This article is also available to readers in PDF format

Labour overtime in Chile:

An exploratory model *   

Jan Cademartori
Catholic Univ. of the North, Antofagasta, Chile.
Email: jcademar@ucn.cl

Daniel Cáceres
Catholic Univ.of the North, Antofagasta, Chile.
Email: dcaceres@ucn.cl

Received: 5.27.2009 Accepted: 9.23.2009

* * *

Introduction

The economic growth is not necessarily associated with people's quality of life if it is obtained with long working days with problems associated to family life and mental health.  In spite of the above some entrepreneural elites and sectors contiguous to the Chilean Government recommend to make flexible the labour market rules to promote employment. It should then be asked if this would affect the the workday's duration and the quality of life.

This subject matters in many countries where there is labour flexibility. In Great Britain, the 1980s reforms increased the workday (Bell and Hart, 1999). In the United States the 1980s and 1990s policies increased the effective extension of the workday (Golden and Jorgesen, 2002). In Australia's mining, the number of worked hours increased between 1983 and 1998 from 46.0 to 51.7 and this happened in parallel to the out-sourcing increase (ILO, 2002). In Chile there has occured something similar since this country led the ranking of 49 countries where there are more working hours, in 1996, 1998 and 2002 (Work Directorate, 2003).

This research confirms Chile's flex time. If our general average of 46.6 hours and its median of 48 hours were representative of the data mass, we would have scarce variance. However our average has a strong dispersion around it (13.0 as standard deviation). In 2003, there were so many cases up and down the 40-50 hours stretch (Graphic 1.1). 31% of the cases worked over 48 standard hours. It is surprising that in this stretch, instead of falling below the curve abruptly after 48 hours, there is a long row above the 53 hours, row covering 65 % of those who are over 48 hours (Graphic 1.2). As a result of the above, this subset worked in average 58,3 hours.

This reality of Chile, which postulates to the developed world, can be compared with a 2007 directive of the European Union which limited the workday to 48 hours including overtime. The Chilean case also contrasts with Holland where it is allowed an average of 40 hours in 13 weeks (ILO, 2007). In 2004, the average workday duration of people at full time was 38.9 in France, 40.3 in Spain, 42.0 in the USA and 42.7 in Britain (Bruyere et al., 2006).

GRAPHIC 1.1: FREQUENCY (%)

BY STRETCH OF WORKED HOURS

Source: Own development with CASEN 2003.

GRAPHIC 1.2: FREQUENCY (%)

WITHIN THE SEGMENT HIGHER THAN 48 HOURS

Source: Own development with CASEN 2003

In this work we will defend that flex time is linked with labour precarity and lack of income. Indeed, in Chile there has decreased the power of trade unions to control the workday and improve wages. The post dictatorship reforms did not re-establish so important trade union gains before the 1973 as: prohibition of replacing the strikers during their cessation, negotiation by branch of economic activity, compulsory affiliation to the trade union, imperative character of the trade union's dues, extension of the collective agreement to the new worker, automatic extension of readjustment of the public sector to the private sector. Thus, the unionization declined from 14.5% in 1991 to 10.5% of employment in 2001 (Pizarro 2005) against nearly 40% at the time of the Popular Unity Government (Unidad Popular).

Equally in Chile the subcontracting that replaces the permanent worker has been extended to a temporary worker. Nearly 35% of Chilean labour works under subcontracting regime after several years of sustained increase (Silva 2007).  The contract duration is submitted to the projects' end. The opportunities for trade union are less because on the same establishment there appear multiple smaller enterprises which often hide their dependence on a same employer. These factors are exacerbated with "the staff provision", this is, when the "temporary" worker supplies services, sometimes for a long time, under the orders of a different company with which he has signed the labour contract.

This research will try to analyze the relation between overtime and some explanatory variables especially with labour precarity and low wages. In this line there are few econometric models therefore we consider this work as a first exploratory effort that has deficiencies. Anyway, in the first sections are presented some basic relations to finish with a model of several variables.

