163.38***

 

−26.41

 

0.83**

 

−0.02

  

(45.090)

 

(31.915)

 

(0.335)

 

(0.233)

Other origin

 

17.71

 

−35.60

 

0.03

 

0.33

  

(41.158)

 

(27.038)

 

(0.501)

 

(0.492)

Crisis effect:

        

Post-crisis

−26.22

−26.22

−41.03**

−41.08**

0.19

0.20

0.15

0.15

 

(27.208)

(27.217)

(22.910)

(22.895)

(0.269)

(0.268)

(0.230)

(0.230)

Interaction terms:

        

Post-crisis* Foreign-born

−23.47**

 

−6.78

 

−0.09

 

−0.05

 
 

(13.595)

 

(10.503)

 

(0.126)

 

(0.098)

 

Post-crisis*Africa

 

−58.93***

 

−22.60**

 

−0.32**

 

−0.21**

  

(15.856)

 

(11.213)

 

(0.187)

 

(0.112)

Post-crisis*Asia

 

−36.04

 

−7.25

 

−0.17

 

0.14

  

(45.952)

 

(17.729)

 

(0.333)

 

(0.289)

Post-crisis*Europe

 

7.76

 

−13.08*

 

0.17

 

−0.10

  

(11.183)

 

(7.986)

 

(0.151)

 

(0.092)

Post-crisis*Latin-Am.

 

−29.97**

 

3.33

 

−0.17

 

0.04

  

(18.057)

 

(13.961)

 

(0.134)

 

(0.137)

Post-crisis*Other origin

 

55.90*

 

12.82

 

0.11

 

−0.44

  

(37.220)

 

(22.600)

 

(0.512)

 

(0.535)

Personal/Job characteristics

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Regional FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Time trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Ind./Occ. FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Ind./Occ. Time trend

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Observations

835,737

835,737

835,737

835,737

835,737

835,737

835,737

835,737

R-squared

0.391

0.392

0.516

0.516

0.500

0.501

0.524

0.525

Notes: Injury rates are per 10,000 workers and fatality rates are per 100,000 workers. The sample includes only individuals aged 16 and older who are employed or unemployed, except for military personnel. Regressions include a constant term and the personal and job characteristics shown in Tables 5 and 6. Observations are weighted using the individual weights in the EPA. Standard errors (in parentheses) are clustered on industry or occupation. * significant at 10% ** significant at 5%; *** significant at 1% in one-tailed or two-tailed tests.

However, including those not working in the analysis does eliminate the observed reduction in work fatalities among Latin American immigrants. Consequently, workforce composition biases might have something to do with their experienced reduction in work fatalities after 2008. Nevertheless, the fact that work injuries continue to be present once we account for workforce composition biases suggest that workload reductions may have also been an important driver of the observed reduction in work injuries among Latin American immigrants following the onset of the economic downturn.

6.2.2. Are higher work injury and fatality rates a unique feature among immigrants?

A final question worth asking is whether higher work injury and fatality rates are only characteristic of immigrant workers or whether they are also observed among other disadvantaged groups of workers, such as low-skilled native youth. If both immigrants and low-skilled native youth are exposed to alike work risks, the employment conditions endured by immigrants might have nothing to do with their nativity but, rather, with their skills. To assess if that is the case, Table 9 replicates the analysis performed in Table 8, but focusing strictly on natives. Specifically, we evaluate if, just like immigrants in Table 8, low-skilled native youth endure worse working conditions than other natives. As in Table 8, we include all working and non-working individuals to address any workforce composition biases.
Table 9

The economic downturn and work injury and fatality rates of low-skilled youth: working and non-working natives

 

Injury rates

Fatality rates

 

Occupation rate

Industry rate

Occupation rate

Industry rate

Demographic variables:

    

Low-skilled youth

37.31

1.42

−0.04

−0.05

 

(32.195)

(7.477)

(0.260)

(0.103)

Crisis effect:

    

Post-crisis

−22.59

−39.73*

0.22

0.15

 

(26.028)

(22.842)

(0.264)

(0.235)

Interaction terms:

    

Post-crisis* Low-skilled youth

−87.17***

−19.97**

−0.39*

−0.08

 

(18.136)

(8.962)

(0.194)

