You are on page 1of 23

Rethinking Vocational Education in the Philippines: Does It Really Lead to Higher

Wages?
Author(s): Rosechin Olfindo
Source: Journal of Southeast Asian Economies , Vol. 35, No. 1, Special Issue:
Commemorating the 50th Anniversary of the ISEAS – Yusof Ishak Institute (April
2018), pp. 79-100
Published by: ISEAS - Yusof Ishak Institute

Stable URL: https://www.jstor.org/stable/10.2307/26539199

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2307/26539199?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

ISEAS - Yusof Ishak Institute is collaborating with JSTOR to digitize, preserve and extend
access to Journal of Southeast Asian Economies

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
Journal of Southeast Asian Economies Vol. 35, No. 1 (2018), pp. 79–100 ISSN 2339-5095 print / ISSN 2339-5206 electronic

DOI: 10.1355/ae35-1g

Rethinking Vocational Education in


the Philippines
Does It Really Lead to Higher Wages?

Rosechin Olfindo

Vocational education is often seen as a means of enhancing the earning potential of


disadvantaged workers — those with lower levels of general formal education, or without
adequate skills to integrate into the labour market. International evidence on the effects of
vocational education on earnings is mixed. An earlier study on the Philippines indicated that
this type of education can increase the wages of low-educated workers. However, by using
a more recent household survey and employing alternative statistical techniques, this paper
finds contrasting evidence. The new estimates indicate that workers who obtained vocational
education do not earn significantly more than those who did not. The differences in estimates
imply that the wage effects of vocational education among Filipino workers remain unclear.
Therefore, caution must be exercised while making policies that aim to promote vocational
education as an alternative to general formal education.

Keywords: Vocational education, wage effects, Philippines.

Article received: April 2017; Revised: December 2017; Accepted: December 2017

1. Introduction
Many economists believe that the accumulation of human capital — in the form of education — increases
wages because workers acquire skills and knowledge while in school, which increase their productivity
(Schultz 1961; Becker 1962; Mincer 1974). Many empirical studies have pointed out the superiority
of education over other determinants of earning differentials (e.g., Psacharopoulos 1972, 1985, 1994;
Psacharopoulos and Patrinos 2004; Peet, Fink and Fawzi 2015). Education also serves as a signal of
workers’ innate abilities as well as other unobservable productivity-related characteristics (Phelps 1972;
Arrow 1973; Spence 1973). While these studies typically refer to general formal education, there is mixed
evidence regarding the role of vocational education in enhancing the earning prospects of workers.

Rosechin Olfindo is Senior Economist at Oriental Consultants Global Co. Ltd., Tokyo Opera City Tower 9F, 20-2,
Nishishinjuku 3-chome, Shinjuku-ku, Tokyo, 163-1409, Japan; and Non-resident Research Fellow at Ateneo School
of Government, Quezon City, Philippines; email: rosechin@post.harvard.edu

© 2018 ISEAS – Yusof Ishak Institute 79

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 79 3/4/18 10:24 AM
80 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

In most countries, vocational education forms an integral part of the education system as a supplier of
job-specific skills. It aims to link workers to available jobs by providing them the skills that employers
seek but are unable to find. Vocational education is also viewed as a means of improving the earning
prospects of workers with low levels of general formal education. By providing them the specific skills
that are needed in the labour market, either as self-employed or wage workers, vocational education serves
as a social intervention to facilitate the labour market inclusion of vulnerable workers (Psacharopoulos
1997). In recent years, governments around the world have renewed their interest in investing in vocational
education in light of rapid technological changes, which demand specific types of technical skills that may
not be readily provided by the general education system (UNESCO 2012; World Bank 2012).
A number of empirical studies have shown favourable results, indicating that vocational education
yields higher wages than obtaining general education alone (Moenjak and Worswick 2003; Sakellariou
2003; Bishop and Mane 2004; El-Hamidi 2006; Meer 2007; Almeida et al. 2015). Other studies have
qualified these positive wage effects, as they depend on the types of vocational courses obtained (Neuman
and Ziderman 1991) or the level of general education already attained by the workers (Kahyarara and Teal
2008). Some studies have shown mixed results, suggesting that the wage effects of vocational education
are comparable to those of general education (Lechner 2000; Riboud, Savchenko and Tan 2007; Malamud
and Pop-Eleches 2010; Newhouse and Suryadarma 2011; Tripney and Hombrados 2013). The variation in
results reflect, for the most part, differences in: specifications; quality of data; and institutional structures
underlying the vocational education system in the countries under study (for review, see Bennel 1996;
World Bank 2012).
Some researchers, however, are less enthusiastic about vocational education (Bennel 1996;
Psacharopoulos 1987, 1997; Psacharopoulos and Patrinos 1993; Horowitz and Schenzler 1999). They
argue that private returns to the general education track are higher than the vocational education track. In
addition, although vocational education may increase wages, this effect does not last over time (Hawley
2003; Hanushek et al. 2016). The concept-based, rather than skill-based, knowledge makes workers more
adaptable and capable of acquiring new skills to meet changing labour market needs. Not only are these
effects less desirable at the worker level, they also make the country’s labour force less flexible — which
may lead to slower economic growth (Krueger and Kumar 2004).
Thus far, the claim that vocational education leads to higher wages remains an empirical question.
Country studies are important inputs in the debate, as they provide better understanding of specific
contexts at which vocational education leads to better labour market outcomes. The Philippines is an
interesting case for analysing the wage effects of vocational education for the following reasons:

1. It is educationally advanced as a developing country, with about a quarter of the labour force having
tertiary education (Hossain and Psacharopoulos 1994);
2. Returns to higher education remain high in the country despite significant expansion of education
over the years (di Gropello, Tan and Tandon 2010);
3. The labour market has been found to have substantial tertiary “diploma effects”, whereby workers
who completed tertiary formal education earn disproportionately higher wages (Hossain and
Psacharopoulos 1994; Schady 2003; Olfindo 2018); and
4. Post-secondary vocational education is stigmatized by Filipinos as an inferior track to formal tertiary
education, because the former is often taken up by individuals who cannot afford, or are unable, to
take degree courses (UNESCO 2010).

Prior estimates of wage effects of vocational education in the Philippines using the “Labour Force
Survey” indicated positive and significant effects among workers with elementary and secondary
education (Choi 2016). Intuitively, however, these results are rather surprising, given the country context

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 80 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 81

described above. By using a newer Labour Force Survey, and employing two alternative methods —
(1) ordinary least squares estimation of the Mincerian wage function; and (2) propensity score matching
approach — to estimate the wage effects, this paper presents new estimates. These estimates provide
less support for the view that workers with vocational education earn higher wages than workers who
pursued only general formal education, controlling for individual and job characteristics for which data
is available.
The main argument of this paper is that empirical research on the effectiveness of vocational education
is still developing. Whether or not obtaining vocational education leads to higher wages remains
inconclusive. The differences in estimates in this paper and those of the earlier study suggest that the
wage effects of vocational education in the Philippines largely depend on the quality of data and methods
used. Hence, caution must be exercised while making policy prescriptions. More research and better data
quality are necessary in order to gain insights on how vocational education can help in achieving social
objectives.
The following section presents a brief background of vocational education in the Philippines. Subsequent
sections discuss the data and its limitations, empirical methods, and results. The final section presents the
conclusion.