Model of Analysis

Unfortunately the econometric models to which we had access in main current magazines (index ISI) are concentrated in more developed countries (ILO, 2007). Among these works outstands the one of Engelland and Riphahn (2005) where it is proved that in Switzerland the labour effort depends on the employment's temporality. This effort translates into extra free hours to improve the reputation and obtain permanent employment. Meanwhile Trejo (1993) validates a model of labour demand where trade union prevents the free overtime, and thus reduces its purchase. This is also confirmed by Green (2002) for England in 1993.  On the other hand, the first two authors control by variables that are taken in this article.

In our model the dependent variable was the number of worked hours in the week prior to the survey. The explanatory variables are Social-Economic: Labour Precarity and Wage. The control variables are the personal features (sex, age and marital status), the geographical context (Urban or Rural Zone, Metropolitan Region) and productive (Mining Zone, Company's Size). It was also analyzed as dependent variable the percentage of people who worked more than the standard workday (48 hours) however the econometric results with this dichotomous variable are unsatisfactory at the moment.

TABLE 1: VARIABLES AND RANGES

Source: Own Development with CASEN 2003

TABLE 2: DESCRIPTIVE STATISTICS

OF VARIABLES

Source: Own development with CASEN 2003

It would have been interesting to incorporate other specific variables such as the unionization, the subcontracting, and the compliance of the extra hour. Unfortunately, the unionization is not adequately captured by our database (CASEN) because the question contains exclusive alternatives regarding other forms of organizing. Nor could it be isolated the presence of subcontracting nor to distinguish the free overtime of the paid one. However it was constructed as proxy solution an IC labour precarity indicator detailed in the next section. This IC indicator operacionalizes the concept of labour precarity, which has multiple dimensions and various definitions.

In this work, labour precarity is understood as a perception of insecurity in the employment's duration, fed by an objective position of increased vulnerability. It can be assumed that the fear of loosing the job forces to work more hours to accumulate savings because in Chile the unemployment insurance is almost non-existent. Likewise, this insecurity encourages a greater effort without need of payment of the extra tour, with the aim of achieving a better reputation to obtain a permanent, employment, as evidenced by the model of Engellandt and Riphaun (2005) quoted above. This could explain that in Hart and Yue (2009) employment's seniority lowers the paid overtime. These two reasons, savings and reputation search make us think that at higher labour precarity workers accept more hours.

Concerning the worker's wage the theory says that there exist forces that affect the dependent variable with opposite signs. In our model we postulate that low wages force people to work more hours to cover their basic needs. Instead, Trejo (1993), using the wage as control variable, obtained an association with positive sign in the United States. The effect of these socio-economic variables is discussed in detail in the following sections.

The geographical and productive context also influences. In Cademartori (2009) it is shown that the mining Region of Antofagasta, with its high productive levels and strong presence of staff subcontracting, shows the highest level of worked hours among the regions of Chile (CASEN without regional weighting). Also ILO (2002) indicates that mining is a productive sector with long workdays in Australia

Besides, urban areas and especially the Metropolitan Region, provide better opportunities than rural areas to work long workdays as there exist greater amount of companies and therefore of work. In addition, the survey was made between the months of October and November, this means,. when agricultural work does not require extra hours compared to the period of harvest.

Another aspect which can intervene is the company's size. (Engellandt et al. 2005)). In our model this variable has ambiguous effects. It is expected that at larger size, the unionization protects the payment of extra hour, this encourages its purchase by workers but not by the company. However the larger size allows the company to obtain resources for expanding production and hiring extra hours. In the reverse many small enterprises compensate their organizational and financial weakness using precisely the extra unpaid hours. Likewise this variable relates with small subcontractor companies and with labour precarity (negatively) and with the worker's wage (positively) so that at smaller size, less extra hours.