(0.113)

Personal/Job characteristics

Yes

Yes

Yes

Yes

Regional FE

Yes

Yes

Yes

Yes

Time trend

Yes

Yes

Yes

Yes

Ind./Occ. FE

Yes

Yes

Yes

Yes

Ind./Occ. Time trend

Yes

Yes

Yes

Yes

Observations

779,493

779,493

779,493

779,493

R-squared

0.385

0.498

0.492

0.515

Notes: Injury rates are per 10,000 workers and fatality rates are per 100,000 workers. The sample includes only nationals aged 16 and older who are employed or unemployed, except for military personnel. Regressions include a constant term and the personal characteristics shown in Tables 5 and 6. Observations are weighted using the individual weights in the EPA. Standard errors (in parentheses) are clustered on industry or occupation. * significant at 10% ** significant at 5%; *** significant at 1% in one-tailed or two-tailed tests.

As can be seen from the figures in Table 9, unlike immigrants, low-skilled native youth do not seem to work in jobs with higher injury or fatality rates than other natives. However, low-skilled native youth have been hard hit by the recent economic downturn. Given their experienced workload reductions,16 we would expect the recession to have lowered their work injury and fatality rates to a greater extent that those of other natives. And, indeed, the economic downturn reduced both the work injury and fatality rates of low-skilled native youth by 87 accidents and 0.39 deaths per 100,000 workers. According to the figures in Additional file 1: Table S3, these declines amount to approximately a 9.5 percent and a 4.5 percent reduction in their occupation work injury and fatality rates, respectively.

In sum, while the recession reduced the work injury and fatality rates of low-skilled native youth in a similar way as those of immigrants, there are two key differences between the two aforementioned groups. First, unlike immigrants, low-skilled native youth do not generally endure worse working conditions than their remaining native counterparts. Second, the fact that both work injury and fatality rates experienced by low-skilled native youth declined during the downturn hints on workload reductions as the primary driver, whereas misreporting seems to be the main reason for the decline in the job injury rate experienced by the average immigrant in Tables 7 and 8.

7. Summary and concluding remarks

We examine whether immigrants appear to have worked in riskier jobs –as captured by injury and fatality rates measured at the occupation and industry level– than natives during the 2001–2010 decade in Spain –a period of intensified immigration encompassing the most recent economic downturn. We further explore how any differences in work safety by nativity may have been exacerbated or narrowed by the recent economic downturn. Specifically, we examine if immigrants –who endure a higher risk of being unemployed than natives and a higher opportunity cost to being unemployed due to their lower likelihood of qualifying for some safety nets, such as unemployment benefits17– experienced a greater reduction in work injury and fatality rates associated to reductions in workload, workforce composition biases, or misreporting relative to natives.

The results clearly indicate that, on average, immigrants –by origin: Africans, followed by Europeans and Latin Americans– work in more dangerous occupations than natives, even after accounting for observable characteristics, such as educational attainment, time in Spain or whether Spanish is their native language. Indeed, depending on the model specification, the work injury and fatality rates in occupations held by immigrants are roughly 30 percent and 10 percent higher than those in occupations held by natives, respectively.

Furthermore, while the recession only appears to have slightly lowered industry injury rates, it has had a differential impact on the work safety of immigrant and native workers. In particular, work injuries –but not fatalities– declined exclusively among immigrants after 2008. While a variety of factors are typically likely to be at play, the fact that only work injuries, and not fatalities, dropped following the onset of the economic downturn suggests that workload reductions and workforce composition biases are unlikely to be the leading explanation for the observed declines in work injuries. Rather, greater fear of dismissal and, to a much lesser extent, workforce composition biases may be responsible for the exclusive decline in work injury rates experienced by the vast majority of immigrants. A closer look by immigrant origin reveals how that has certainly been the case for European and Asian immigrants. However, African and Latin American immigrants experienced reductions in both work injuries and fatalities following the onset of the crisis –a pattern that hints on other factors as the main explanations affecting both injuries and fatalities, such as workload reductions and workforce composition biases.

