Open Access

Dropout trends and educational reforms: the role of the LOGSE in Spain

  • Florentino Felgueroso1,
  • Maria Gutiérrez-Domènech2 and
  • Sergi Jiménez-Martín3Email author
IZA Journal of Labor Policy20143:9

https://doi.org/10.1186/2193-9004-3-9

Received: 13 November 2013

Accepted: 6 March 2014

Published: 9 May 2014

Abstract

Over the last 50 years, some important reforms in European countries were aimed at improving the system of vocational studies. By contrast, the Spanish educational law (LOGSE) from 1990 moved in the opposite direction. While the LOGSE increased the number of compulsory schooling years from 8 to 10, it also eliminated vocational studies of first grade (FP-I, ages 14 to 16), thereby reducing flexibility. The LOGSE was rolled out at different times across regions within Spain. This is used as an identification strategy to analyse whether the change in law contributed to stop the declining trend in dropout rates observed in the last two decades (they declined from 70% in 1977 to 30% in 1995, but remained at roughly 30% until recent years, twice the EU27 average). Results show that the new compulsory secondary education increased dropout rates for men and decreased them for women. Alternatively, the removal of the lower vocational track had negative effect for both genders, being the effect stronger for women. Finally, by reducing the track choice opportunities for students, the reform reduced the probability of following the vocational track after completion of the compulsory stage.

JEL codes

I20, J24.

Keywords

School dropoutEducational lawVocational studies

I. Introduction

School dropout is one of the main problems of education systems and it is therefore high on the policy agenda in most OECD countries.1 Concerns about school dropout are based on its potential adverse consequences with respect to perspectives on the labour market, the health condition and the participation in society (Heckman & Lafontaine, 2007). Indeed, social costs associated with school dropout can be large in terms of social exclusion and, for instance, involvement in criminal activity.

While dropout rates in Europe have been falling continuously in recent years, they remained stable in Spain between 1995 and 2008. Dropout rates only started to fall in 2008 once the severe crisis expulsed youngsters with low education from the labour market.2 Nevertheless, dropout rates continue to be significantly high in Spain, 24.9% in 2012, among the highest in Europe, and roughly twice the EU28 average, which stands at 12.7%.

A closer look to long-term dropout rates in Spain (see Figure 1) reveals a clear break in its negative trend in the mid 90s. In this sense, it is very interesting to analyse what may have happened in the mid 90s that contributed to deterring the declining process of dropout rates in Spain. The hypothesis investigated in this paper is whether the implementation of a new educational law in academic year 1991-92 (Ley Orgánica de Ordenación General del Sistema Educativo, LOGSE, passed in 1990), amongst other factors, contributed to suspend the declining trajectory of dropout rates observed years earlier. This law replaced the previous one (Ley General de Educación, LGE, 1970). The LOGSE was rolled out at different times across regions. This allows us to build up a natural experiment that uses this variation to identify the likely effect of the new educational law on dropout rates and also on vocational track enrolment. We use individual data from the Spanish Labour Force Survey (Encuesta de Población Activa, EPA) from 1987 to 2010, and restrict the sample to individuals from 18 to 24 years old born in the cohorts potentially affected by the reform, between 1977 and 1986. We then test the effect of the educational law LOGSE on the individuals’ probability of being a school dropout.3
Figure 1

Evolution of school dropouts in Spain. 1992-2011. Source: European Labour Force Survey (Eurostat).

The LOGSE in 1990 increased the number of compulsory schooling years from 8 to 10 and, hence, it provided two more years of academic curricula. This way the starting age of non-compulsory secondary education (Bachillerato) increased from 14 to 16 years old and the ending age of primary education moved from 14 to 12. It also delayed the choice between academic or vocational paths for two years, until the age of 16. As a result, vocational studies of first grade (FP-I, ages 14 to 16) were eliminated and all students from that range of age were forced to study secondary compulsory education. Hence, with the reform a more flexible system was replaced by a more rigid new system. In this context, students less keen on continuing their studies towards a more academic type of degree were obliged to proceed and were not given the chance to move towards a more vocational type of training.

Overall, the results of the paper suggest that the LOGSE did not have any positive effect on dropout rates. However, this general result hides large differences by gender. For males the introduction of the LOGSE has negative implications on dropout rates, while the reverse is true for women. Alternatively, the removal of the lower vocational track is negative for both, but only significant for women. Note, finally, that the latter effect has also substantial implications on track choice, especially for women.

The rest of the paper goes as follows. In Section 2, we summarise the literature on school dropout. Section 3 describes the trends and factors underlying school dropout in Europe. Section 4 explains the main educational reforms in Europe over the last 50 years, including the Spanish one from the early 90s (LOGSE), and shows the differences across regions in its implementation. In Section 5, we present the data, the methodology as well as the identification strategy; we discuss the validity of our natural experiment and present some descriptive statistics. The main results obtained from the analysis are described in Section 6. Finally, Section 7 concludes.

2. Literature

This section summarises the main research about school dropout distinguishing four main lines. First, some papers investigate the gains from additional schooling and use new compulsory school laws as instruments to estimate the counterfactual for students that would otherwise have left sooner (Angrist and Krueger (1991) and Acemoglu and Angrist (2000) for the US; Harmon and Walker (1995) for the UK; Oreopoulos (2007) for the US, UK, and Canada; Brunello and Rocco (2010) for the European Countries). Overall, they find that compulsion school raises earnings to individuals that otherwise would be dropouts. Other studies find non-pecuniary losses of school dropout such as committing crime or ending up in jail (Lochner and Moretti (2004)) and higher mortality (Lleras-Muney (2002)). For the Spanish case, Felfe et al. (2012) study the effect of extending universal compulsory schooling to age 3 (also introduced by the LOGSE) on childen’s cognitive development. They find sizeable effects on reading skills and weak evidence of a reduction in grade retention.

Second, some analyse the choice of post-secondary studies (attending high school, staying at home or employed in various occupations) (Keane and Wolpin (1997), Eckstein and Wolpin (1998) and Arcidiacono (2004, 2005)). For the Spanish case, López-Mayan (2010) analyses the choice between academic and vocational tracks, and she finds that there is a strong association between decreasing the dropout rate in upper secondary education and the increase of annual wage of medium vocational education qualifications.

Third, there is research on the determinants of dropout: family background, type of schools, transmission across generations (Dustmann (2004) and Blanden (2009)); the situation of local labour markets (Duncan (1965) and Clark (2007)); and cohort size (Felgueroso and Jiménez-Martín (2009)). For the Spanish case, Mora, Escardíbul and Espasa (2010) look at the impact of expenditure per pupil, class size, and pupil-teacher ratio on dropout rates at the end of compulsory education and find the expected signs, although the magnitude of the effects is relatively small. Lacuesta et al. (2012) find that the increase in Spain of the unskilled to mid-skill wages ratio observed between 1993 and 2001 led to an increase in the fraction of youth who completed at most primary education (or less), but mainly at the expense of a reduction of youth completing upper high school.

Finally, some authors evaluate education policies aimed at reducing dropout. A large number of studies look at the financial incentives to continue studying (Dynarski (2003) in the US and Schultz (2003) in Mexico, Angrist and Lavy (2004) in Israel, Dearden et al (2009) in the UK). Overall, the message of all these papers is mixed. Our paper belongs to this group of research since it analyses the impact of the introduction of a new law, the LOGSE, on school dropout.