The personal features here considered are: sex, age and marital status. It is reasonable to expect that female workers work fewer extra hours than men because they often assume the additional household burden. In our sample 32% of men worked more than 48 hours (against 23% of (women). Regarding people living with a partner, usually women must maintain children and sometimes the couple who does not work.

Another aspect which it considers is the worker's age. We suppose that elder people are more sensitive to fatigue and health problems when their workday lengthens. In addition, older people already have housing and other durable goods that young people seek to finance working more hours.  These factors generate a negative relation between age and extra hours. This does not occur in the first age stretch where it is faced less pressure to form a family. Then the model will be controlled with an inverted U curve because the data suggest a negative effect at greater distance of an intermediate age that is between 40 and 45 years old (Figure 1 in Annex).

Variables' Measurement and Data

We postulate that the perception of insecurity affecting the worker on his employment's duration depends on his objective situation. We have proposed an objective indicator of labour precarity (IP) that sums the ten variables detailed in Table 3.  These variables that were originally discrete, have become dichotomous for each individual: they assume the value one in presence of precarity and, otherwise, zero value. The sum of these variables allows to obtain an IP with values between 0 and 10.  The most precarious workers were coded with value 1 in all variables with which their IP is equal to 10. Subsequently the IP was transformed into quintiles (QIP) to obtain an index having less isolated data in the end values and to obtain representative averages of each group.  

TABLE 3: VARIABLES OF

LABOUR PRECARIOUSNESS INDEX

Source: Own development with CASEN 2003

TABLE 4: PEARSON'S CORRELATIONS

AMONG PRECARITY VARIABLES

Source: Own development

These ten variables that comprise the IP present positive and significant Pearson's binary correlations at 1 %. In addition, the factor analysis of the main components (Table 1 in Annex) delivers as first component positive charges for all these variables. This component explains twice the variance compared to a second component.  Thus it is reasonable to conclude that the ten variables belong to the same dimension that we call labour precarity.

It was also considered the wage per hour obtained by work (income of the main occupation in CASEN). The data do not allow to distinguish the wage for the normal workday of the value paid for the extraordinary hour therefore we conform with the wage per W hour that corresponded to the total of the workday. We have estimated W considering that the law obliges in week an extra payment of the hour higher than 48 hours of 50% over W (except holidays). Thus, the weekly income Y (extracted from CASEN) of the worker with overtime is the sum:

 Our data was extracted from the 2003 National Survey of Economic Characterization (CASEN) in charge of the Government. This survey was applied to a sample of 58,000 households of all communes of the country.  The original database was weighted by a factor that allows to adjust the regional sample to its national participation. The people who are not waged workers were removed from it, this is, independent workers and entrepreneurs. There were also eliminated those who are less than 16 and over 65 years old, ages that officially are not part of the labour force. Likewise there were withdrawn of the database those who did not respond some of the questions on labour precarity. The domestic service workers and the Armed Forces were included. The main disadvantage of CASEN survey is that it is a survey aimed at households and not to workplaces therefore the information on the employee and his work environment is indirect.

In the two following sections are highlighted some thick relations. The sample has been divided into sub-groups containing a similar percentage of the population. This allows to reduce the variance of individual data and to operate with averages by group.

Precariousness and Worked Hours

In Chile 2003 the standard legal workday was 48 hours. The following Graphic 2 shows the IP of three groups of people against HDICOT, the percentage who worked more, equal or less than 48 hours. The line that starts up the bar marks the range of data whose IP exceeds 75% of the data and the line that starts down starts at the inferior range at 25%. Within each bar it has been marked its median that accumulates 50% of the data. This Graphic shows the similarity between the first and the third bar. Something must be in common between these two extreme groups. We sustain that it is labour precarity.