2. Vocational Education in the Philippines: A Brief Background


In the Philippines, the education and training system consists of basic education (kindergarten, elementary
and secondary levels), vocational education, and higher education (tertiary and post-baccalaureate levels).
Vocational education courses vary in duration, content, method, delivery and target clientele. Some
courses are short-term, which last for a few days or weeks, whereas long-term courses can last up to
three years. They also cover a variety of subjects related to agriculture, industry and services sectors.
While some courses are highly technical in nature, such as those based on information technology and
construction sectors, others are more service-oriented, such those related to tourism and food sectors.
Vocational courses are delivered by a wide variety of public and private providers, including vocational
institutes, private companies, and local community organizations.
Vocational education in the Philippines is primarily considered a type of post-secondary non-tertiary
education, but there are courses that also cater to individuals with lower education levels. A majority
of such courses lead to a National Certificate, which is regarded as a proof of possession of a certain
qualification. This certificate has four levels of qualifications, which are obtained by completing vocational
courses that develop skills to perform:

1. routine and predictable tasks (Level I);


2. a wide range of functions involving known routines and procedures (Level II);
3. a wide range of tasks and work with a complexity of choices in a given function (Level III); and
4. a wide range of applications and responsibilities that are complex and non-routine (Level IV).

Courses that lead to a National Certificate require the completion of secondary education and are
delivered by vocational institutes or private companies. Vocational courses that cater to individuals
who did not complete secondary education lead to a Certificate of Competency, which is equivalent to
completing only a sub-level of a National Certificate.
Among the individuals who take vocational courses, about 80 per cent take courses that lead to a
National Certificate. However, most of them only opt for lower levels of qualifications (Level I or II).
The remaining 20 per cent take shorter, mostly service-oriented, courses that lead to a Certificate of
Competency. It may be noted that among existing courses, a majority of enrollees are in the tourism (travel

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 81 3/4/18 10:24 AM
82 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

guides, hotel and accommodation, health and wellness), information technology (computer programming,
call centre), and electronics (appliance repair) sectors.1
Based on the Labour Force Survey of 2015, individuals who obtained vocational courses account for
only about 5 per cent of the working age group in the Philippines. This proportion is quite consistent
across several rounds of the survey. About 70 per cent of workers with vocational education are employed.
Many of those who have vocational education but do not participate in the labour force believe that
there is no work available for them. Among the employed workers, 50 per cent are employed in private
establishments, while 30 per cent are self-employed. They are employed mainly in service industries,
most notably in trade and repair, and transport and storage services.
In addition to the Labour Force Survey, the Philippine government also undertakes a survey of the
employment status of vocational education graduates. The latest survey was conducted in 2013. Among
the 15,709 surveyed graduates, 70 per cent participated in the labour force, of which only 65 per cent
were employed (TESDA 2014). The labour force participation rate among vocational education graduates
is slightly higher than the national average of 64 per cent, as estimated from the Labour Force Survey.
Given these low figures, a number of studies have raised concerns about the quality of vocational
education in the Philippines and its relevance to labour market needs (World Bank 2016; Paqueo and
Orbeta 2017; Orbeta and Esguerra 2016; Orbeta and Abrigo 2012; di Gropello, Tan and Tandon 2010).
For instance, only six in ten individuals who take vocational education courses are certified as being
competent in their field. In more technical fields (like information technology), this proportion is even
lower. Moreover, among those who found employment, only 30 per cent report that their jobs are related
to the skills they acquired from vocational education.

3. Data
The data used to estimate the wage effects of vocational education among Filipino workers has been
drawn from the Labour Force Survey in April 2015. The Philippine Statistical Authority conducts the
household survey once every quarter of the year (in January, April, July and October). It includes a
comprehensive set of questions on the respondent’s labour market status (employed, unemployed, or not
in the labour force), as well as details about wages, number of hours worked, occupation, highest formal
education level obtained, and whether or not the respondent received vocational education. The survey
covered a total of 201,495 individuals belonging to 43,270 households. Prior estimates of the wage effects
of vocational education (Choi 2016) were based on the Labour Force Survey of 2014.
About 70 per cent of respondents belong to the working age population (15 years and older), of which
60 per cent are employed. The employed workers are classified into two groups: wage workers (60 per
cent) and non-wage workers (40 per cent). The former are employed either in private establishments
(77 per cent), government (14 per cent), private households (8 per cent), or family-operated business
(1 per cent). Non-wage workers are those who are employers in their own businesses (with or without
employees) or unpaid family workers.
The analysis in this paper is limited to wage workers in private establishments only. This is because of a
number of reasons. First, the survey does not provide earnings for the non-wage workers. Moreover, wage
workers in government are paid according to an officially determined pay scale, which may not reflect the
wage rates in a competitive labour market. Third, wage workers in private households or family-operated
businesses are usually paid with non-monetary compensation such as board and lodging, transportation,
paid leave, educational support and other non-formal modes of payments, all of which may not be reflected
in the reported earnings. It might be useful to note here that the analysis in Choi (2016) covered all wage
workers. This may have created a downward bias in estimating the mean wages because wage workers in
private households and family-owned businesses tend to report lower wages — not taking accounting for

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 82 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 83

the non-monetary compensation that they receive. However, for comparison purposes, the estimates using
all wage workers are also reported in this paper.
Considering the above factors, the sample size falls to only 28,293 workers, of whom 1,781 workers
(or about 6 per cent of the sample) obtained vocational education. A majority of workers who obtained
vocational education had completed secondary education and a considerable proportion had attained
tertiary education (incomplete or complete). This profile reflects the fact that vocational education in
the Philippines is primarily a post-secondary intervention. A substantial share of workers who obtained
vocational education are young (aged 15–24 years), and account for 60 per cent share. Table 1 shows the
descriptive statistics of the sample.
Figure 1 shows the wage distributions of workers with and without vocational education, or technical
and vocational education and training (TVET). Without controlling for other factors, the wages of workers
with vocational education are higher than that of workers without it. This pattern is true for both male
and female workers. However, if other factors like education level and length of work experience are
considered, the wage distribution may be different.
Only recent rounds of the Labour Force Survey have information on whether or not the respondent
obtained vocational education, and this new addition to the questionnaire is still being developed. For
instance, the information on when the vocational education was obtained — whether it was before or after
the formal general education — is not provided. Additionally, the type of vocational courses obtained are

TABLE 1
Summary Statistics

With Without
All Workers Vocational Vocational
Education Education
Hourly wage (pesos)
 Mean 44 44 43
 Median 38 39 38
  Standard deviation 36 35 36
No. of hours worked in one week
 Mean 42 43 42
 Median 48 48 48
  Standard deviation 15 15 15
Age
 Mean 34 33 34
 Median 32 30 32
  Standard deviation 13 11 13
No. of years of formal education*
 Mean 19 11 19
 Median 10 10 10
  Standard deviation 13 12 13
Notes: * Years 1 to 6 refer to elementary; Years 7 to 10 refer to secondary; and Years 11 to
14 refer to tertiary.
Source: Labour Force Statistics 2015; Author’s calculations.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 83 3/4/18 10:24 AM
84 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

Female workers
Wage Distribution of Workers with and without Vocational Education
FIGURE 1

Source: Labour Force Statistics 2015; Author’s calculations.