3. Trends and drivers of school dropout in Europe and the role of vocational studies

Dropout rates are very different across countries. More importantly, Figure 2 shows that most countries reduced their dropout rates over the period 1997-2007, except for Spain and the Scandinavian countries. In fact, Spanish dropout rates decreased from 70% in 1977 to 30% at the end of the 1990s, but remained at roughly 30% until 2008. School dropout rates recently declined to 25% since the recession brought back to study some youngsters. By contrast, Italy and Portugal, the two other EU15 countries with very high incidence of school abandonment, improved their rates between 1997 and 2007. This is especially the case for Italy, whose dropout rate declined ten points. Ireland also experienced a substantial decline in school abandonment over this period, around 7 points, and the rates in UK, Greece and the Netherlands progressed as well around five points. Despite having experienced a slight increase in the prevalence of dropout, the Scandinavian countries continue to hold the lowest dropout rates in Europe.
Figure 2

Evolution of early school leavers by country (1997, 2007). Source: European Labour Force Survey (Eurostat).

A large number of early school leavers can lead to a dual composition of the population by educational levels: a high share of individuals with high and low educational attainment and by a low share of individuals with medium attainment. As shown in Figure 3, Spain is clearly characterised by this dual composition. Although the fraction of low educated people decreased between 1996 and 2008 (about 17 percentage points), Spain remained well behind other countries (Portugal, Italy and Greece).
Figure 3

Composition of the population by educational levels in EU 15 (population aged 25-64, 1996-2008). tes: Lower level corresponds to the % of population with at most lower secondary education; higher level means the % of population with tertiary education (first or second stage). Source: European Labour Force Survey (Eurostat).

Moreover, Figure 3 shows that the proportion of population in Spain with medium degree continues to be amongst the lowest in Europe, despite having increased around 7 percentage points over this period. By contrast, the share of highly educated individuals evolved very favourably (around 10 percentage points) and it is currently among the highest, certainly above the EU 15 average. Hence, this data reveals that Spain was quite successful in achieving a large share of population with tertiary qualifications, but did not manage to shift the fraction from low to medium (Felgueroso and Jiménez-Martín (2009)). Other northern European countries are better located in this respect.

The composition of the population by educational levels determines a country’s productivity and livings standards. In fact, studies point out that it is more efficient to devote resources to reach a very large proportion of individuals with secondary education rather than the aforementioned dual composition with a very large number of dropouts and relatively high tertiary rates (World Bank (2005)). Clearly, the reduction of the incidence of school dropout is key to achieve a larger share of individuals with secondary education.

Although there are many factors that affect school dropout (the productive model, the socio-demographic characteristics and the economic cycle), the structure of upper education is key (Blondal and Adalbjarnardottir (2009); Pereita and Pastor (2000)). Studies show that one of the key aspects in order to prevent early school abandonment and hence boost medium educational levels is a well-functioning vocational training system (Gangl (2003), Haywood and Tallmadge (1995), Digest (1987), Plank et al. (2005) and Field et al. (2007)). They also emphasise that it is positive to use more work-oriented vocational programs. There are other advantages of a well-functioning vocational system: students with practical abilities are better off following a vocational degree (Meer (2007)); countries with an extensive vocational training system tend to have better transitions between education, employment and apprenticeship than countries without these systems (Brzinsky-Fay (2007)).

Spain is not very well placed with regards to its vocational system. A report from the OECD (2008) recommends increasing the attractiveness of vocational studies in this country and improving its quality. Cross-country comparison shows that the differences in the weight of vocational studies on the upper secondary level are large (Figure 4). The variation is also significant in terms of how vocational studies are structured: systems that combine school and work programs and others with only school-based vocational programs (see Table 1).
Figure 4

Distribution of upper secondary education (2007). Notes: Lower level corresponds to the % of population with at most lower secondary education; higher level means the % of population with tertiary education (first or second stage). Source: OECD (2008) and own calculations.

Table 1

The role of the educational system (2007)

 

Distribution of upper secondary education

Weight of vocational studies

  

High

Medium

Low

Distribution of Vocational Programs

High work-based

Austria , Czech Republic

Germany, Denmark, France

 

Low work-based

Belgium, Netherlands, Finland

Poland, Spain

Ireland, Hungary

 

Medium work-based

Italy

Sweden

United Kingdom, Greece, Portugal

Source: Field et al. (2007) and own calculations.

We observe, for example, that Austria and Czech Republic are characterised by a high proportion of vocational studies in upper education and they also have a high rate of vocational programs that combine school and work. Other countries with a relatively high prevalence of vocational studies are Belgium, the Netherlands and Finland. On the lower spectrum are located Greece, United Kingdom and Portugal.4 The educational system in each country is the result of educational reforms that took place over the last century (see next section). In this sense, it is very interesting to analyse the implementation of the Spanish educational reform in 1990. The new law somehow moved against this principle by abolishing the possibility of undertaking vocational studies at earlier stages.

4. Educational reforms in Europe

Over the last 50 years there have been many educational reforms across countries affecting multidisciplinary aspects. Amongst them, those reforms that were probably more important for the incidence of dropout were those affecting the number of years of compulsory school and those regarding the main organisational and curricular aspects of secondary and post-secondary non-tertiary studies (ISCED 2-4).

Simultaneously, there were changes in other educational levels (pre-primary, primary and tertiary education) and dimensions (for instance, with respect to teachers, school autonomy and special education). Although those were also relevant for the well-functioning of the educational system, in general, their impact on early school-abandonment was probably smaller. For simplification, these reforms are not incorporated in the educational calendar in Figures 5 and 6 but a description of them can be read in Garrouste (2011).
Figure 5

Reforms in compulsory years of schooling. Source: Garrouste (2011).

Figure 6

Main organisational and curricular laws in secondary and post-secondary non-tertiary in Europe. Source: Garrouste (2011).

Figure 5 shows the reforms in the number of years of compulsory education that took place over the last fifty years in several countries in Europe. The range of compulsory years is between 9 and 12. The majority moved compulsory education upwards over this period of time. In particular, it is remarkable the gradual shift from 5-6 to 10 in Greece, Italy and Spain. Most other countries had done this shift earlier on. However, increasing the number of years in compulsory education, in spite of being a positive move in terms of guaranteeing somehow better minimum standards, it does not affect the ultimate level of dropout. More important is how secondary school is organised.

Figure 6 shows the reforms in the main organisational and curricular laws of secondary and post-secondary non-tertiary (ISCED 2-4). We observe that most countries implemented changes to the vocational systems to adapt them to new needs and to make their degrees more compatible to academic programs. By contrast, in Spain there was a reform in 1990 (although effective since 1991-1992) that abolished vocational studies of first grade and, consequently, somehow moved in the opposite direction.

In Figure 7, we place together the most important reforms over time and the evolution of dropout rates for several countries in order to have a first look on whether there is a potential relationship between them. For the Spanish case, the timing of this law matches the break in the mid 90s of the declining trajectory of the dropout rates. This law was introduced gradually from 1992 onwards being almost fully implemented (100% of the first two courses and 90% of the last two) in the academic year 1998-1999. In Italy and Portugal, we observe a certain accentuation of their declining profile of dropout rates when they undertook reforms to boost vocational studies. For other countries, the relationship is less clear but none seems to contradict this hypothesis. Therefore, overall Figure 7 shows some signs that indeed the way in which the educational system evolves may be important for the incidence of dropout.
Figure 7

Evolution of dropout and main reforms in Europe. Source: Garrouste (2011), Eurostat and own calculations. In the case of Spain the reform was introduced in 1991-92 but was no fully in place until 1998-1999.

