DLSU Research Congress 2022

De La Salle University, Manila, Philippines

July 6 to 8, 2022

Making TUPAD Happen: An Impact Evaluation Study on Tulong

Pangkabuhayan sa Ating Disadvantaged Workers Program

Jay Pee A. Maalihan1 and Dr. Mitzie Irene P. Conchada2

1Graduate Student, DLSU School of Economics
2 Associate Dean, DLSU School of Economics

*Corresponding Author: [email protected]

Abstract: In 2009, the Department of Labor and Employment (DOLE) introduced

TUPAD or Tulong Panghanapbuhay sa Ating Disadvantaged Workers, an emergency
employment program under DOLE’s main livelihood and welfare program. TUPAD
seeks to provide assistance to displaced workers, underemployed, and unemployed
poor through a cash-for-work scheme. This study assessed and evaluated the impact
of TUPAD to its beneficiaries by utilizing the DLSU-AKI Community-Based
Monitoring (CBMS) data from 2016-2017. This impact evaluation aimed to reinforce
and solidify the TUPAD program in order to bring forth welfare-based strategic
directions that will make decent living standards happen. Thereby, econometric
methods such as probit regression and propensity score matching (PSM) were
performed to determine objectives of interest. Probit regression showed that TUPAD
has significant optimistic effects on education and labor force by establishing positive
statistical relationships. For every increase in the non-continuance of education, there
is a 0.820 increase in TUPAD program participation. Similarly, labor force has the
same variable situation with education. Thus, for every unit increase in the non-
participation on labor, there is 0.032 unit increase in the program participation.
Correspondingly, PSM proved that TUPAD can increase its annual average income by
about 18%. The impact evaluation of TUPAD program proved the gradual
improvement of the living standards of a household through the additional interim
income stream from the provided or available work set by program administrators.
Still, it is important to acknowledge that education plays a vital role in somehow
avoiding the effects brought about by any adverse economic circumstances.

Key Words: impact evaluation; probit regression; propensity score matching; TUPAD

1. INTRODUCTION DOLE concluded that four out of every 10 workers in

the Philippines were deemed vulnerable as supported
by their statistical result regarding employment rate
The Philippines, as an emerging economy, amounting to 42.6 percent. Accordingly, during the
has always been striving for stability and downswing years of 2000-2009, the conjointly
sustainability especially in the human capital and declining GDP growth rate and the rising
employment sector. However, the Philippines has unemployment rate and population. Henceforth, due
always been a vulnerable country to both socio- to the repercussions of economic crises and the rapid
economic crises and natural disasters, affecting the growth in population, job generation slowed down
livelihood conditions of millions of Filipinos. In 2009, significantly (Urrutia et al., 2017). Moreover, almost

1
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

every year, natural disasters are the major outcome of interest. Moreover, this study then extends
contributors to regional economic decline leading to policy recommendations upon the evaluation results
the displacement of workers. In light of the country’s to strengthen and expand the program.
economic behavior towards the employment growth, Correspondingly, this study offers persistent
the Philippine government started zeroing in on the perspective on government’s social protection
provision of social protection, and inequality and programs to Economics students and researchers, and
poverty reduction through Department of Labor and this study can be valuable to DOLE as the results and
Employment (DOLE). The government’s initiative on policy recommendations may be a tool into reforming
alleviating the risks of the ramifications has resulted a more effective, equitable social protection policy
into commencing DOLE Integrated Livelihood and strategy.
Emergency Employment Program (DILEEP) in
2009—a program with a principal objective of 2. METHODOLOGY
livelihood restoration through immediate assistance
to unemployed, underemployed, disadvantaged, and
displaced workers.
2.1 Results Chain
Tulong Panghanapbuhay sa Ating Over the years of strengthening social
Disadvantaged Workers or TUPAD is a program protection, governments of developing countries had
initiated and implemented by DOLE. TUPAD is a veered away from commodity-based assistance and
community-based program that provides assistance replaced it with cash-based assistance through several
package for displaced workers, underemployed, and transfer modalities. Theoretically, cash-based
unemployed poor through a cash-for-work scheme.
assistance is preferable to commodity-based as it is
Initially, program offers emergency employment from
various projects for a period of 10-30 days. However, deemed to be more economically efficient (Tabor,
due to the reformation of special provisions under the 2002). Moreover, considering that certain
2019 General Appropriations Act, the program period assumptions hold, cash-based assistance gives the
was changed to 10-90 days. Projects include social, beneficiaries the freedom of choice to manage their
economic, and agro-forestry projects. In addition, household expenditures covering all necessities,
DOLE provides TUPAD enrollees of the following especially human capital investments. Thus, the
benefits: basic orientation regarding health and
beneficiaries acquire a higher level of satisfaction with
safety, personal protective equipment, micro-
insurance enrolment, and TUPAD ID. Beneficiaries cash-based assistance than the case with commodity-
would be paid 100% of the current minimum wage in based (Hanlon et al., 2010). In response, the
the private sector as per time records. Ultimately, Philippine government appointed DOLE for the
TESDA provides free accessibility to the agency’s implementation of DILEEP in order to alleviate
trainings in order to gear up the beneficiaries towards poverty caused by environmental, socioeconomic
self- and wage employment. (DOLE, 2014)
shocks through cash transfers and cash-for-work
TUPAD primarily aims to improve the
current situation of the disadvantaged workers schemes.
through cash-for-work which consequently translates
to an improvement in the essential expenditures of the TUPAD operates on cash-for-work scheme
beneficiaries. As such, this study assessed one of the that enables disadvantaged workers to continue
main objectives of TUPAD which is the additional investing in human capital development pursuits for
interim income stream. Correspondingly, it answered the stability of their household. Further, the
the question “What positive impact does TUPAD have emergency employment by TUPAD helps the
on the annual income of its beneficiaries?” Its beneficiaries in terms of labor force by filling in the
evaluation was obtained by utilizing secondary data
gaps caused by unavailability and loss of work.
from the DLSU-AKI Community-Based Monitoring
System (CBMS) data. However, due to the paucity of TUPAD hypothesizes that the emergency employment
the available data, evaluation is limited to a 1-year capacitates the beneficiaries to increase and improve
period from 2017-2018 and variables such as total their household expenditures for the meantime.
expenditure will be regarded as whole all throughout Correspondingly, Figure 1 shows the results chain in
due to the lack of component breakdown in the which TUPAD has the following theory of change:
dataset. The impact evaluation of this study applies
theory of change backed with hypothesis towards