Male workers

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 84 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 85

also self-reported, and not classified according to their official names and levels of competency. Hence,
there is limited information on content, duration, and method of delivery of the vocational education
attained by the respondents. Further research is warranted once more information is available in the survey.

4. Empirical Methods
To estimate the wage effects of vocational education, this paper first employs ordinary least squares (OLS)
estimation of the Mincerian wage function (Mincer 1974), which relates the differences in hourly wages to
different observable individual and job characteristics of the workers. The wage function includes a binary
variable that indicates whether or not the worker obtained vocational education. The OLS estimates are
then validated using propensity score matching approach (Rosenbaum and Rubin 1983), which estimates
the difference in the expected hourly wage of workers with vocational education (treatment group) and
that of workers without vocational education who display the same observable characteristics as those
with it (the control group). The details of these two methods are discussed below.

4.1 Ordinary Least Squares Estimation


The reduced-form of the Mincerian wage function is specified in Equation (1) as follows:

logWi = a + b1Vi + Σnj=2bjXi, j–1 + ui (1)

where the dependent variable is the natural log of earnings per hour W and the independent variables
include: binary variable V that takes a value of 1 if worker i obtained vocational education, 0 otherwise; X
that represents several characteristics of worker i; and ui stands for the error term. The worker characteristics
include the number of years of experience and its square, dummy variables that represent the highest
general education obtained, and dummy variables that represent individual and job characteristics of
worker i.
Since the information on worker’s experience is not provided in the Labour Force Survey, it is computed
using the worker’s age less the approximate age when the worker finished schooling. For instance, a
worker whose highest formal education attained was complete secondary education was likely to finish
schooling at the age of sixteen. Hence, a twenty-year-old survey respondent with similar educational
background is assumed to have four years of work experience at the time of the survey. The individual
characteristics comprise the worker’s gender and marital status, while the job characteristics include the
worker’s sector of employment, occupation, and location.
To account for the relative wage effects of vocational education depending on the level of formal
education obtained by the worker, the wage function in equation (1) is extended to allow for the interaction
between V and dummy variables representing the highest formal education. This extended equation is
specified in equation (2) as follows:

logWi = a + b1Vi + Σ4j=2bjVi, j–2(CSi + ITi + CTi) + Σnj=5bjXi, j–4 + ui (2)

where the dummy variables CS, IT, and CT represent complete secondary, incomplete tertiary, and
complete tertiary education, respectively (incomplete secondary education or lower as reference); and
other variables as in equation (1).
The coefficients of variable V as well as the interaction terms with V are of particular interest. A
positive and significant coefficient of the dummy variable V represents the impact of vocational education
on wages among workers with incomplete secondary education or lower, while a positive and significant

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 85 3/4/18 10:24 AM
86 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

coefficient of the dummy variable V with CS, IT, and CT, represents the relative wage effects of vocational
education among workers with complete secondary, incomplete tertiary, and complete tertiary education,
respectively.
In order to control for individual heterogeneity, wherein a worker’s innate ability may be creating
an upward bias in the coefficients on vocational education, equations (1) and (2) are estimated with an
additional dummy variable, which proxies this ability using their father’s education (1 for having a father
who completed at least secondary education). However, this procedure reduces the sample size because
only the workers who reported to be the children in the household roster (son or daughter) are included
in the estimation. Considering the additional equations with father’s education dummy, there are a total
of four equations to be estimated using the OLS method. In addition, equations (1) and (2) are estimated
using a larger sample covering all wage workers.

4.2 Propensity Score Matching


The propensity score matching (PSM) approach is a method that is frequently used in impact evaluation
studies to estimate the average effect of a particular programme among the participants. For this paper,
PSM attempts to compare the wages of workers with vocational education (treatment group) and those
without (control group), but limits the estimation only to workers who display the same observable
characteristics (Rosenbaum and Rubin 1983). The control group is created in two steps. First, a propensity
score is assigned to each observation by estimating the likelihood of obtaining vocational education
using a logit regression. Second, the observations to be included in the control group are selected using
the nearest-neighbour matching method, or those with propensity score nearest to that of a treatment
observation and fall within a certain range, or the region of common support. The difference between the
expected hourly wage of the treatment group and the statistical control group represents the estimated
wage effects of vocational education.
To limit the wide variations in the covariates between treatment and control groups, the propensity
score method is applied separately across four groups of workers, batched together according to highest
level of formal general education obtained. These groups are: incomplete secondary or lower; complete
secondary; incomplete tertiary; and complete tertiary. For each group, the logistic regression to estimate
the propensity score for each observation in the sample is specified in equation (3):
exp(xb)
Pr[ Vi = 1 | X ] = (3)
1 + exp(xb)
where the dependent variable is a binary variable V and the independent variables (represented by X) are
the same as in equation (1) in the OLS method.
After estimating the propensity scores, the observations for each of the four groups are stratified into
blocks, and a test for balancing property is performed for each block. If the block satisfies the test,
it indicates that the distance in the marginal distributions of the covariates between the workers with
vocational education and those without it is minimal. Once the control group is formed, the expected
hourly wage of the treatment group is compared with that of the control group.2

5. Results
Estimates using the OLS and propensity score matching methods indicate that workers with vocational
education are not likely to earn higher wages than workers who only obtained general education.
Estimates of significant wage effects of vocational education are found among workers with incomplete
tertiary education using the propensity score matching approach, but these effects are negative. These

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 86 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 87

results are in contrast with prior estimates of wage effects of vocational education in the Philippines (Choi
2016), where positive and significant impacts were found among workers with elementary and secondary
education. A more detailed discussion of the results is presented below.