A more delicate analysis is accomplished with the hours worked per week (H). This relation generated three zones (Graphic 3.1-3.2 and Table 5). In the first zone, H increases with QIP until it is reached the second zone where this relation becomes saturated. In the third zone the worked hours decrease with QIP. To represent these graphics we propose a third-degree polynomial allowing a goodness fit greater than 90 % regardless of whether there are ten (Figure 3.1) or five IP categories (Graphic 3.2). This three degree and its r2 greater than 0.9 repeat if the H dependent variable is replaced by HDICOT, the percentage of people who work more than 48 hours (Graphic 3.3). It is therefore a scarcely sensitive relation to the way of measuring the concept.

 Our interpretation is as follows. The higher levels of precarity force people to work longer for the exposed reasons: savings and reputation.  However if precarity levels are too high, there do not exist opportunities to purchase extra hours. Many of these people have low productivity and/or their companies are not very profitable. Anyway the dispersion of the average increases with precariousness (CV, Table 5).

  GRAPHIC 2: PRECARITY QUARTILES

ACCORDING TO WORKDAY'S EXTENSION

Source: Own development with CASEN 2003

TABLE 5: LABOUR PRECARITY VS.

WORKED HOURS

Source: Own development with CASEN 2003

GRAPHIC 3.1: WEEKLY HOURS WORKED

BY LABOUR PRECARITY DECILE

Source: Own development with CASEN 2003

GRAPHIC 3.2: WEEKLY HOURS WORKED

BY LABOUR PRECARITY QUINTILE

Source: Own development with CASEN 2003

GRAPHIC 3.3: WEEKLY WORK HIGHER THAN 48 HOURS

(% of workers)

 Source: Own development with CASEN 2003

As it will be seen in the next section, there is neither a simple relation between income and worked hours. Indeed labour precarity and income levels show statistical association what makes probable that there are presented non simple curves. We start with a brief theoretical discussion which is used to interpret the graphics.

Worked hours and income per hour

The standard theory relates the workday length with the wage per hour. For this it must be assumed that extra hours are recharged with a multiple of the base wage that we assume constant and independent of the wage level. Thus, by adding in each case the extra hours with the standard hours, the total hours demand of the companies and the total hours offer of the workers, can be rewritten based on the base wage.

Regarding the demand, at greater wage, the companies should buy less hours by replacement effect and more hours by income effect. These hours are for them more expensive compared to other inputs such as new labour saving machinery or to subcontract external services (replacement effect).  However the hours purchase is incremented when the companies are more successful, this is, they receive more customers orders and hire more productive and better remunerated employees thanks to that larger demand and that higher productivity. Thus, these mutually related variables generate a positive association between wage and total hours (income effect). Also, in companies where wages and productivity are higher it is likely that the company agrees to amortize the initial investment in professional training distributing it in more extra hours instead of hiring new inputs.

Also from the point of view of the labour supply, the increase in the standard wage has a contradictory effect on worked hours. By replacement effect, it could be expected that at higher wage, workers have greater incentives to prolong their workday because the opportunity cost of the rest time increases. In other words at greater wage more time is worked. The income effect generates an opposite incentive because at greater wage the workers' needs remain covered without the obligation of using overtime.

In textbooks it is assumed that this last effect is less important than the substitution effect by which it is drawn a curve where the labour supply in hours increases with wage except for the minority of high income. In this section the stylized facts seem to adjust to a labour supply where the income effect has much influence and the substitution effect is marginal. This offer seems to combine with a labour demand whose opposing forces are annuled so the demand is irrelevant or its replacement's effect is more substantial than the others. This would allow to conclude that the worked hours follow the labour supply this and this latter presents a curve with negative slope because those receiving a lower wage work more hours (Graphic 4 and 5).

The previous analysis can be opposed because it extrapolates the normal hours demand to overtime, which is restricted to special periods of the year of high demand for the final product (Trejo 1993). However it seems to us that the extra hours demand is independent of wage also in these special periods. This can be explained in the following way. In the peack periods the company has a guaranteed positive operational margin, subtracting even the additional overtime cost and its fee with surcharge. This is achieved by the employer by negotiating the selling price of the final good in front of the extraordinary order or by the natural rise of this value when the market demand for the final good seasonally expands. Indeed this margin will be positive in cases where the employer circumvents the surcharge to the wage required by law.