2
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

beneficiaries and non-beneficiaries of the TUPAD

program.

2.4 Estimation Procedure and Model

Specification
The comparison of the outcomes in
Fig. 1. Results Chain
households with one TUPAD participant and
households with matched characteristics and
2.2 Evaluation Design
presumed outcomes but not a TUPAD participant is a
This study utilizes a matching design,
valid measurement of the impact of TUPAD. With
particularly propensity score matching (PSM), to
this, the assessment of the impact can be governed by
determine the impact of the TUPAD intervention to
the concept that any intervention that requires
the income of the participants. Generally, PSM is a
assessing shall be inferred through outcomes of both
suitable method to compare the income of the TUPAD
participants and non-participants. Accordingly, this
participants with their counterfactual group that did
concept can be translated into an empirical form as
not avail the program. Moreover, this method is
various studies such as Heckman et. al (1997) and
considered to be the most pragmatic strategy in
Smith & Todd (2001) asserted that in this conversion
identifying the impacts of the program of interest
𝑌1 is the mean of conditional outcome regarding
given the circumstances of data availability. The
participation, which is on the treatment group, while
evaluation design will be dependent of the secondary
is the mean of conditional outcome regarding non-
data available. Hence, no firsthand tasks or activities
participation which brought about by the control
were conducted in order to gather data for the
group. Thus, accounting for both aforementioned
intervention. The evaluation process design in
parameters shall present a change in the mean
determining the impacts of TUPAD are as follows: (1)
outcome that concludes the impact of participation in
identifying the source data, (2) estimation procedure
the program.
or empirical approach, (3) defining the variables, and
In this study, the change in the mean
(4) running the program that would cater the PSM.
outcome of both participants and non-participants
shall conclude the impacts of the TUPAD intervention
2.3 Data Source in income. The empirical form is given by:
In this study, the data utilized was obtained Δ𝑌 = 𝑌1 − 𝑌0 (1)
from DLSU-AKI which dated in 2016-2017. The data
Propensity score matching (PSM) as the main
is gathered through Community-Based Monitoring econometric method further examines the
System (CBMS), a diagnostic tool for assessment of abovementioned change in the mean outcome and this
the poverty in various levels such as barangay, method has the prediction of its propensity scores is
municipal, city, and provincial. The data includes dependent on either a probit or logit model. In this
demographics of the community, programs they study, it is based on a multivariate regression
enrolled into, or they are aware of, and all of the approach in terms of probit. The model is shown
necessary information that may be used for planning, below:
program implementation, and impact monitoring to 𝑒𝑒𝑝𝑡𝑢𝑝𝑎𝑑𝑖𝑛𝑑 = 𝛽0 + 𝛽1 𝑎𝑔𝑒𝑦𝑟 + 𝛽2 𝑒𝑚𝑝𝑙𝑦𝑑𝑦𝑟𝑠
whichever level it is needed. (2)
+𝛽3 𝑒𝑑𝑢𝑐𝑖𝑛𝑑 + 𝛽4 𝑙𝑎𝑏𝑓𝑜𝑟 + 𝛽5 𝑟ℎ𝑠𝑖𝑧𝑒 + 𝜀
The CBMS survey was conducted in 2016-
2017 among poor and non-poor in selected barangays Correspondingly, the predicted scores of the
in the following municipalities: City of Manila, City of propensity matching are utilized for the measurement
Marikina, Lipa Batangas, Dasmarinas Cavite, of the treatment effect. Comprehensively, the
Ozamiz Misamis Occidental, and Bago Negros treatment effect in this study is the average treatment
Occidental. The study focused on the poor both effect on the treated (ATT) as it is a widely known
parameter of interest when an analysis regarding