5.1 Ordinary Least Squares Estimation


Table 2 presents the estimation results for equations (1) and (2). Annex Table A1 presents the successive
introduction of variables into equation (2). The estimated coefficients for the variables representing
individual and job characteristics have the expected signs. They also reveal the convexity in the wage returns

TABLE 2
Estimated Coefficients in Regression of Log of Hourly Wage

Equation (1) Equation (2)


Variables Estimated Standard Estimated Standard
Coefficient Errors Coefficient Errors
a –2.787*** 0.011 –2.786*** 0.011
V –0.024*** 0.011 –0.055*** 0.046
Experience –0.010*** 0.080 –0.010*** 0.000
Experience2 (× 100) –0.015*** 0.001 –0.015*** 0.001
Complete secondary –0.151*** 0.007 –0.147*** 0.007
Incomplete tertiary –0.243*** 0.011 –0.253*** 0.011
Complete tertiary –0.477*** 0.011 –0.485*** 0.011
V*CS –0.000*** 0.048
V*IT –0.144*** 0.057
V*CT –0.201*** 0.063
Male –0.172*** 0.006 –0.173*** 0.006
Married –0.071*** 0.006 –0.071*** 0.006
Industry –0.150*** 0.009 –0.149*** 0.009
High-skill services –0.070*** 0.011 –0.070*** 0.011
Low-skill services –0.197*** 0.012 –0.196*** 0.012
Professional –0.590*** 0.014 –0.586*** 0.014
Technician –0.333*** 0.017 –0.330*** 0.017
Clerk –0.307*** 0.013 –0.304*** 0.013
Operator –0.169*** 0.008 –0.168*** 0.008
Other service workers –0.060*** 0.011 –0.061*** 0.011
Urban location –0.202*** 0.006 –0.202*** 0.006
R-squared 10.4165*** 10.4172***
F-test 1187.21*** 1011.82***
No. of observations 128,293*** 128,293***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 87 3/4/18 10:24 AM
88 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

to education in the Philippines, i.e., workers with higher levels of general education earn significantly
higher wages. Using equation (1), the estimated coefficient for the binary variable V is significantly
different from zero, which suggests that workers who obtained vocational education are likely to earn
slightly more than workers who only obtained general education. This suggests that obtaining vocational
education generally gives workers a labour market advantage, but this advantage is still inferior compared
to the wage effects of obtaining higher general education.
Using equation (2), however, the estimated coefficient for the binary variable V shows no significance,
while the estimated coefficients for the interaction terms show significance only among workers with
incomplete tertiary and complete tertiary education, but these effects are negative. This implies that the
wage effects of vocational education vary by the formal education level already attained by the worker.
In particular, among workers with tertiary education, those with vocational education are likely to earn
substantially less than those without it.
So far, the above estimates have not controlled for individual heterogeneity using father’s education
as a proxy. Table 3 shows the estimates using equations (1) and (2), but only among a sub-sample of
workers who reported to be sons or daughters in the household roster. The estimated coefficients for the
binary variable V and its interaction terms with the general education obtained by the workers are no
longer significant. These findings suggest that, if individual’s ability is controlled for (proxied by father’s
education), no significant earning difference exists among workers who obtained vocational education and
those who only obtained general education. Annex Tables A2 and A3 show the results using the sample of
all wage workers, displaying similar patterns on the wage effects of vocational education.

5.2 Propensity Score Matching


The first step of this method is to estimate the propensity scores for all observations using the logistic
regression as specified in equation (3). The estimated coefficients are shown in Table 4 for each group of
workers. Across education groups, there seems to be no definite pattern that can predict the likelihood of
obtaining vocational education. Generally, workers with incomplete secondary or lower education, those
employed in services sector, or working as technicians or operators, are more likely to obtain vocational
education. A slightly different pattern is observed among workers with secondary education, whereby
individual characteristics and occupation are better predictors of obtaining vocational education than
the sector of employment. Males, married workers and those employed in higher skill occupations are
also more likely to obtain vocational education. Meanwhile, many variables did not show significance in
estimating the propensity score among workers with tertiary education.
The next step is to match each treated observation with one or more control observations using the
nearest-neighbour matching method within a region of common support. Table 5 presents the t-test results
of the differences in means for the variables before matching, while Table 6 presents the t-test results after
matching. Although many variables that were previously statistically different between workers with and
without vocational education turned out to have the same means after matching, some variables remain
significantly different between the two groups. This implies that data quality can be improved to obtain
a better matching.
Table 7 shows the estimated ATT (see Imbens 2004) for the four groups of workers. The test for
balancing property was satisfied by all groups, except for the one with complete secondary education.
However, when the analysis was limited to children only, the test for balancing property was satisfied for
all four groups. The estimated ATT for workers with incomplete secondary education or lower was found
to be statistically insignificant. This result is consistent with the OLS estimates showing no statistically
significant effect of vocational education among workers with lower education levels. Also, given the
wide variations in individual and job characteristics of workers with vocational education in the treatment

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 88 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 89

TABLE 3
Estimated Coefficients in Regression of Log of Hourly Wage, Children Only

Equation (1) Equation (2)


Variables Estimated Standard Estimated Standard
Coefficient Errors Coefficient Errors
a –2.799*** 0.017 –2.798*** 0.017
V –0.000*** 0.017 –0.040*** 0.085
Experience –0.013*** 0.001 –0.013*** 0.001
Experience2 (× 100) –0.019*** 0.004 –0.019*** 0.004
Complete secondary –0.160*** 0.012 –0.154*** 0.012
Incomplete tertiary –0.238*** 0.017 –0.253*** 0.017
Complete tertiary –0.457*** 0.018 –0.461*** 0.018
V*CS –0.081*** 0.088
V*IT –0.139*** 0.099
V*CT –0.062*** 0.107
Male –0.113*** 0.010 –0.114*** 0.010
Married –0.076*** 0.011 –0.076*** 0.011
Educated father –0.149*** 0.010 –0.148*** 0.010
Industry –0.131*** 0.015 –0.130*** 0.015
High-skill services –0.024*** 0.018 –0.024*** 0.018
Low-skill services –0.161*** 0.020 –0.159*** 0.020
Professional –0.474*** 0.023 –0.473*** 0.023
Technician –0.274*** 0.027 –0.276*** 0.027
Clerk –0.277*** 0.020 –0.276*** 0.020
Operator –0.172*** 0.014 –0.172*** 0.014
Other service workers –0.072*** 0.017 –0.072*** 0.017
Urban location –0.191*** 0.009 –0.191*** 0.009
R-squared 0.4470*** 0.4479***
F-test 517.35*** 445.00***
No. of observations 11,540*** 11,540***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.

group, only a small number of workers were found to have nearly equivalent characteristics. Re-estimating
the ATT for this group, but only among the children who reported father’s education, yields the same
results. Annex Table A4 shows the results using all wage workers.
Among workers with complete secondary education, the estimated ATT indicates that vocational
education is likely to increase wages by about 5 per cent. The estimate is statistically significant and differs
from the OLS estimates as shown in Table 4. However, the test for balancing property was not satisfied
by this group. When the same estimation procedure is applied among the sub-sample of workers who are
children in the household roster, the test for balancing property was satisfied, but the ATT estimate shows
no significance.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 89 3/4/18 10:24 AM
90

TABLE 4

18-J03487 JSEAE 07.indd 90


Logistic Regression for the Propensity Score

Group 1:
Group 2: Group 3: Group 4:
Incomplete Secondary
Variables Complete Secondary Incomplete Tertiary Complete Tertiary
or Lower
Est. coeff. Std. error Est. coeff. Std. error Est. coeff. Std. error Est. coeff. Std. error
Constant –5.779 0.512*** –2.792 0.139*** –2.074 0.196*** –2.063 0.295
Experience 0.006 0.032 0.016 0.010 0.026 0.013** 0.019 0.014
Experience2 (× 100) –0.015 0.053 –0.056 0.022** –0.074 0.031** –0.036 0.034
Male –0.498 0.269* 0.520 0.073*** 0.361 0.089*** 0.290 0.086***
Married 0.287 0.263 –0.128 0.072* –0.074 0.090 0.009 0.094
Sector of employment:
Employed in industry 0.316 0.403 0.039 0.130 –0.118 0.197 0.214 0.289
Employed in high-skill services 0.836 0.437* 0.224 0.140 0.127 0.199 0.145 0.294
Employed in low-skill services 1.114 0.533** 0.249 0.154 0.237 0.209 0.063 0.294
Occupation:
Employed as professional 1.173 1.044 0.924 0.192*** –0.175 0.258 –0.513 0.182***