Thus, regarding the entrepreneural demand, the higher wages also have no adverse effects on the overtime recruitment in peak periods, then leaving the problem in the hands of a labour supply dominated by the income effect. This could be explaining Graphic 4.

In this Graphic, the DW variable represents the W wage decile to which the worker belongs and the first decile groups 10 % of the people of lower income. H symbolizes the average number of worked hours.  The lack of income forces people to work more hours even if these latter do not linearly depend of the wages per hour W. This is ratified in Graphic 5 where HDICOT represents the percentage of people who worked more than the 48 hours standard workday.

Graphics 4 and 5 suggest that the workday clearly decreases with DW up to the fourth decile (the workday increases with poverty). Instead, the following deciles (4, 5, 6) show a softer descence in the number of hours (Graphic 4), and certain stability for the percentage (Graphic 5). The final three quintiles redisplay a negative relation in hours. These data are in accordance with a third degree polynomial in the variable DW (r2 = 0.93) and a second degree parable in the variable HSDICOT (r2 = 0. 94).   

In this way (Graphic 5), there exists a negative relation between income per hour and worked hours for nearly 80% of the population. This contradicts the significant portion of the labour supply curve of orthodox microeconomics textbooks where the substitution effect dominates the income effect in most of the curve which is crescent (positive slope).

The most consistent interpretation for a country of lower income with strong social inequality for that income is that people work in excess to cover their basic needs and the high debt levels of the families. This thesis is consistent with the indirect privatization of basic services as education which obliges families to increasingly assume their funding. This would explain that it works 55 hours the average of the first three deciles (Graphic 5) and that only the richest decile of workers (probably professionals and managers) works the oficial workday of developed countries. 

Tus, in the presence of long working workdays, the average income per capita is misleading as welfare action of a country, but not only because if its unequal distribution, but because this income is accomplished with very different workdays and labour conditions that depend precisely of its unequal distribution. For this same reason, even the known indexes of income inequality (10/10 Gini) underestimate the true inequality per working hour.  That without considering other variables that probably depend on labour precarity and the social status such as: work accidents, work intensity, monotony, promotion possibilities, participation, etc. In the following section it is proposed a more complete model that considers the simultaneous effect of several variables and makes useful the previous sections.

GRAPHIC 4: WORKED HOURS

VERSUS INCOME DECILE

Source: Own development with CASEN 2003

GRAPHIC 5: WORKERS WORKING OVER STANDARD WORKDAY (%) VERSUS INCOME DECILE

Source: Own development with CASEN 2003

Model with several variables

The previous sections help to build hypothesis. However, not controlling by the control variables, could make the signs of relations change. In addition, a model with a single explanatory variable does not allow to compare the relative importance of each determinant. To avoid these problems it is proposed the following linearized model where the Worked Hours depende on:

The two first explanatory variables, labour precarity quintile (QIC) and the wage decile (DW), were divided in third-degree polynomials for the reasons indicated in previous sections. The rest are control variables presented in table 1 of the first section. Among these control variables, only the age variable is found as polynomial (of second degree) to reflect the inverted U curve suggested in Graphic 1 in Annex. It was used the SPSS program to obtain a linear regression by the method of ordinary minimum squares.

The CASEN database gives the option to multiply each data by regional ponderers to reconcile the weight of the region's population with the weight of the individual in the sample. In the previous sections it was opted to use the regional ponderers. However, this rose n to 3.061.128. In this section the base regression weighted regionally but each data of the regional expansion was amplified by the factor (50,212/3,061,128) with which n = 50, 212 is maintained.

The coefficients and their level of significance of the proposed model in (3) are shown in the first column of the following Table. In the second column are shown the Students test of the parameters.