We further explore which of the latter two explanations dominates among immigrants from each of those two origins. We find that, among African immigrants, workload reductions are likely to be the more prominent reason behind the observed declines in work injuries and fatalities, as the two persist after accounting for workforce composition biases. In contrast, addressing workforce composition biases eliminates the significant reduction in work fatalities experienced by Latin American immigrants after 2008. Therefore, in addition to workload reductions explaining their persistent decline in work injuries, workforce composition biases may have been a main explanation for their lower fatality rate after the economic downturn.

A couple of concluding remarks are worth making. First, to the extent that: (a) work injury and fatality rates by industry and by occupation combine immigrants and natives, and (b) informality, which is more common among immigrants, is more likely to result in misreporting, our estimates likely represent lower bounds. Second, we are unable to examine with the data at hand whether immigrants receive a compensating wage differential for working in riskier jobs than natives. If immigrants are taking these jobs due to misinformation or lack of alternative employment opportunities, they might not earn the same compensating wage differential as natives. In that case, corrective measures addressing these disparities in work safety dangerously concealed by the economic downturn might be warranted.

Endnotes

1 Even after the economic downturn, the foreign-born accounts for more than 12 percent of the population (Instituto Nacional de Estadística, 2012).

2. (Note however that Bonin et al 2009) find that, after controlling for age, education, and family characteristics, first-generation immigrants in Germany are more risk-averse than natives.

3 According to the European Commission (2009), up to 1,232,000 irregular immigrants were present in Spain at the beginning of 2005. And, although this figure significantly declined with the 2005 amnesty, this number still stood at 354,000 in early 2008.

4 Furthermore, the information cannot be used to locate undocumented workers.

5 As noted by Alonso-Villar and del Rio (2010) or Amuedo-Dorantes and De la Rica (2010), among others, doing so minimizes any seasonal effects.

6 Statistical Classification of Economic Activities in the European Community.

7 International Standard Classification of Occupations for European Union purposes.

8 We had to homogenize NACE Rev. 1, used for 2001-2008, and NACE Rev. 2, used for 2009 and 2010. We ended up with 44 different industry clusters and 61 occupation categories.

9 Additional file 1: Table S2 further disaggregates the share of immigrants employed in the industry and occupation categories included in Table 3 by region of origin.

10. (Benavides et al 2009) examine whether the implementation of preventive measures by the regional governments were responsible for the declining trend in industry and job injuries and fatality rates from 2000 onwards. However, they do not find any significant results. The authors then propose alternative explanations, including: (a) increased safety inspections, (b) changes in workforce compositions from high-risk to low-risk industries, and (c) changes in reporting standards introduced by insurance companies.

11We include these characteristics sequentially to assess how the estimated coefficient changes as we include some variables that could be potentially considered endogenous, as is the case with contract type or job tenure.

12 In this second specification, we do not include the full set of year fixed-effects since they are reported to be collinear to the crisis dummy. We are, however, able to include year dummies for the years 2001-2007 along with the crisis dummy and the results (available from the authors) do not change.

13 We only observe marginally significant reductions in industry injury rates –hinting on the pro-cyclicality of accident rates previously noted on the literature for other countries ( Kossoris e.g. 1938, Fairris 1998. or, more recently, Boone et al 2011).

14 We also find that work injuries increased for migrants from other origins, e.g. Oceania. This could occur if, for instance, firms invest less in safety and training following the onset of the crisis, resulting in higher work injury rates. Nevertheless, the very small number of observations in that category prevents us from making any significant inferences.

15 On average, according to the figures from specification no. 1, immigrant work injury rates now drop by less than before (by 23 versus 32 per 10,000 workers). However, the reduction continues to be statistically different from zero after including non-working individuals. Therefore, as we had already concluded from the figures in Table 7, employment biases are not likely to be the main factor driving the reduction in work injuries experienced by the average immigrant following the onset of the recession. That is also the case for European and Asian immigrants (specification no. 2).

16 It is estimated that unemployment rates among young low-skilled natives have risen to a record high 50 percent during the recession that started in 2008.

17. (Vazquez et al 2009) note that, although the Spanish unemployment benefit scheme is one of the most generous in Europe, the fact that it requires 360 days of contribution over the past 6 years results in fewer benefit-entitled immigrants, especially among recent cohorts.