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
Employed as technician 1.372 0.797* 1.206 0.157*** 0.227 0.179 –0.435 0.220**
Employed as clerk –– –– 0.874 0.131*** –0.004 0.148 –0.359 0.189*

All use subject to https://about.jstor.org/terms


Journal of Southeast Asian Economies

Employed as operator 1.352 0.294*** 0.668 0.089*** 0.449 0.126*** 0.063 0.200
Employed as other service workers 0.000 0.441 0.259 0.110** –0.040 0.138 –0.173 0.205
Employed in an urban location 0.745 0.219*** –0.036 0.063 0.127 0.085 –0.068 0.095
No. of observations 10,057 10,354 12,688 15,128
Pseudo R2 0.0785 0.0259 0.0438 0.0533
LR chi2 088.70*** 208.14*** 163.23*** 159.84***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level. Agriculture is the reference for dummy variables on sector of employment.
Farmers and labourers are the reference for dummy variables on occupation.
Source: Labour Force Survey 2015; Author’s calculations using Stata.
Vo l . 3 5 , N o . 1

3/4/18 10:24 AM
18-J03487 JSEAE 07.indd 91
April 2018

TABLE 5
t-Test of Mean Differences between Treatment and Control Groups Before Matching

Group 1:
Group 2: Group 3: Group 4:
Incomplete Secondary or
Complete Secondary Incomplete Tertiary Complete Tertiary
Variables Lower
Mean Mean Mean Mean
Treat. Cont. Treat. Cont. Treat. Cont. Treat. Cont.
diff. diff. diff. diff.
Experience (no. of years) 28.12 26.06 16.53 17.16 * 15.58 15.23 15.95 14.23 *
Experience2 (no. of years) 976.14 891.44 394.98 431.66 ** 339.41 355.13 362.24 309.32
Male 0.79 0.81 0.77 0.69 *** 0.79 0.63 *** 0.69 0.46 ***
Married 0.72 0.61 ** 0.55 0.56 0.58 0.55 0.52 0.44 *
Educated father 0.11 0.08 0.72 0.65 *** 0.62 0.57 0.69 0.74
Employed in agriculture 0.11 0.38 *** 0.07 0.12 *** 0.04 0.07 0.02 0.01
Employed in industry 0.53 0.40 *** 0.39 0.40 0.29 0.29 0.27 0.15 ***
Employed in high-skill services 0.28 0.16 *** 0.36 0.34 * 0.40 0.41 0.35 0.32

This content downloaded from


Employed in low-skill services 0.06 0.03 0.15 0.12 *** 0.25 0.21 0.33 0.49 ***

152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC


Employed as professional 0.00 0.00 0.02 0.01 ** 0.01 0.03 0.26 0.40 ***

All use subject to https://about.jstor.org/terms


Employed as technician 0.01 0.00 ** 0.05 0.02 *** 0.07 0.05 0.06 0.08
Employed as clerk 0.00 0.00 0.10 0.07 *** 0.16 0.19 0.26 0.33
Employed as operator 0.60 0.25 *** 0.35 0.29 *** 0.34 0.20 *** 0.04 0.16 ***
Employed as other service workers 0.09 0.09 0.22 0.25 ** 0.24 0.30 * 0.15 0.09 *
Employed as labourer 0.27 0.63 *** 0.22 0.33 *** 0.20 0.14 ** 0.09 0.03 ***
Employed in an urban location 0.63 0.35 *** 0.61 0.59 ** 0.74 0.68 * 0.72 0.78
O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s

No. of observations 101 10,022 1,356 9,010 204 2,486 118 5,010
Note: Two sample t-test of equal variances: difference = mean (treatment) – mean (control) is equal to zero at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.
91

3/4/18 10:24 AM
92

18-J03487 JSEAE 07.indd 92


TABLE 6
t-Test of Mean Differences between Treatment and Control Groups After Matching

Group 1:
Group 2: Group 3: Group 4:
Incomplete Secondary or
Complete Secondary Incomplete Tertiary Complete Tertiary
Variables Lower
Mean Mean Mean Mean
Treat. Cont. Treat. Cont. Treat. Cont. Treat. Cont.
diff. diff. diff. diff.
Experience (no. of years) 28.12 27.45 16.53 15.77 ** 15.58 13.57 ** 15.95 12.63 ***
Experience2 (no. of years) 976.14 895.09 394.98 363.22 ** 339.41 285.48 * 362.24 234.92 ***
Male 0.79 0.91 *** 0.77 0.80 *** 0.79 0.78 0.69 0.54 ***
Married 0.73 0.79 0.55 0.56 0.58 0.56 0.52 0.42 *
Educated father 0.20 0.06 ** 0.48 0.41 *** 0.62 0.65 0.69 0.81 *
Employed in agriculture 0.11 0.23 *** 0.07 0.10 *** 0.04 0.02 0.02 0.00 *
Employed in industry 0.53 0.64 ** 0.39 0.47 *** 0.29 0.32 0.27 0.16 ***
Employed in high-skill services 0.27 0.11 *** 0.36 0.34 0.49 0.40 ** 0.35 0.23 ***
Employed in low-skill services 0.06 0.00 *** 0.15 0.07 *** 0.25 0.16 *** 0.33 0.59 ***

This content downloaded from


Employed as professional 0.00 0.00 *** 0.02 0.00 *** 0.01 0.01 0.26 0.37 **

152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC


Employed as technician 0.01 0.00 ** 0.05 0.01 *** 0.07 0.02 *** 0.06 0.04

All use subject to https://about.jstor.org/terms


Employed as clerk 0.00 0.00 *** 0.10 0.04 *** 0.16 0.13 0.26 0.40 ***
Journal of Southeast Asian Economies

Employed as operator 0.59 0.60 0.35 0.38 0.34 0.34 0.05 0.16 ***
Employed as other service workers 0.09 0.02 *** 0.22 0.26 *** 0.24 0.38 *** 0.15 0.07 **
Employed as labourer 0.27 0.36 * 0.22 0.28 *** 0.14 0.09 ** 0.09 0.03 **
Employed in an urban location 0.63 0.44 *** 0.61 0.61 0.74 0.86 *** 0.72 0.89 ***
No. of observations 101 936 1,353 5,115 204 459 118 447
Note: Two sample t-test of equal variances: difference = mean (public) – mean (private) is equal to zero at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.
Vo l . 3 5 , N o . 1