4.1 Educational reform in Spain: the LOGSE

The hypothesis tested in this paper is that the implementation of a new educational law in 1991-92, the LOGSE, that replaced the LGE, prevented dropout rates from falling further. Figure 8 summarises the educational path of the two systems.
Figure 8

Schooling levels in Spain under the LGE and the LOGSE. Source: López-Mayan (2010) and own elaboration.

Under the LGE, compulsory education was between 6 and 14 years old and covered the EGB (Educación General Básica) or primary education. Individuals who did not manage to graduate from EGB were obliged (Art 3.2. of the LGE) to move towards lower vocational studies (Formación profesional) until the age of 16 (FP-I). However, the enforcement Art 3.2 of the LGE was not very effective until a new law (LOGSE) was introduced (see Egido (1995), page 79). A student who completed EGB could either choose secondary education – BUP (Bachillerato Unificado Polivalente) or lower vocational studies. A secondary education degree could be followed by academic high school – COU (Curso de Orientación Universitaria), a specialised track to attend university.

On the other hand, the LOGSE made compulsory primary education EPO (Educación Primaria Obligatoria) until 12 years old and compulsory secondary education ESO (Educación Secundaria Obligatoria) until 16 years old. After that, the student could either choose Academic High School (Bachillerato) for two years (until 18 years old), and get prepared to attend university, or move to vocational studies (Formación Profesional de Grado Medio).

Hence, the LOGSE set up several organizational changes, two of them key for the analysis of this paper. First, it increased effective compulsory schooling age for all students by two years, from 14 to 16. Second, it shortened primary education from 14 to 12 and introduced a new four years stage of secondary compulsory education (ESO) from 12 to 16. Therefore, the starting age of Academic High School (Bachillerato) increased from 14 to 16 years old. As a result, vocational studies of first grade (FP-I, ages 14 to 16) were eliminated and all students from that range of age were forced to study secondary compulsory education. The latter implied that less academic oriented students were obliged to proceed and were not given the chance to move towards a more vocational type of training. This could have possibly generated negative peer effects since the less-academic and less-motivated students were left together with the rest, and the chances were high that they hold the general progress back. At the same time, by forcing students to remain in the academic track for longer periods could have deterred enrolment in the vocational track.

One of the key aspects of the reform that can be used for the analysis is that there were differences across regions in the implementation of the educational law. Besides demographic and labour market factors, how fast regions implemented the LOGSE possibly played a crucial role in explaining the different regional trends in dropout. In this sense, our analysis uses the variation across regions in when the LOGSE became in operation to test whether the shift from LGE to LOGSE, amongst other factors, contributed to suspend the declining trajectory of dropout rates observed years earlier. This is a reasonable natural experiment exercise since the differences in timing and velocity of carrying out the LOGSE were uncorrelated with the initial level of dropout rates of these regions.

On these grounds, Figure 9 presents two indicators of how rapidly the LOGSE was introduced during the nineties and also and indicator of how long it took the old system to disappear. The first indicator is the fraction of the 12-17 population studying under the LOGSE over the period 1990-2000. We observe that some regions started to implement it in academic year 1991-92, but others did it later. By the academic year 1999-2000, the LOGSE was full in place in all regions. The second indicator is the fraction of 12-13 population studying, under the new system, the first two courses of the ESO. The graph makes it clear that there is a lot of variation across regions and that the new system was fully introduced by 1997. The last indicator is the fraction of population 14-15 still studying under the old system lower vocational studies. It is clear that there is a lot of variation across regions and cohorts in the removal of the previous educational system.
Figure 9

The introduction of the LOGSE and the removal of the old lower vocational studies. SOURCE: Ministry of Education and own calculations.

5. Econometric model

We would like to test the effect of the educational law LOGSE on the individuals’ probability of being a school dropout. The probability of dropout is associated with the degree of exposure of each individual to the LOGSE (the treatment indicator), amongst other factors.

In more detail the basic model we consider is as follows:
y iqr * = α + X iqr ' β + γ T iqr + δ Z qr + u iqr

Where y iqr * denotes the propensity to dropout of individual i from birth cohort q in region r, X is a vector of individual and family characteristics, T denotes treatment (1 if exposed to the LOGSE, and zero if exposed to the LGE) and Z is a vector or regional controls at the time of entry in the labour market.

Since y iqr * is unobservable, we define y iqr = 1 y iqr * > 0 . Hence, our dependent variable takes value one if the individual is a dropout, that is, he/she is between 18 and 24 years old (21 in some experiments in order to minimize the potential sample selection problem due to nest leaving) with at most lower secondary degree and not in further education or training; otherwise, it takes value zero. Assuming normality of the error term u we can (up to scale) identify the parameters of the model by estimating a probit model by maximum likelihood. Given that our policy variables have variation at the regional level, we cluster standard errors at that level of variation.

It is worth noticing that in our “natural” experiment there are two groups of subjects: a control group who did not have exposure to the LOGSE and a treatment group who did have exposure to this law. Since we do not have complete information on whether an individual studied under the LGE or LOGSE, the indicator for the control and treatment group is determined by an instrument that allocates the level of exposure to the LOGSE to each individual. Had we known for each individual the system under he/she studied we would have used that variable as our treatment for evaluating the impact of the law.

Hence, we can only use an instrument for the treatment that is calculated using external information for the individual related to the region and time. This allows classifying each student according to the level of exposure for the law using three dimensions: year, region and date of birth. For instance, we know that in a particular region and year, the proportion of students within a certain age band under the LOGSE system was κ. We then allocate this κ to that individual according to his/her date of birth, which becomes his/her index of exposure to the LOGSE as in Duflo (2000).

As said, since the introduction of the LOGSE progressed differently across schools and regions we do not know exactly whether an individual was exposed to the treatment or not. However, we can clearly consider three periods: before 1991 only the LGE was in place and hence treatment = 0 for all the individuals; between 1991 and 1999 both the LGE and LOGSE coexist with differences across regions and hence some individuals will be allocated to 0 and others to 1 depending on their region; from 2000 onwards only LOGSE is in place so treatment = 1 for all the individuals.

As with any independent variable, we would like to check whether the measured effect of the index of exposure to the LOGSE is casual. That is, is the difference in the likelihood to be a dropout between individuals who were exposed to the LOGSE and those who were not due to this exposure, or is this indicator simply reflecting something else? It might be that the individuals exposed to the LOGSE would have had, on average, higher dropout even if they had not been exposed to the LOGSE. The more ad-hoc was the allocation of the incidence of the LOGSE across individuals, the more likely that our estimates for the index of exposure to the LOGSE will show the appropriate relationship between this indicator and the probability of dropout.

In this sense, it is very important to control for as many factors as possible that might be related to whether an individual was exposed to the LOGSE (see Table 2 for the descriptive statistics of these variables for the entire sample as well as disentangling for men and women). In our regression, besides the individual index of exposure of the LOGSE we include other variables (more information can be read in Appendix): demographic characteristics such as whether the individual as a child cohabited with a father or mother, parental education, cohort effects and origin; labour market and economic factors that vary across regions and time (expected wage at age 30 by education level, unemployment rates, the share of employment in construction and the rate of regional growth).
Table 2

Descriptive statistics

 

Males

Females

 

Mean

St.dev.