Abbreviations

EAT: 

Estadística de accidentes de trabajo

EPA: 

Encuesta de población activa

ISCO: 

International standard classification of occupations

MCVL: 

Muestra continua de vidas laborales

NACE: 

Nomenclature générale des activités économiques dans les communautés européennes.

OLS: 

Ordinary least squares

Declarations

Acknowledgements

This paper has benefited from funding from the University of Seville (24/02/2011) and from comments provided by Madeline Zavodny and participants at the Seminar on Demography and Economics at the University of Seville, the 2012 ESSLE, and the 2012 Symposium of Economic Analysis. Any remaining errors are our own.

Responsible Editor: Denis Fougère

Authors’ Affiliations

(1)
Department of Economics, San Diego State University
(2)
Department of Economics and Economic History, University of Seville

References

  1. Ahonen EQ, Benavides F: Riesgo de lesión por accidente laboral en trabajadores extranjeros. Gac Sanit 2008,22(1):44–47. 10.1157/13115110View ArticleGoogle Scholar
  2. Alonso-Villar O, Del Rio C: Occupational segregation of immigrants in Spain. Papeles de Trabajo del Instituto de Estudios Fiscales no. 09/2010. 2010.Google Scholar
  3. Amuedo-Dorantes C: Work safety in the context of temporary employment: The Spanish experience. Ind Labor Relat Rev 2002,55(2):262–285. 10.2307/2696208View ArticleGoogle Scholar
  4. Amuedo-Dorantes C, De la Rica S: Immigrants’ responsiveness to labor market conditions and their impact on regional employment disparities: evidence from Spain. SERIEs 2010,1(4):387–407. 10.1007/s13209-009-0002-yView ArticleGoogle Scholar
  5. Antecol H, Bedard K: Unhealthy assimilation: Why do immigrants converge to American health status levels? Demography 2006, 43: 337–360. 10.1353/dem.2006.0011View ArticleGoogle Scholar
  6. Bauer TK, Million A, Rotte R, Zimmermann KF: Immigration Labor and Workplace Safety. 1999. IZA DP nº 16Google Scholar
  7. Benavides FG, Garcia AM, Lopez-Ruiz M, Gil J, Boix P, Martínez JM, Rodrigo F: Effectiveness of occupational injury prevention policies in Spain. Public Health Rep 2009, 124: 180–187.Google Scholar
  8. Berger MC, Gabriel PE: Risk aversion and the earnings of US immigrants and natives. Appl Econ 1991, 23: 311–318. 10.1080/00036849100000138View ArticleGoogle Scholar
  9. Bonin H, Constant A, Tatsiramos K, Zimmermann KF: Native-migrant differences in risk attitudes. Appl Econ Lett 2009,16(15):1581–1586. 10.1080/13504850701578926View ArticleGoogle Scholar
  10. Boone J, van Ours JC: Are recessions good for workplace safety? J Health Econ 2006, 25: 1069–1093. 10.1016/j.jhealeco.2006.03.002View ArticleGoogle Scholar
  11. Boone J, van Ours JC, Wuellrich J-P, Zweimüller J: Recessions are bad for workplace safety. J Health Econ 2011, 30: 764–773. 10.1016/j.jhealeco.2011.05.013View ArticleGoogle Scholar
  12. Díaz-Serrano L: Do Legal Immigrants and Natives Compete in the Labour Market? Evidence from Catalonia. 2010. IZA Discussion Paper No. 4693Google Scholar
  13. European Commission: Irregular Migration in Spain. Clandestino Research Project. 2009. Available at: http://clandestino.eliamep.gr Google Scholar
  14. Fahr R, Frick B: On the Inverse Relationship between Unemployment and Absenteeism: Evidence from Natural Experiments and Worker Heterogeneity. 2007. IZA Discussion Papers 3171Google Scholar
  15. Fairris D: Institutional change in shop floor governance and the trajectory of postwar injury rates in the U.S. manufacturing, 1946–1970. Ind Labor Relat Rev 1998, 51: 187–203. 10.2307/2525213View ArticleGoogle Scholar
  16. Gonzalez L, Ortega F: How do very open economies absorb large immigration flows? Recent evidence from Spanish regions. Labour Econ 2011, 18: 57–70. 10.1016/j.labeco.2010.06.001View ArticleGoogle Scholar