3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 93

TABLE 7
Propensity Score Matching: Differences in Log Hourly Wage

Group 1: Group 2: Group 3: Group 4:


Incomplete Complete Incomplete Complete
Secondary or Secondary Tertiary Tertiary
Lower
All workers
ATT 0.057 0.053*** –0.104*** –0.152*
  Standard error 0.065 0.014*** –0.031*** –0.102*
  t-value 0.871 3.802*** –3.399*** –1.490*
No. of observations 1,037 6,468*** –3.663*** –1.565*
 Treatment 101 1,353*** –3.204*** –1.118*
 Control 936 5,115*** –3.459*** –1.447*
Test of balancing property Satisfied Not satisfied Satisfied Satisfied
Children only
ATT –0.041 0.037*** –0.150*** –0.035*
  Standard error 0.111 0.020*** –0.109*** –0.096*
  t-value –0.365 1.858*** –1.378*** –0.368*
No. of observations 186 2,035*** –3.219*** –1.196*
 Treatment 29 1,582*** 1–3.90*** –1.152*
 Control 157 1,453*** –3.129*** –1.144*
Balancing property Satisfied Satisfied Satisfied Satisfied
Note: ATT was estimated using the nearest-neighbour matching method with common support option. Standard
errors are bootstrapped. t-values are significant at ***1% level; **5% level; and *10% level. Number of observations
in treatment and control groups refers to actual nearest neighbour matches. Covariates for sample using children only
include father’s education.
Source: Labour Force Survey 2015; Author’s calculations using Stata.

On the other hand, results among workers who obtained incomplete tertiary education show negative
and significant wage effects of vocational education. The results are consistent when the effects of father’s
education are controlled for. This implies that vocational education does not increase the earnings of
workers who have already obtained tertiary education. In fact, it almost proves to be a “penalty” for
them. Moreover, consistent with the OLS estimation is the result showing no significant wage effects
of vocational education among workers with complete tertiary education. However, for both of these
groups (incomplete tertiary and complete tertiary), the sample sizes are too small, requiring caution in the
interpretation of results.

6. Conclusion
The estimation results presented in this paper offer less support to the view that vocational education
in the Philippines has the potential to enhance earning prospects of workers, controlling for observable
individual-level and job characteristics. This implies that obtaining vocational education, in addition to
general formal education, does not provide the worker a labour market advantage. These estimates also
deviate from those of an earlier study (Choi 2016), where the results showed positive and significant

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 93 3/4/18 10:24 AM
94 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

wage effects of vocational education among low-educated workers. The differences in estimates may be
attributed to differences in statistical techniques used and/or the quality of data. Therefore, evidence on
the effectiveness of vocational education in increasing wages, at least in the Philippines’ context, remains
inconclusive. As such, caution must be exercised in making policy prescriptions that aim to promote
vocational education as an alternative to general formal education.
Given the policy attention placed on vocational education, there is clearly a need to improve the quality
of data. One possible way could be for the Labour Force Survey to include more questions regarding the
exact vocational courses taken by the workers. Alternatively, the impact evaluation study that is regularly
conducted by the Technical Education and Skills Development Authority could be expanded to facilitate
comparison of estimates among students who did not take vocational education.
Moreover, further research is necessary to identify factors for improving the effectiveness of vocational
education on increasing the workers’ wages. For instance, some vocational courses may be more relevant
than others, depending on what the labour market needs. Similarly, some courses may serve as better
complements to general education than others, depending on the level of cognitive and non-cognitive
abilities of the worker. So far, the empirical research on the topic is scant, most likely due to limited
availability of good quality data. The estimates presented in this paper could pave the way for more
experiential work to support policy-making in the Philippines.
While the estimates presented in this paper are preliminary in nature, they point to a potential policy gap
in the delivery of vocational education in the country. Findings of this article suggest that there is a need
to evaluate whether or not it is cost-effective to expand access to vocational education among early school
leavers. This is because obtaining general formal education remains as the most important factor that
leads to higher wages in the Philippines. These results are also corroborated by other studies on returns to
education in the Philippines (Hossain and Psacharopoulos 1994; Schady 2003). Perhaps, expanding access
to basic education (elementary and secondary levels), and allowing individuals to complete it, gives them
better chances of finding a well-paying job.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 94 3/4/18 10:24 AM
TABLE A1
Estimated Coefficients in Regression of Log of Hourly Wage

18-J03487 JSEAE 07.indd 95


April 2018

Equation (2a) Equation (2b) Equation (2c) Equation (2d)


Variables Estimated Std. Estimated Std. Estimated Std. Estimated Std.
coefficients errors coefficients errors coefficients errors coefficients errors
a –2.939*** 0.009 –2.825*** 0.010 –2.755*** 0.011 –2.786*** 0.011
Experience –0.020*** 0.000 –0.014*** 0.000 –0.012*** 0.000 –0.010*** 0.000
Experience2 (×100) –0.029*** 0.001 –0.020*** 0.001 –0.017*** 0.001 –0.015*** 0.001
Complete secondary (CS) –0.191*** 0.007 –0.202*** 0.007 –0.168*** 0.007 –0.147*** 0.007
Incomplete tertiary (IT) –0.335*** 0.011 –0.353*** 0.011 –0.315*** 0.011 –0.253*** 0.011
Complete tertiary (CT) –0.808*** 0.009 –0.863*** 0.009 –0.784*** 0.010 –0.485*** 0.011
V –0.128*** 0.050 –0.126*** 0.049 –0.092*** 0.048 –0.055*** 0.046
V*CS –0.014*** 0.052 –0.024*** 0.051 –0.002*** 0.050 –0.000*** 0.048
V*IT –0.175*** 0.061 –0.200*** 0.060 –0.170*** 0.059 –0.144*** 0.057
V*CT –0.339*** 0.068 –0.378*** 0.067 –0.326*** 0.065 –0.201*** 0.063
Male –0.173*** 0.006 –0.159*** 0.006 –0.173*** 0.006
Married –0.081*** 0.007 –0.078*** 0.006 –0.071*** 0.006
Industry –0.240*** 0.008 –0.149*** 0.009
High-skill services –0.084*** 0.009 –0.070*** 0.011

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
Low-skill services –0.294*** 0.011 –0.196*** 0.012

All use subject to https://about.jstor.org/terms


Professional –0.586*** 0.014
Technician –0.330*** 0.017
Clerk –0.304*** 0.013
Operator –0.168*** 0.008
Other service workers –0.061*** 0.011
O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s

Urban location –0.269*** 0.006 –0.273*** 0.006 –0.224*** 0.006 –0.202*** 0.006
R-squared 10.3169*** 10.3366*** 10.3645*** 10.4172***
F-test 1311.91*** 1195.69*** 1081.23*** 1011.82***
No. of observations 128,293*** 128,293*** 128,293*** 128,293***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.
95

3/4/18 10:24 AM
96 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

TABLE A2
Estimated Coefficients in Regression of Log of Hourly Wage
(Sample: All wage workers)

Equation (1) Equation (2)