Mean

St.dev.

Dropout

0.368

0.483

0.230

0.421

Vocational track

0.251

0.433

0.198

0.398

Demographics

    

Born abroad

0.022

0.148

0.029

0.167

Father present

0.830

0.375

0.798

0.402

Mother present

0.917

0.275

0.884

0.320

Father dropout

0.454

0.497

0.443

0.497

Mother dropout

0.548

0.498

0.530

0.499

Cohort size

0.990

0.111

0.988

0.112

Regions

    

Andalucía

0.204

0.402

0.206

0.405

Aragón

0.040

0.196

0.040

0.195

Asturias

0.024

0.153

0.025

0.155

Balears (Illes)

0.022

0.146

0.021

0.145

Canarias

0.057

0.232

0.057

0.232

Cantabria

0.024

0.154

0.023

0.149

Castilla y León

0.078

0.269

0.076

0.265

Castilla-La Mancha

0.096

0.294

0.097

0.296

Cataluña

0.095

0.293

0.094

0.291

Comunitat Valenciana

0.083

0.277

0.086

0.280

Extremadura

0.040

0.197

0.042

0.200

Galicia

0.064

0.245

0.065

0.247

Madrid

0.057

0.233

0.056

0.230

Murcia

0.035

0.184

0.033

0.179

Navarra

0.020

0.140

0.021

0.142

País Vasco

0.048

0.213

0.047

0.212

LaRioja

0.013

0.113

0.012

0.110

Labour Market and Business Cycle (regional)

    

Expected real wage ratio at age 30

1.696

0.224

2.061

0.213

Employment in construction

0.103

0.019

0.103

0.019

Unemployment rate

0.204

0.067

0.204

0.068

Growth rate

0.023

0.019

0.023

0.019

Educational Model

    

Competences in education

0.394

0.470

0.387

0.469

Adequacy rate at age 15 (%)

65.564

8.008

76.626

6.571

Incidence of the LOGSE

    

% students in LOGSE 12-17

0.392

0.319

0.391

0.319

% students in FP-I

0.104

0.072

0.105

0.072

% students in ESO-I

0.350

0.429

0.348

0.429

Observations

53738

51884

SOURCE: Individual data from Spanish Labour Force Survey (EPA). Educational variables from the Ministry of Education (MEC). Rest of the variables from http://www.ine.es.

The analysis also includes educational characteristics such as the adequacy rate at age 15. The larger the adequacy rate the lower will be the incidence of negative peer effects. In fact, data shows a decreasing trend of the adequacy rate in recent years, probably due to the fact that students cannot be expelled from the system until they are 16 years old. Another variable is whether a region had full competences in education or not. Regions who have full competence since the beginning of the period of analysis are Catalonia, Bask Country, Galicia, Andalusia, Canary Islands, Community of Valencia and Navarre. The rest are under the ruling of the Ministry of Education (MEC), but many of them started having competences during the transition period (1991-1999). In addition to the previous variables, the study also incorporates region dummies to control for other possible factors that are region specific and may be related to dropout but are not included in the regression. Similarly, we include year of birth dummies to take into account for systematic differences that may vary over time and could have an impact to dropout but are not in the model.

5.1 Description of the data and the use of a natural experiment

This paper uses data from the Spanish Labour Force Survey (Encuesta de Población Activa, EPA) from 1987 to 2010. We restrict our sample to individuals from 18 to 24 years old born in the cohorts potentially affected by the reform, between 1977 and 1986. Since the first year of implementation of the LOGSE was in academic year 1991-92, this means that individuals born in 1976 were not affected by the change in law. Similarly, the last academic year of the pre-LOGSE system was 2001-02, which implies that all individuals born from 1987 onwards were fully under the LOGSE. As a result, cohorts between 1977 and 1986 were affected by the reform depending on the region and constitute the basis of our sample. This methodology has been appointed by others (see Lacasa (2006)) but it has not been formally analysed. One crucial condition for our analysis is that the differences in the implementation of the LOGSE across regions were independent from dropout rates at the initial state. That is, for our econometric model to work it is essential that there is no rationale behind the differences in the implementation of the new educational law across regions. For this purpose, we calculated the correlation between dropout rates across regions in the first period of the analysis and the degree of implementation of the LOGSE the same year and found that the two events were uncorrelated. This finding reinforces the approach followed in the paper.

5.2 The proxies for the treatment

Clearly, the construction of an indicator to account for the degree of exposition of a region (or individual) to the LOGSE is very important since this is the variable that we would like to test in this paper. To avoid relying on a single indicator or proxy for the treatment we consider the three indicators presented in Figure 9. Clearly, the main indicator is the % students in LOGSE 12-17. However, we also use two other more specific that look, for instance, at the exposure of the new law in terms of the abolition of the former vocational studies of first grade (% students in FP-I), and the rate of students in the first two years of the new compulsory secondary education (% students in ESO-I)5. The way we assign indicators to individuals is as follows. For example, for an individual born in 1980, that turns 12 in 1992 and 13 in 1993, the exposure to ESO-I is constructed as the average of the fraction studying ESO in his region of residence in years 1992 and 1993. And we construct his exposition to FP-I as the average of the fraction studying FP-I in 1994 (when he turns 14) and 1995 (when he turns 15).

In this context, it is important to point out that the increase in the compulsory age from 14 to 16 years old favours our identification strategy since all the students during the period 1991-2000 period were obliged to study until 16 year old, either under the old track (LGE) or the new one (LOGSE). This means that we can observe all individuals from 14 years old to 16 for both educational systems.

6. Results

Table 3 shows the results from the estimation using individual data for men and women aged 18-24 at the time of the interview6. For each gender, we present two specifications which differ in the indicators used to proxy the degree of exposition to the LOGSE: (1) The fraction of students under the LOGSE at ages 12-17; and (2) the fraction of students under LOGSE at ages 13-14 (ESO-I). In both specifications we include the fraction of students enrolled under FP-I (which disappeared with the introduction of the LOGSE).
Table 3

Probability of being dropout (All sample)

 

Males

Females

 

(1)

(2)

(1)

(2)

Demographics

    

Born Abroad

0.1036***

0.1037***

0.0699**

0.0694**

 

(3.019)

(3.026)

(2.006)

(1.997)

Father present

-0.1899***

-0.1900***

-0.1624***

-0.1623***

 

(-22.795)

(-22.803)

(-19.794)

(-19.740)

Mother present

-0.1420***

-0.1420***

-0.2094***

-0.2095***

 

(-13.457)

(-13.482)

(-16.726)

(-16.731)

Father dropout

0.1918***

0.1919***

0.1165***

0.1165***

 

(21.240)

(21.219)

(14.037)

(14.044)

Mother dropout

0.0976***

0.0976***

0.0801***

0.0800***

 

(16.828)

(16.858)

(14.694)

(14.675)

Cohort effect

0.0925

0.1107

0.1291

0.0923

 

(1.290)

(1.496)

(1.574)

(1.186)

Labour Market and Business Cycle (regional)

Expected real wage

0.0156

0.0207

-0.0499***

-0.0492***

ratio, workers age 30

(0.410)

(0.574)

(-3.237)

(-2.997)

Share of employment

0.1821

0.0527

-0.5081

-0.3097

in construction

(0.478)

(0.134)

(-1.298)

(-0.821)