  17. Hamermesh DS: Immigration and the quality of jobs. In Help or Hindrance? The Economic Implications of Immigration for African Americans. Edited by: Hamermesh DS, Bean FD. New York: Russell Sage Foundation; 1998.Google Scholar
  18. Hersch J, Viscusi K: Immigrant status and the value of statistical life. J Hum Resour 2010,45(3):749–771. 10.1353/jhr.2010.0014View ArticleGoogle Scholar
  19. Instituto Nacional de Estadística : Extranjeros en España. Madrid: Cifras INE; 2004.Google Scholar
  20. Instituto Nacional de Estadística: Encuesta Nacional de Inmigrantes 2007: Una Monografía. Madrid: Instituto Nacional de Estadística; 2009.Google Scholar
  21. Instituto Nacional de Estadística: Extranjeros en la UE y en España. Madrid: Cifras INE; 2012.Google Scholar
  22. Kossoris M: Industrial injuries and the business cycle. Mon Labor Rev 1938,46(3):579–594.Google Scholar
  23. Leeth J, Ruser J: Safety segregation: The importance of gender, race, and ethnicity on workplace risk. J Econ Inequality 2006,4(2):123–152. 10.1007/s10888-005-9008-2View ArticleGoogle Scholar
  24. Loh K, Richardson S: Foreign-born workers: trends in fatal occupational injuries, 1996–2001. Mon Labor Rev 2004,127(6):42–53.Google Scholar
  25. Marvasti A: Occupational safety and English language proficiency. J Labor Res 2010,31(4):332–347. 10.1007/s12122-010-9096-zView ArticleGoogle Scholar
  26. Ministerio de Trabajo e Inmigración: Estadística de Accidentes de Trabajo. Datos Anuales; 2001–2010. Published on line: http://www.mtin.es/estadisticas/eat/welcome.htm Google Scholar
  27. Moral De Blas A, Martín Román A, Rodríguez Caballero JC: La antigüedad y las diferencias de esfuerzo entre trabajadores de distintas zonas geográficas: un estudio de los accidentes de trabajo. Estudios de Economía Aplicada 2010, 28: 201–221.Google Scholar
  28. Orrenius PM, Zavodny M: Do immigrants work in riskier jobs. Demography 2009,46(3):535–551. 10.1353/dem.0.0064View ArticleGoogle Scholar
  29. Orrenius PM, Zavodny M: Immigrants in risky occupations. 2012. IZA DP No. 6693Google Scholar
  30. Reher DS, Cortés Alcalá L, González Quiñones F, Requena M, Sanz Gimeno A, Sánchez Domínguez M, Stanek M: Informe Encuesta Nacional de Inmigrantes (ENI – 2007). Madrid: Instituto Nacional de Estadística; 2008.Google Scholar
  31. Rosen S: The Theory of Equalizing Differences. In Handbook of Labor Economics. Edited by: Ashenfelter O, Layard R. North Holland, Amsterdam; 1986.Google Scholar
  32. Sesé A, Palmer AL, Cajal B, Montaño JJ, Jiménez R, Llorens N: Occupational safety and health in Spain. J Safety Res 2002,33(4):511–552. 10.1016/S0022-4375(02)00054-3View ArticleGoogle Scholar
  33. Shea J: Accident rates, labor effort, and the business cycle. SSRI Working Paper 9028. Madison: University of Wisconsin; 1990.Google Scholar
  34. Solé M, Díaz-Serrano L, Rodriguez Martínez M: Work, Risk and Health: Differences between Immigrants and Natives in Spain. 2010. IZA Discussion Papers 5338Google Scholar
  35. Vasileva K: 6.5% of the EU population are foreigners and 9.4% are born abroad. Eurostat: Statistics in Focus 34/2011. Eurostat, European Union; 2011.Google Scholar
  36. Vázquez P, Alloza M, Vegas R, Bertozzi S: Impact of the Rise in Immigrant Unemployment on Public Finances. 2009. FEDEA Working Paper No. 2009–15Google Scholar
  37. Viscusi WK: The value of risks to life and health. J Econ Lit 1993, 31: 1912–1946.Google Scholar

Copyright

© Amuedo-Dorantes and Borra; 2013

This article is published under license to BioMed Central Ltd. licensee Springer. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.