Variables Estimated Standard Estimated Standard
coefficient errors coefficient errors
a –2.670*** 0.011 –2.669*** 0.011
V –0.047*** 0.011 –0.069*** 0.046
Experience –0.013*** 0.007 –0.013*** 0.000
Experience2 (× 100) –0.019*** 0.001 –0.019*** 0.001
Complete secondary –0.165*** 0.007 –0.161*** 0.007
Incomplete tertiary –0.272*** 0.011 –0.281*** 0.011
Complete tertiary –0.633*** 0.011 –0.640*** 0.011
V*CS –0.019*** 0.048
V*IT –0.126** 0.057
V*CT –0.182*** 0.058
Male –0.243*** 0.006 –0.243*** 0.006
Married –0.113*** 0.006 –0.113*** 0.006
Industry –0.085*** 0.010 –0.085*** 0.010
High-skill services –0.109*** 0.011 –0.109*** 0.011
Low-skill services –0.012*** 0.010 –0.012*** 0.010
Professional –0.781*** 0.012 –0.777*** 0.012
Technician –0.459*** 0.015 –0.456*** 0.015
Clerk –0.413*** 0.011 –0.410*** 0.011
Operator –0.241*** 0.008 –0.240*** 0.008
Other service workers –0.107*** 0.009 –0.106*** 0.009
Urban location –0.178*** 0.005 –0.178*** 0.005
R-squared 10.4731*** 10.4736***
F-test 1957.39*** 1667.43***
No. of observations 137,085*** 137,085***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 96 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 97

TABLE A3
Estimated Coefficients in Regression of Log of Hourly Wage
(Sample: All wage workers, children only)

Equation (1) Equation (2)


Variables Estimated Standard Estimated Standard
coefficient errors coefficient errors
a –2.717*** 0.017 –2.717*** 0.017
V –0.016*** 0.017 –0.003*** 0.085
Experience –0.017*** 0.001 –0.017*** 0.001
Experience2 (× 100) –0.031*** 0.004 –0.032*** 0.004
Complete secondary –0.170*** 0.012 –0.164*** 0.012
Incomplete tertiary –0.265*** 0.017 –0.279*** 0.017
Complete tertiary –0.564*** 0.017 –0.569*** 0.017
V*CS –0.067*** 0.087
V*IT –0.147*** 0.099
V*CT –0.117*** 0.102
Male –0.169*** 0.009 –0.170*** 0.009
Married –0.108*** 0.011 –0.108*** 0.011
Educated father –0.145*** 0.009 –0.144*** 0.009
Industry –0.071*** 0.016 –0.071*** 0.016
High-skill services –0.101*** 0.018 –0.102*** 0.018
Low-skill services –0.086*** 0.017 –0.087*** 0.017
Professional –0.696*** 0.021 –0.693*** 0.021
Technician –0.421*** 0.024 –0.421*** 0.024
Clerk –0.388*** 0.018 –0.379*** 0.018
Operator –0.229*** 0.015 –0.229*** 0.015
Other service workers –0.059*** 0.016 –0.059*** 0.016
Urban location –0.188*** 0.009 –0.188*** 0.009
R-squared 0.4753*** 0.4761***
F-test 693.59*** 596.39***
No. of observations 13,801*** 13,801***
Note: Estimated coefficients are significant at ***1% level; ** 5% level; and *10% level.
Source: Labour Force Survey 2015; Author’s calculations.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 97 3/4/18 10:24 AM
98 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

TABLE A4
Propensity Score Matching: Differences in Log Hourly Wage
(Sample: All wage workers)

Group 1: Group 2: Group 3: Group 4:


Incomplete Complete Incomplete Complete
Secondary or Secondary Tertiary Tertiary
Lower
All workers
ATT –0.038 0.071*** –0.108* –0.143***
Standard error –0.063 0.019*** –0.079* –0.033***
t-value –0.594 3.748*** –1.368* –4.326***
No. of observations –1,164 7,299*** –0.790* –1,328***
Treatment –0.124 1,641*** –0.263* –0.224***
Control –1,040 5,658*** –0.527* –1,104***
Test of balancing property Satisfied Not satisfied Satisfied Satisfied
Children only
ATT –0.088 0.036*** –0.031* –0.150***
Standard error –0.133 0.019*** –0.087* –0.098***
t-value –0.664 1.908*** –0.361* –1.533***
No. of observations –0.190 2,267*** –0.254* –0.292***
Treatment –0.134 1,664*** –0.105* –0.177***
Control –0.156 1,603*** –0.149* –0.215***
Test of balancing property Satisfied Satisfied Satisfied Not satisfied
Note: ATT was estimated using the nearest-neighbour matching method with common support option. Standard
errors are bootstrapped. t-values are significant at ***1% level; **5% level; and *10% level. Number of observations
in treatment and control groups refers to actual nearest neighbour matches. Covariates for sample using children only
include father’s education.
Source: Labour Force Survey 2015; Author’s calculations using Stata.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 98 3/4/18 10:24 AM
April 2018 O l f i n d o : R e t h i n k i n g Vo c a t i o n a l E d u c a t i o n i n t h e P h i l i p p i n e s 99

NOTES

1. Data taken from the Technical Education and Skills Development Authority.
2. For more discussion on the propensity score method as well as its practical implementation and limitations, see
Nopo (2008) and Caliendo and Kopeining (2005).