Unemployment rate

0.3080

0.3397

0.3940***

0.3831***

 

(1.295)

(1.429)

(3.361)

(3.078)

Growth rate

0.3840

0.3451

0.0625

0.1006

 

(1.421)

(1.248)

(0.236)

(0.378)

Educational Model

    

Educational

-0.0300

-0.0282

-0.0103

-0.0151

competences

(-1.345)

(-1.398)

(-0.332)

(-0.509)

Adequacy rate at 15 (%)

-0.0016

-0.0011

0.0008

0.0007

 

(-1.517)

(-1.138)

(0.491)

(0.417)

Incidence of the LOGSE

% studying LOGSE

0.1344

 

-0.1831**

 
 

(1.019)

 

(-3.226)

 

% studying FP-I

-0.0132

-0.0555

-0.2393**

-0.1562*

 

(-0.111)

(-0.494)

(-2.439)

(-1.680)

% studying ESO-I

 

0.1254***

 

-0.0827*

  

(2.745)

 

(-1.827)

R-squared

0.064

0.064

0.078

0.078

Observations

53738

53738

51884

51884

Individuals aged 18-24 interviewed between 1995 and 2010. Marginal effects.

NOTES: The difference between (1) and (2) relies in the set of educational indicators included in the specification. ***significant at 1%; **significant at 5%; *significant at 10%. Omitted variables: regional & cohort dummies. Marginal effects after probit estimation. Reported t-stats constructed from standard errors clustered by region.

First of all, it is noteworthy the fact that the estimates that account for the exposure to the LOGSE respond differently depending on the gender. On the one hand, we observe a positive and significant effect for men in column (2), since those who were exposed to the first stage of the LOGSE (ESO-I) are more likely to dropout. One very plausible explanation is that the abolishment of Lower Vocational studies or FP-I for children aged 14 to 16 years old that took place after the implementation of the LOGSE had a negative impact on the willingness to continue studying for men. This is because it is very likely that an important group of males that were less academic oriented had their more appropriate educational path closed. That is, boys that would have done much better in vocational studies were forced to follow a more academic track; they lost motivation and quit studies at 16 years old. Under the previous law (LGE) this type of students would have proceed Lower Vocational Studies until 16 years old and perhaps an important share would have continued until High Vocational Studies after the age of 16. For example in 1990-91 there were 266 thousand males enrolled in FP-I and 207 thousand females (39.1% and 31.9% of the 14-15 population, respectively).

On the other hand, results show that females under the LOGSE had a lower probability to dropout. The rationale of this finding for females is very different from that for men. For females, the negative correlation between the implementation of the LOGSE and dropout is probably due to the fact that less women tended to choose FP-I and hence were less affected by its abolition. Another plausible explanation could be that at the end of the 90s there was a change in the production model that affected men and women asymmetrically with more male human capital demanded in construction. In this sense, in the mid 90s the share of people employed in the construction sector was stable at 10%, but started to escalate in 1998, and by 2000 the ratio was at roughly 12%.

To that, the decreasing level of complexity in the curriculum required in secondary school in the LOGSE compared to the LGE (Delibes-Liniers (2006)) also might have contributed to relax the incidence of dropout after the introduction of the new law. This has made easier to complete secondary school and continue towards non-compulsory education. For males, however, it is possible that the simplification in the curriculum did not offset other characteristics of the LOGSE that reduced their incentives to study further.

With respect to other indicators of the exposition of the LOGSE, we find that the presence of lower vocational track (FP-I) has a small negative but non-significant effect on the dropout rate for males and a negative effect for females. One explanation is that the existence of FP-I under the LGE helped reducing the negative peer effects in the secondary school since it selected into the more academic track only those students who were more motivated, so the probability of completing not compulsory education is higher for them.

Overall, the average of effect of the LOGSE seems to be at least not positive. Averaging marginal effects in specifications 1 and 2 we obtain. Although the joint model is clearly rejected by the data (the complete set of results is available upon request),7 we want to stress that the average effect of the LOGSE variables (the % studying LOGSE and the % studying FP-I in specification 1 and the % studying ESO and the % studying FP-I in specification 2) is very close to zero. Alternatively, the effect of the fraction studying FP-I, is larger but also non-significant in the two specifications considered. Averaging estimated marginal effect over the two population leads to a similar message, since the estimated average marginal effect is a 0.011 increase in the probability of dropout after the introduction of the LOGSE. Summing up, the effect of the LOGSE is by no means positive.

Besides the specifications of the LOGSE, other factors were also important to explain the evolution of dropout. With regards to the demographic indicators, we observe that the pattern is similar across genders, although the magnitude is different. In this context, results clearly show that non-natives have a higher probability of dropout. Data also reveal the importance of the parental role since individuals whose parents were not present in the household were more likely to abandon further education, as well as individuals whose parents had been a dropout.

The regression also includes cohort effects to check whether there is a relationship between the relative number of individuals who have the same age and dropout to account for the possibility that perhaps more competition reduces returns to education and then the willingness to study longer (and hence increases dropout). However, we find weak evidence for this hypothesis since the estimates are positive but not significant.

Another indicator that may affect the decision of abandoning school is the perception of the opportunities given by the labour market. In this sense, we observe that the higher the expected earnings at the age of 30, the lower female dropout, although this relationship is insignificant for males. With regards to the production model, we observe that the larger the weight of the construction sector, the higher (although the effect is not significant) the probability to abandon further education for men, but the opposite for women8.

The small impact of employment opportunities in the construction sector (or low skill occupations in general), which may surprise the reader, can be due to the fact that our sample period (1991-1999) does not cover the core of the boom of the construction sector. In fact it is quite likely that the boom of the construction sector has stronger effect after year 2000, when the evolution of the Spanish construction sector started to clearly diverge from other European countries. This is line with the research by Aparicio (2010) who finds that the improvement in the labour market prospects of low educated workers motivated by the increases in employment and wages in the construction sector during the recent housing boom raised men’s propensity to dropout of high-school, relative to women.

Finally, regarding the educational model, we observe that men and women who studied in the regions with early competences in education tend to have lower incidence of dropout, but this effect is not significant.

6.1 Robustness check: excluding immigrants from the sample

In Table 4 we present the results obtained excluding immigrants (about 2 percent of the sample) since they may have not fully studied under the Spanish system or may have extra language difficulties. Although this fact did not play a crucial role in deterring the improving rates of dropout in the last decade, it will probably do over the next years since the proportion of foreigners attending compulsory secondary school will grow considerably (see Zynovieva et al, 2008).
Table 4

Probability of being dropout

 

Males

Females

 

(1)

(2)

(1)

(2)

Demographics

    

Father present

-0.1937***

-0.1938***

-0.1666***

-0.1665***

 

(-22.87)

(-22.88)

(-19.03)

(-18.99)

Mother present

-0.1580***

-0.1579***

-0.2159***

-0.2161***

 

(-16.52)

(-16.58)

(-17.41)

(-17.42)

Father dropout

0.1926***

0.1928***

0.1162***

0.1161***

 

(21.60)

(21.58)

(14.63)

(14.65)

Mother dropout

0.0982***

0.0983***

0.0797***

0.0797***

 

(17.38)

(17.42)

(16.29)

(16.26)

Cohort effect

0.0724

0.0913

0.1296

0.0893

 

(0.98)

(1.22)

(1.47)

(1.04)

Labour Market and Business Cycle (regional)