The model in (4) does not improve by substituting the Precariousness Index for the first component of the above mentioned Factor Analysis. The signs of the polynomial coefficients are all negative, what is more difficult to interpret. In addition the coefficient estimator that accompanies the second degree term of the precarity polynomial ceases to be statistically significant at 20% while the third degree coefficient is not statistically significant at 5%.  On the other hand we see that with the new variable, the goodness fit (r2) diminishes regarding the previous regression.

With the purpose of evaluating the coefficients' stability and reduce the sample's size (avoiding T test of too accordant) ten regressions were made, choosing randomly in each one 4 % of the sample with original regional weighting (n = 3, 061 million).

Of the base regression and ten simulations (Table 6) it is concluded that:

1. The base regression has a low r2 but this and the average of the ten are globally significant at 1 %.
2. The coefficients of Labour Precarity and Income polynmials have the expected signs and orders of magnitude according to the curves proposed in previous sections.
3. The Age variable presents unexpected signs for an inverted U. However, as shown in Graphic 1 of Annex, the Age is related to the percentage of people who worked more than 48 hours (HDICOT).
4. On the the ten regressions' average, the variables which we call of control, have the expected signs and are significant at 1 % with some nuances:

- The pertainance to the Metropolitan Region and Sex do not have the expected sign.

- The mining Region variable is not significant at 10 %. 

- The Age, Metropolitan Region and Company's Size variables are significant only at 10 %.

- The coefficients' variability, measured by the Variation Coefficient (CV) is quite higher than the average in Sex and Company's Size variables although sex is significant at 1 %.

These last 3 and 4 problems advise to improve the model in future works. In addition, Table 2 in Annex presents the Pearson correlations among the explanatory variables. It is observed that labour precarity is related to wage and the company's size. Another problem is that it is not discarded the self-correlation among the residues according to the D-W test at 1 %. 

TABLE 6: SUMMARY OF REGRESSIONS

Source: Own development

TABLE 7: REGRESSIONS WITH RANDOM SAMPLES.

Source: Own development.

Conclusions

In this work were analyzed the working hours behavior in Chile based on a households survey conducted in 2003. The sample covered nearly fifty thousand workers with representation in the different regions of the country.

The results show that an important segment (30%) of workers laboured most of the 48 standard hours in that year, and for that subset the average was of 58,3 hours a week, figures much higher than the one regulated by the European Union.  To explain the workday there were analyzed the work conditions, the wage income; the characteristics of the person, region and company were used as control variables

The perception of insecurity in employment we call it labour precarity and this affects in the employees' pressure that leads them to work more hours both by reputation effect as by the need to prepare themselves for bad times. In this work we quantified subjective precarity through the objective conditions of work. Thus, the objective conditions were summarized by an index comprising ten variables related to labour stability. These ten variables present significant and positive correlations and can be grouped as the primary axis of a factor analysis. Subsequently the working population was grouped in quintiles according to its relative position in this index. We suppose that grouping by averages stratas of large data masses there exists a strong statistical association between the subjective dimension and the objective dimension.

With this new variable we obtained that people work more hours at greater precariousness and only when precarity levels are very high the employees cannot "buy" overtime to compensate their precarity. This significant relation between the workday extension and labour precarity generates large variations in the workday's duration and therefore the income inequality per worked hour would tend to increase with a greater employers' flexibility. In addition, the legal workday decreasement will be scarcely effective while the background causes are not solved: labour precariousness and social inequality.

These results not only highlight the relation between labour precarity and over-employment but the relation between labour precarity and sub-employment, a typical phenomenon of the peripheral countries. Unfortunately in Chile it is considered occupied a person who worked a few hours weekly prior to the survey by which the theme of sub-employment or disguised unemployment is being lost. In our análisis, instead, both the over-employment and the sub-employment are manifestations of a same problem which is labour precarity. In this framework of analysis, favourable actions to employers' flexibility could be facilitating both the excessive work for a good quality of life as the disguised unemployment.