REFERENCES

Almeida, Rita, Leandro Anazawa, Filho Naercio and Ligia Maria Vasconcellos. “Investing in Technical and Vocational
Education and Training: Does it Yield Large Economic Returns in Brazil?”. Policy Research Working Paper
No. 7246. Washington, D.C.: World Bank, 2015.
Arrow, Kenneth. “Higher Education as a Filter”. Journal of Public Economics 2, no. 3 (1973): 193–216.
Becker, Gary. “Investment in Human Capital: A Theoretical Analysis”. Journal of Political Economy 70, no. 5 (1962):
9–49.
Bennel, Paul. “General versus Vocational Secondary Education in Developing Countries: A Review of the Rates of
Return Evidence”. Journal of Development Studies 33, no. 2 (1996): 230–47.
Bishop, John and Ferran Mane. “The Impacts of Career-Technical Education on High School Labor Market Success”.
Economics of Education Review 23, no. 4 (2004): 381–402.
Caliendo, Marco and Sabine Kopeinig. “Some practical guidance for the implementation of propensity score
matching”. Journal of Economic Surveys 22, no. 1 (2008): 31–72.
Choi, Seonkyung. “Effects of Vocational Education on Wage: Case of the Philippines”. Journal of Economics and
Sustainable Development 7, no. 10 (2016).
El-Hamidi, Fatma. “General or Vocational Schooling? Evidence on School Choice, Returns, and ‘Sheepskin’ Effects
from Egypt 1998”. Journal of Policy Reform 9, no. 2 (2006): 157–76.
di Gropello, Emanuela, Hong Tan and Prateek Tandon. Skills for the Labor Market in the Philippines. Washington,
D.C.: World Bank, 2010.
Hanushek, Eric, Guido Schwerdt, Ludger Woessmann and Lei Zhang. “General Education, Vocational Education, and
Labor Market Outcomes over the Life-Cycle”. Journal of Human Resources (2016).
Hawley, Joshua. “Comparing the Payoff to Vocational and Academic Credentials in Thailand Over Time”. International
Journal of Educational Development 23, no. 6 (2003): 607–25.
Horowitz, Andrew and Christoph Schenzler. “Returns to General, Technical and Vocational Education in Developing
Countries: Recent Evidence from Suriname”. Education Economics 7, no. 1 (1999): 5–20.
Hossain, Shaikh and George Psacharopoulos. “The Profitability of School Investments in an Economically Advanced
Developing Country”. International Journal of Educational Development 14, no. 1 (1994): 35–42.
Imbens, Guido W. “Nonparametric Estimation of Average Treatment Effects Under Exogeneity: A Review”. Review
of Economics and Statistics 86, no. 1 (2004): 4–29.
Kahyarara, Godius and Francis Teal. “The Returns to Vocational Training and Academic Education: Evidence from
Tanzania”. World Development 36, no. 11 (2008): 2223–42.
Krueger, Dirk and Krishna Kumar. “Skill-Specific rather than General Education: A Reason for US–Europe Growth
Differences?”. Journal of Economic Growth 9, no. 2 (2004): 167–207.
Lechner, Michael. “An Evaluation of Public-Sector-Sponsored Continuous Vocational Training Programs in East
Germany”. Journal of Human Resources 35, no. 2 (2000): 347–75.
Malamud, Ofer and Christian Pop-Eleches. “General Education versus Vocational Training: Evidence from an
Economy in Transition”. Review of Economics and Statistics 92, no. 1 (2010): 43–60.
Meer, Jonathan. “Evidence on the returns to secondary vocational education”. Economics of Education Review 26,
no. 5 (2007): 559–73.
Mincer, Jacob. Schooling, Experience and Earnings. New York: Columbia University Press, 1974.
Moenjak, Thammarak and Christopher Worswick. “Vocational Education in Thailand: A Study of Choice and
Returns”. Economics of Education Review 22, no. 1 (2003): 99–107.
Newhouse, David and Daniel Suryadarma. “The Value of Vocational Education: High School Type and Labor Market
Outcomes in Indonesia”. World Bank Economic Review 25, no. 2 (2011): 296–322.
Neuman, Shoshana and Adrian Ziderman. “Vocational Schooling, Occupational Matching, and Labor Market Earnings
in Israel”. Journal of Human Resources 26, no. 2 (1991): 256–81.
Ñopo, Hugo. “Matching as a tool to decompose wage gaps”. Review of Economics and Statistics 90, no. 2 (2008):
290–99.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 99 3/4/18 10:24 AM
100 Journal of Southeast Asian Economies Vo l . 3 5 , N o . 1

Olfindo, Rosechin. “Diploma as signal? Estimating sheepskin effects in the Philippines”. International Journal of
Educational Development 60 (2018): 113–19.
Orbeta, Aniceto and Michael Abrigo. “An Assessment of TESDA Scholarship Programs”. Philippine Institute for
Development Studies Discussion Paper Series No. 2012-25 (2012).
Orbeta, Aniceto and Emmanuel Esguerra. “The National System of Technical Vocational Education and Training
in the Philippines: Review and Reform Ideas”. Philippine Institute for Development Studies Discussion Paper
Series No. 2016-07 (2016).
Paqueo, Vicente and Aniceto Orbeta. “Unlocking the Filipino People’s Potential in the Next Six Years and Beyond”.
Albert del Rosario Institute for Strategic and International Studies, Quezon City, Philippines, 2017.
Peet, Evan, Günther Fink and Wafaie Fawzi. “Returns to Education in Developing Countries: Evidence from the
Living Standards and Measurement Study Surveys”. Economics of Education Review 49 (2015): 69–90.
Phelps, Edmund. “The Statistical Theory of Racism and Sexism”. American Economic Review 62, no. 4 (1972):
659–61.
Psacharopoulos, George. “Rates of Return on Investment in Education around the World”. Comparative Education
16, no. 1 (1972): 54–67.
———. “Returns to Education: A Further International Update and Implications”. Journal of Human Resources 20,
no. 4 (1985): 583–604.
———. “To Vocationalize or Not to Vocationalize? That is the Curriculum Question”. International Review of
Education 33, no. 2 (1987): 187–211.
———. “Returns to Education: A Global Update”. World Development 22, no. 9 (1994): 1325–43.
———. “Vocational Education and Training Today: Challenges and Responses”. Journal of Vocational Education and
Training 49, no. 3 (1997): 385–93.
——— and Harry Patrinos. “Secondary Vocational Education and Earnings in Latin America”. Vocational Aspect of
Education 45, no. 3 (1993): 229–38.
——— and Harry Patrinos. “Returns to Investment in Education: A Further Update”. Education Economics 12, no. 2
(2004): 111–34.
Riboud, Michelle, Yevgeniya Savchenko and Hong Tan. The Knowledge Economy and Educational Training in South
Asia. Washington, D.C.: World Bank, 2007.
Rosenbaum, Paul and Donald Rubin. “The Central Role of the Propensity Score in Observational Studies for Causal
Effects”. Biometrika 70, no. 1 (1983): 41–55.
Sakellariou, Chris. “Rates of Return to Investments in Formal and Technical/Vocational Education in Singapore”.
Education Economics 11, no. 1 (2003): 73–87.
Schady, Norbert. “Convexity and Sheepskin Effects in the Human Capital Earnings Function: Recent Evidence for
Filipino Men”. Oxford Bulletin of Economics and Statistics 65, no. 2 (2003): 171–96.
Schultz, Theodore. “Investment in Human Capital”. American Economic Review 51, no. 1 (1961): 1–17.
Spence, Michael. “Job Market Signaling”. Quarterly Journal of Economics 87, no. 3 (1973): 355–74.
Technical Education and Skills Development Authority. “Study on the Employability of TVET Graduates”. TESDA,
2014.
Tripney, Janice and Jorge Hombrados. “Technical and Vocational Education and Training (TVET) for Young People
in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis”. Empirical Research in
Vocational Education and Training 5, no. 1 (2013): 1–14.
United Nations Educational, Scientific and Cultural Organization (UNESCO). “Shanghai Consensus: Transforming
Technical and Vocational Education and Training”. Third International Congress on Technical and Vocational
Education and Training in Shanghai, People’s Republic of China, 14–16 May 2012. 2012.
———. “Increasing Public Awareness of TVET in the Philippines: A Case Study”. Bonn: United Nations Educational,
Scientific and Cultural Organization, 2010.
World Bank. The Right Skills for the Job? Rethinking Training Policies for Workers. Washington, D.C.: World Bank,
2012.
———. Philippines Labor Market Review: Employment and Poverty. Washington, D.C.: World Bank, 2016.

This content downloaded from


152.118.24.10 on Thu, 19 Sep 2019 09:03:52 UTC
All use subject to https://about.jstor.org/terms
18-J03487 JSEAE 07.indd 100 3/4/18 10:24 AM

You might also like