Expected real wage

0.0064

0.0122

-0.0437**

-0.0422**

ratio, workers age 30

(0.16)

(0.32)

(-2.56)

(-2.34)

Share of employment

0.1884

0.0553

-0.4503

-0.2404

in construction

(0.48)

(0.13)

(-1.12)

(-0.59)

Unemployment rate

0.3074

0.3439

0.3349***

0.3285**

 

(1.24)

(1.39)

(2.87)

(2.40)

Growth rate

0.3868

0.3449

0.0907

0.1278

 

(1.40)

(1.22)

(0.34)

(0.47)

Educational model

    

Competences in

-0.0346

-0.0330

-0.0054

-0.0111

education

(-1.42)

(-1.50)

(-0.17)

(-0.36)

Adequacy rate 15 (%)

-0.0015

-0.0011

0.0005

0.0005

 

(-1.44)

(-1.04)

(0.32)

(0.29)

Incidence of the LOGSE

    

% studying LOGSE

0.1402

 

-0.1911***

 
 

(1.01)

 

(-3.28)

 

% studying FP-I

-0.0164

-0.0600

-0.1756

-0.0823

 

(-0.13)

(-0.53)

(-1.61)

(-0.77)

% studying ESO-I

 

0.1364***

 

-0.0553

  

(2.91)

 

(-1.33)

R-squared

0.07

0.07

0.08

0.08

Observations

52531

52531

50393

50393

Individuals aged 18-24 interviewed between 1995 and 2010. Robustness checks. No immigrants. Marginal effects.

NOTES: The difference between (1) and (2) relies in the set of educational indicators included in the specification. ***significant at 1%; **significant at 5%; *significant at 10%. Omitted variables: regional & cohort dummies. Marginal effects after probit estimation. Reported t-stats constructed from standard errors clustered by region.

Note that the elimination of individuals born abroad does not alter the key conclusions of the analysis. The educational coefficients, that is, % studying LOGSE and the fraction studying ESO-I, remain significant for females and males. However, the coefficient of the fraction studying FP-I, that was significant at 5 percent in the whole sample, losses significance. One possible explanation could be that the positive effect of the existence of FP-I on continuing further education was relatively larger for the immigrant population.

6.2 Implications on the change in the educational law on the track choice

The introduction of the LOGSE and, in particular, the progressive removal of the lower vocational track (generally available for students aged 14-16 before the LOGSE) may have had consequences on the path (either vocational or academic) choice. In order to check whether this is the case, we estimate the probability of taking vocational studies after compulsory education.

We observe in Table 5 that the coefficient of lower vocational studies is positive and highly significant for females and positive but not significant at 5 percent for males (in fact, many of the important variables appear to be more significant for females than they are for males). The marginal effect is very large, especially for women.9 It implies that an increase of one percent in the fraction studying FP-I increases the probability of studying lower vocational studies by 0.41. Consequently, by eliminating the lower vocational track, the LOGSE largely reduced the fraction of women studying the vocational track. The effect for women can be quantified between 4 and 10 percentage points and for men between 2 and 5 percentage points, depending on the year taken as a reference.
Table 5

Track choice: the probability of taking vocational studies vs. the academic track after compulsory education (marginal effects)

 

Males

Females

 

(1)

(2)

(1)

(2)

Demographics

    

Born Abroad

-0.0464*

-0.0464*

-0.0793***

-0.0793***

 

(-1.81)

(-1.81)

(-5.54)

(-5.53)

Father present

-0.0633***

-0.0632***

-0.0555***

-0.0555***

 

(-7.25)

(-7.24)

(-7.87)

(-7.84)

Mother present

-0.0425***

-0.0426***

-0.1035***

-0.1036***

 

(-3.64)

(-3.67)

(-12.47)

(-12.45)

Father dropout

0.0632***

0.0633***

0.0503***

0.0503***

 

(7.03)

(7.07)

(5.75)

(5.73)

Mother dropout

0.0140

0.0140

0.0079

0.0080

 

(1.57)

(1.57)

(1.06)

(1.06)

Cohort effect

0.1600**

0.1740**

0.1644***

0.1691***

 

(2.10)

(2.15)

(4.34)

(3.82)

Labour Market and Business Cycle (regional)

Expected real wage ratio,

-0.0218

-0.0223

-0.0446***

-0.0432***

workers age 30

(-0.77)

(-0.85)

(-2.85)

(-2.74)

Share of employment

-0.3834

-0.5175

-0.4537

-0.4826

in construction

(-0.99)

(-1.36)

(-1.29)

(-1.35)

Unemployment rate

0.0805

0.0570

0.5190**

0.5319***

 

(0.35)

(0.23)

(2.55)

(2.64)

Growth rate

-0.0979

-0.1086

0.3611*

0.3422

 

(-0.30)

(-0.33)

(1.65)

(1.57)

Educational model

    

Competences in

0.0239

0.0296

0.0296

0.0298

Education

(0.62)

(0.84)

(0.88)

(0.99)

Adequacy rate 15 (%)

-0.0002

-0.0002

-0.0005

-0.0003

 

(-0.18)

(-0.16)

(-0.40)

(-0.26)

Incidence of the LOGSE

    

% studying LOGSE

0.1193

 

0.0333

 
 

(1.19)

 

(0.28)

 

% studying FP-I

0.2817*

0.2392

0.4093***

0.4059***

 

(1.81)

(1.56)

(3.38)

(2.88)

% studying ESO-I

 

0.0277

 

0.0511

  

(0.78)

 

(1.22)

R-squared

0.02

0.02

0.02

0.02

Observations

33988

33988

39941

39941

Full sample (18-24) that has completed compulsory education.

NOTES: The difference between (1) and (2) relies in the set of educational indicators included in the specification. ***significant at 1%; **significant at 5%; *significant at 10%. Omitted variables: regional & cohort dummies. Marginal effects after probit estimation. Reported t-stats constructed from standard errors clustered by region.

The rest of the results are by no means less important. Particularly interesting are the coefficients of father’s present and father’s dropout (that go in opposite directions), cohort size and the expected real wage ratio (between educational levels 3 and 1) at age 30. In this sense, individuals whose father was present have 6 percent less probability of choosing the vocational track while those whose father was a dropout have a probability of choosing the vocational track 6 percent higher. Likewise, belonging to a larger cohort increases the probability of selecting the vocational path by about 16 percent, and the higher the expected real wage of educated worker at age 30 the lower the probability of choosing the vocational track, although the latter coefficient is only significant for women.

7. Conclusions

Dropout rates in Spain are markedly high compared to other European countries. After two decades (the 70s and the 80s) of continuous falls they were roughly 30% since 1995 and only decreased further to around 25% with the recession. Obviously, this very high dropout rates may have strong consequences in the future labour opportunities of Spanish workers and may compromise the future growth path of the Spanish economy.

This paper explores the role of the educational law (LOGSE) from 1990, and other factors, in explaining why school dropout stopped its declining trend in the last two decades. In this paper we have argued that, by eliminating lower vocational studies, the law moved the educational system in the opposite direction than other educational reforms in Europe. Differences in the timing of implementation of the LOGSE across different regions are used to identify whether this law contributed in deterring the rate of improvement of the dropout rates observed in years before. Our empirical strategy makes use of the demographic information available in the EPA to control for both personal and family characteristics.