Besides, in textbooks it is assumed that normally, at less wage there will be less hours offer by the workers. However we obtained that, in technical terms, the substitution effect has scarce importance in comparison to the income effect.  In a more simple language our results suggest that the workday is extended by the workers themselves with the purpose of compensating the lack of income for 80 % of the population who believe that their needs are unsatisfied. This makes us think that poverty is much higher than the official measure calculated by MIDEPLAN.

Likewise, there exist large variations in the workday duration as a result of the income inequality. In the first poorer decile there are worked 56.3 hours against the 39.4 of the richest decile. This means that the traditional ways to measure income inequality (20 % versus 20 % or Gini) also should incorporate the effort required to earn the wage dividing the people's income by the worked hours.

These conclusions are held with an econometric model in which were added variables that control the personal characteristics of the worker, the characteristics of the region and the productive sector although some of them are weaker than the expected. This should be improved in future studies. There should also be measured the direct relations between unionization, subcontracting and the workday duration.

Bibliography

Bell, D., Hart, R., (1999), “Unpaid Work”, Economica, London School of Economics and Political Science, 66 (262), 271-290.

Bruyère, Mireille, Odile Chagny, Valérie Ulrich (2006),  «Comparaisons Internationales de la Durée du Travail pour Sept Pays en 2004: la place de la France». Données Sociales. L'édition française. INSE. Recuperado de http:\insee.fr.

Cademartori, J. (2008), El Desarrollo Económico y Social de la Región de Antofagasta: Historia y Perspectivas ORDHUM, Departamento de Economía, UCN.

Dirección del Trabajo (2003), “Relaciones de trabajo y empleo en Chile. ENCLA 2002” . Departamento de Estudios. Santiago: Ministerio del Trabajo y Previsión Social.

Engellandt, A., Riphahn, R. (2005), “Temporary Contracts and Employee Effort”, Labour Economics, 12, 281-299.

Golden, L, Jorgesen, H., (2002), “Time after time: Mandatory Overtime in the U.S. Economy”, Washington D.C.: Economic Policy Institute.

Green, F. (2002), “Why has Work Effort Become More Intense?” Studies in Economics 0207. Department of Economics, University of Kent.

Hart, Robert K. y Yue, Ma (2009), “Wage-hours contracts, Overtime Working and Premium Pay”. Labour Economics. Accepted Manuscript 6 April 2009-

ILO (2002), The Evolution of Employment, Working Time and Training in the Mining Industry”, Report for discussion at the Evolution of Employment, Working Time and Training in the Mining Industry, ILO, Geneva.

Lee, S., McCann, D., Messenger, J. (2007), Working Time around the World: Trends in working hours, laws and policies in a global comparative perspectiv ”, ILO, Geneva.

Pizarro, Roberto (2005), “Desigualdad en Chile, desafío económico, ético y político”. Polis, Revista Académica Universidad Bolivariana, Vol.3, Nº 10.

Silva, Consuelo, (2007), La subcontratación en Chile, una aproximación sectorial. Consejo Asesor Presidencial Trabajo y Equidad. Recuperado de http:/trabajoyequidad.cl, Santiago de Chile.

Trejo, S. J. (1993), “Overtime Pay, Overtime Hours, and Labor Unions”, Journal of Labor Economics, 11, 253-278.

Annexes

TABLE 1: MATRIX OF PRINCIPAL COMPONENTS

TABLE 2: PARTIAL CORRELATIONS ZERO ORDER.

GRAPHIC 1: PEOPLE WORKING MORE THAN 48 HOURS

VERSUS AGE

Source: Own development based on CASEN

Notes

* Research supported by the Regional Science and Public Policies Nucleus of the Millennium Scientific Initiative Programme.

GRAPHIC 1.2: …and more

ECUACIÓN: (1) Then (2) if yes no

GRAPHIC 5: decile

ECUACION: (3) age, sex,…..size

(4) sex, age, age,…size

GRAPHIC 1 (Annexes): (age) (age) age

setTimeout(function(){parent.Translate.Run();},0);