Results show a negative relationship between the LOGSE and the incidence of school dropout for men. It is of particular interest the fact that the effect is asymmetric by gender. Although a priori this result may be surprising, there are good reasons to think why this could be the case. On the one hand, we observe a positive and significant effect for men since those who were exposed to the LOGSE are more likely to dropout. One very plausible explanation is that the abolishment of FP-I (Lower Vocational studies) for children aged 14 to 16 years old that took place after the implementation of the LOGSE had a negative impact on the willingness to continue studying for men. This was because this sort of studies was more appealing for them. It is also possible that, for boys, mixing the good and the bad students in the same class (in the last two courses of the new track) had worst negative consequences than for girls. A complementary explanation of the differences across gender is the changes in the production model of the economy. That is, it is quite likely that the construction boom from the end of the 90s until the starting of the recession in 2007 contributed to further deter the improving rates of dropout observed in previous years (up to 1994). However, the 2000’s are out of our sample period, so we have to postpone the analysis of this issue to further research.

On these grounds, the government approved in July 2011 a new law for Vocational Studies aimed precisely at enabling the transition from secondary degree in ESO to professional programs for students who did not complete their degree. Hence, the new law facilitates the transition to specific programs for professional training. Until now, those were limited to those individuals aged 16 or older who did not complete ESO. Instead, under the new law, individuals aged 15 can have access to these programs for professional training and, if they succeed they can continue towards Vocational Studies. Similarly, the new law enables individuals moving from Medium Vocational Studies to High Vocational Studies without the need of completing secondary non-compulsory school (Bachillerato), and after taking a special course. For that, around 20-30% places in High Vocational Studies will be reserved. All these new measures will probably contribute in reducing the high incidence of dropout and are quite in line with our findings.

On the other hand, the paper also explored the choice of educational track. We observe that, after the introduction of the LOGSE, the elimination of the lower vocational track starting at age 14 has significantly reduced the choice of the vocational track for both genders, being the effect particularly strong for women.

Endnotes

1The definition of dropout rates is quite broad. In this paper, a dropout is an individual aged 18-24 with at most lower secondary degree and not in further education or training. According to the International Standard Classification of Education (ISCED), this corresponds to having completed at most ISCED 2.

2In 2008-2012 the incidence of early-school leavers decreased by seven percentage points. This was most likely driven by the recession since youngsters were less attracted by the pool of low-skills jobs that had been easily available before the start of the crisis.

3Undoubtedly, the Spanish production model, very specialized in low-skill low-educational sectors, also contributed to sustain a very high dropout rate. However, this alternative explanatory factor was more influential in the early 2000’s as documented in Felgueroso and Jiménez-Martín (2009).

4See Hasluck (2004), Hannan et al. (1996) and Raffe (2008) for further discussion on educational systems.

5We have also considered the fraction of students in the 3rd and 4th grade of ESO (highly complementary to the fraction studying FP-I). However, the inclusion of this variable does not seem to change the key evidences obtained from the analysis.

6Similar results are obtained when restricting the sample to individuals aged 18-21.

7The joint model hypothesis is rejected by the data. For example the LR of the test for specification 1 is 262.76, which is distributed as a χ 40 2 .

8The characteristics of the construction sector, mainly based on on-the-job training and gender bias jobs, help explain the differential effect we find between men and women.

9The relatively large effect for women can be explained by the importance of some (quasi-segregated) studies within the lower vocational track, such as the administrative, hairdressing or health related branches

Appendix

Construction of the variables for the microeconomic analysis

Dependent variable:

Dropout (a dummy variable that takes value 1 if the individual is a dropout and zero otherwise). An individual is considered to be a dropout if she/he is currently not studying (the variable SIDIV of the survey EPA is different from one) and her/his maximum level of education is primary education or first stage of basic education, age 16 (ISCED < =2).

Track choice: A dummy variable that takes 1 if the individual chooses the vocational track after having completed compulsory education.

The regression is estimated separately for women and men.

Covariates:

Three different measures of the degree of exposure to the LOGSE:

  • 12-17 years old LOGSE index: fraction of the 12-17 population under the LOGSE (ESO-I & ESO-II & Bachillerato).

  • 12-13 years old LOGSE index: fraction of the 12-13 population under the LOGSE (ESO-I).

  • FP-I: fraction of the 14-15 population studying FP-I.

Individual factors

  • Nationality: dummy one if born abroad (non-native).

  • Parental presence: dummy one if mother presence (mother) and dummy one if father presence (father).

  • Parental education: dummy one if mother was a dropout (mother dropout) and dummy one if father was a dropout (father dropout).

  • Cohort effect: a measure that combines the relative number of individuals who have the same age with respect to the region size (direct competition to the individual) and the relative number individuals who are 1 or 2 years around the individual’s age (with a weight of 2/3) reflecting the partial competition. This follows the normalisation of Welch (1979).

Regional and time factors

  • Measure for dropout earnings: the ratio between the expected earnings at age of 30 of educational level 1 and 3 (earnings gap).

  • The proportion of students of 15 years old that are studying in their corresponding year in percentage (adequacy rate at 15) at regional level.

  • Regional unemployment rates.

  • Regional employment share in construction.

  • Regional annual percentage change of the regional PIB.

  • Regional educational competences: a dummy one for those regions who have competences in education.

Declarations

Acknowledgements

Sergi Jiménez-Martín and Florentino Felgueroso thank financial help from the Spanish Ministry of Education, projects #ECO2008-06395-C05 and #ECO2011-30323-C03-02. We thank an anonymous referee, the editor, Antonio Ciccone, Antonio Cabrales, Oriol Aspachs-Bracons and participants at SEA 2011 in Madrid for helpful comments.

Responsible editor: Juan F Jimeno

Authors’ Affiliations

(1)
Universidad de Oviedo, Oviedo, and FEDEA
(2)
“la Caixa” Research
(3)
Universitat Pompeu Fabra and Barcelona GSE, Barcelona, and FEDEA

References

  1. Acemoglu D, Angrist JD: How large are human capital externalities? Evidence from compulsory schooling laws. NBER Macroecon Annu 2000, 2000: 9–59.View ArticleGoogle Scholar
  2. Angrist JD, Krueger AB: Does compulsory school attendance affect schooling and earnings? Q J Econ 1991, 16: 4.Google Scholar
  3. Angrist JD, Lavy VC: The Effect of High Stakes High School Achievement Awards: Evidence from a School-Centered Randomized Trial. IZA DP No. 1146; 2004.Google Scholar
  4. Aparicio A: High-School Dropouts and Transitory Labor Market Shocks: The Case of the Spanish Housing Boom. IZA DP No. 5139; 2010.Google Scholar
  5. Arcidiacono P: Ability sorting and the returns to college major. J Econ 2004, 121: 343–375. 10.1016/j.jeconom.2003.10.010View ArticleGoogle Scholar
  6. Arcidiacono P: Affirmative action in higher education: how do admission and financial aid rules affect future earnings? Econometrica 2005, 73: 1477–1524. 10.1111/j.1468-0262.2005.00627.xView ArticleGoogle Scholar
  7. Blanden J: “How much can we learn from international comparisons of intergenerational mobility?” CEE discussion papers, CEEDP0111: Centre for the Economics of Education. London, UK: London School of Economics and Political Science; 2009.Google Scholar
  8. Blondal KS, Adalbjarnardottir S: Parenting practices and school dropout: a longitudinal study. Adolescence 2009,44(176):729–749.Google Scholar
  9. Brunello G, Rocco L, Arigac K, Iwahashid R: On the efficiency costs of de-tracking secondary schools in Europe. Educ Econ 2010, 20: 1469–5782.Google Scholar
  10. Brzinsky-Fay C: Lost in transition? Labour market entry sequences of school leavers in Europe. Eur Sociol Rev 2007,23(4):409–422. 10.1093/esr/jcm011View ArticleGoogle Scholar
  11. Clark W: Delayed Transitions of Young Adults. Ottawa: Statistics Canada; 2007.Google Scholar
  12. Dearden L, Emmerson C, Frayne C, Meghir C: Conditional cash transfers and school dropout rates. J Hum Resour 2009,44(4):827–857. 10.1353/jhr.2009.0013Google Scholar
  13. Delibes-Liniers A: La gran estafa: El secuestro del sentido común en la educación. Madrid: Unisón editores; 2006.Google Scholar
  14. Digest E: Reducing the Dropout Rate through Career and Vocational Education: Overview. ERIC Digest No. 63; 1987. http://www.ericdigests.org/pre-925/rate.htmGoogle Scholar
  15. Duflo E: Schooling and Labour Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment. NBER Working Paper num. 7860; 2000.View ArticleGoogle Scholar
  16. Duncan B: Dropouts and the unemployed. J Polit Econ 1965, 73: 121–134. 10.1086/259001View ArticleGoogle Scholar
  17. Dustmann C: Parental background, secondary school track choice, and wages. Oxf Econ Pap 2004, 56: 209–230. 10.1093/oep/gpf048View ArticleGoogle Scholar
  18. Dynarski: Does Aid Matter? Measuring the Effect of Student Aid on College Attendance and Completion. NBER Working Paper No. 7422; 2003.Google Scholar
  19. Eckstein Z, Wolpin K: Youth Employment and Academic Performance in High School. CEPR Discussion Papers 1861, C.E.P.R. Discussion Papers; 1998.Google Scholar
  20. Egido I: La evolución de la Enseñanza primaria en España: organización de la etapa y modelo educativo. Tendencias Pedagógicas 1995, 1: 75–86.Google Scholar
  21. Felfe C, Nollenberger N, Rodriguez-Planas N: Can't Buy Mommy's Love? Universal Childcare and Children's Long-Term Cognitive Development: Evidence from a Natural Experiment" with Christina Felfe and Natalia Nollenberger. 2012. IZA discussion paper No. 7053Google Scholar
  22. Felgueroso F, Jiménez-Martín S: The New Growth Model: How and with Whom?. Working Papers 2009–39, FEDEA; 2009.Google Scholar
  23. Field S, Kuczera M, Pont B: Structures and Pathways. In Education and Training Policy: No More Failures: Ten Steps to Equity in Education. Volume 15. OECD; 2007:55–82.Google Scholar
  24. Gangl M: Returns to Education in Context: individual education and transitions outcomes in European labor markets. In Transitions from Education to Work in Europe: the Integration of Youth into EU Labour Markets. Edited by: Müller W, Gangl M. New York: Oxford University Press; 2003:156–185.View ArticleGoogle Scholar
  25. Garrouste C: 100 years of educational reforms in Europe: a contextual database. MPRA Paper no. 31853; 2011.Google Scholar
  26. Hannan DF, Raffe D, Smyth E: Cross-national Research on School to Work Transitions: an Analytical Framework. OECD, Paris: Commissioned by the OECD Secretariat; 1996.Google Scholar
  27. Harmon C, Walker I: Estimates of the economic return to schooling for the UK. Am Econ Rev 1995,85(5):1278–1296.Google Scholar
  28. Hasluck C: The Net Costs to Employers of Modern Apprenticeships. In Apprenticeship: an Historical Re-invention for a Post Industrial World, Proceedings of the Conference. Edited by: Roodhouse S, Hemsworth D. London: University Vocational Awards Council; 2004.Google Scholar
  29. Haywood BJ, Tallmadge GK: Dropout Prevention and Vocational Education. Research Triangle Institute; 1995.Google Scholar
  30. Heckman JJ, Lafontaine PA: The American High School Graduation Rate: Trends and Levels. NBER Working Paper No. 13670; 2007.View ArticleGoogle Scholar
  31. Keane MP, Wolpin KI: The career decisions of young men. J Polit Econ 1997,105(3):473–522. University of Chicago Press 10.1086/262080View ArticleGoogle Scholar
  32. Lacasa JM: El efecto LOGSE y otros cuentos. Instituto FORMA de Investigación Educativa; 2006.Google Scholar
  33. Lacuesta A, Puente S, Villanueva E: The schooling response to a sustained increase in low-skill wages: evidence from Spain 1989–2009. Banco de España WP 1208; 2012.Google Scholar
  34. Lleras-Muney A: The Relationship Between Education and Adult Mortality in the United States. NBER Working Paper No. 8986; 2002.View ArticleGoogle Scholar
  35. Lochner L, Moretti E: The effect of education on crime: evidence from prison inmates, arrests, and self-reports. Am Econ Rev, Am Econ Assoc 2004,94(1):155–189. 10.1257/000282804322970751View ArticleGoogle Scholar
  36. López-Mayan C: Demand for Post-compulsory Education: The Choice Between Academic and Vocational Tracks. CEMFI; 2010. Job Market PaperGoogle Scholar
  37. Meer J: Evidence on the returns to secondary vocational education. J Educ Work 2007, 26: 559–573.Google Scholar
  38. Mora T, Escardíbul J-O, Espasa M: The effects of regional educational policies on school failure in Spain. Revista de Economía Aplicada 2010,18(54):79–106.Google Scholar
  39. OECD: Raising Education Outcomes. In OECD Economics Surveys. Volume 19. Spain: OECD; 2008:73–106.Google Scholar
  40. Oreopoulos P: Do dropouts drop out too soon? Wealth, health and happiness from compulsory schooling. J Public Econ, Elsevier 2007,91(11–12):2213–2229. 10.1016/j.jpubeco.2007.02.002View ArticleGoogle Scholar
  41. Pereita C, Pastor M: The primary school dropout in Spain: the influence of family background and labour market conditions. Educ Econ 2000,2(88):157–168.View ArticleGoogle Scholar
  42. Plank S, DeLuca S, Estacion A: Dropping out of High School and the Place of Career and Technical Education: A Survival Analysis of Surviving High School. St. Paul, MN: National Research Center for Career and Technical Education; 2005.Google Scholar
  43. Raffe D: The concept of transition system. J Educ Work 2008,4(21):277–296.View ArticleGoogle Scholar
  44. Schultz TP: in Mexico School Subsidies for the Poor: Evaluating the Mexican Progresa Poverty Program. Yale University Center Discussion Paper No. 834; 2003.Google Scholar
  45. Welch F: Effects of cohort size on earnings: the baby boom babies’ financial bust. J Polit Econ 1979, 87: s65-s97. 10.1086/260823View ArticleGoogle Scholar
  46. World Bank: Expanding Opportunities and Building Competencies for Young People: A New Agenda for Secondary Education. Washington: World Bank; 2005.View ArticleGoogle Scholar
  47. Zynovieva N, Felgueroso F, Vazquez P: Immigration and students’ achievement in Spain. Fedea DP 2008–37; 2008.Google Scholar

Copyright

© Felgueroso et al.; licensee Springer. 2014

This article is published under license to BioMed Central Ltd. 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 credited.