3
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

PSM is conducted (Becker & Ichino, 2002). The ATT Age Independent / Age of the worker
in this study shall be used to assess the effects of age_yr in years
participation on the TUPAD intervention. ATT is Years of Independent / Years of
calculated through matching participants and non- Employment emplyd_yrs employment in
numbers
participants with the closest propensity scores.
Education Independent / 1 if education
Moreover, radius matching technique, as embedded in educind was continued to
the statistical software package, shall be employed for college, 0
the advantage of comparing the available units within otherwise
the radius that allows the usage of extra units when Labor Force Independent / 1 if actively
there is no availability of good matches. Ultimately, labfor looking for a
job/currently
the verification of the robustness of the PSM is
employed, 0
realized using Stata’s sensatt program, a sensitivity otherwise
analysis specializing on PSM variables. Household Independent / Size of the
Size rhsize household in
2.5 Data Specification numbers
Data used in this study is gathered from
DLSU-AKI CBMS 2016-2017. The variables utilized 3. RESULTS AND DISCUSSION
are as follows: (1) participation in TUPAD, (2) age, (3)
years of employment, (4) labor force, and (5) household
size. In order to identify the impact of the TUPAD 3.1 Summary Statistics
program, enrollees of the program should be
As shown in Table 2, there are 5,791
determined as well as the non-enrollees. Hence, the
program participation variable. observations for the Availed TUPAD variable, and it
Age defines the range of required age for the has a mean of 0.0223 or 2.23% in terms of the
TUPAD program. Age is an important variable participant’s portion. It is defined by 1 and 0 which
because it determines the viability of a worker to represents the people who are program participants
legally participate in the workforce. Years of and who are not, respectively. For age, there are
Employment denotes the years rendered by an 36,030 observations with an average age of ~37 years
employee from its starting period until 2018. This
old. As such, the minimum and maximum age are 18
variable is necessary to determine the impact of
TUPAD because it provides the level of expertise of a and 65, respectively. Years in employment has 26,016
worker so as to land an appropriate job in the observations which presented a minimum of 0 to a
emergency employment program. Education defines maximum of 45 years of work experience. The mean
the worker’s educational level so as if they continued year of employment is 0.9486 or ~1 year. With 35,376
into college or not. This variable is as important as observations, education accounted for a mean value of
years of employment as well because it provides the 0.0706 or 7.06% in terms of the people who continued
educational attainment of a worker so as to land an
their education into college. Labor force and household
appropriate job in the emergency employment
program. Labor force categorizes the worker into two size has the same size of observations with age—
sections such as when (1) worker is actively looking for 36,030. Labor force participation has an average value
a job or currently working, or (2) no work at all. Lastly, of 0.4874 or 48.47% of the population are actively
household size denotes the number of family members seeking for a job or currently working. Lastly, for the
included in the worker’s current household. household size, the mean value is 5.3037 or ~6
members of the household from the minimum number
Table 1. Data Specification
of 1 to a maximum of 19 members.
Variable Type/ Designation Measurement
Availed Dependent / 1 if Yes-
TUPAD eetupadind participants of
TUPAD, 0
otherwise

4
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

Table 2. Summary Statistics the same variable situation with education. Thus, for
Variable Obs. Mean SD Min. Max. every unit increase in the non-participation on labor,
eeptupadind 5,791 0.0222 0.1475 0 1 there is 0.032 unit increase in the program
age_yr 36,030 37.2467 13.1081 18 65 participation. This result is straightforward as it
indicates that if the population is not working or not
emplyd_yrs 26,016 0.9486 4.1530 0 45
connected into any compensating work, they have a
educind 35,376 0.0706 0.2561 0 1
higher probability of availing TUPAD while seeking or
labfor 36,030 0.4874 0.4998 0 1 waiting for job opportunities. Lastly, the household
rhsize 36,030 5.3037 2.1029 1 19 size exhibits a statistically positive relationship with
the program participation. For every unit increase in
the household size, there is a 0.033 unit increase in
3.2 Probit Regression the program participation as well. This may describe
Prior to executing PSM is the probit that when the family or the members in the household
regression analysis. The variables of interest in this is getting bigger, the more likely that one of them
study undergone multivariate probit regression as would avail the TUPAD program to suffice their
depicted in Table 3. needs.
First, the results of age have exhibited a
significant positive relationship with the program Table 3. Multivariate Probit Regression
participation. The coefficient of regression shows that Variable (1)
for every unit increase in age, there is a 0.005 unit Multivariate Probit Regression
Age 0.005**
increase in the participation in TUPAD. These results
(0.003)
indicate that age affects the TUPAD in a very minimal Years of Employment -0.010***
manner but in certain sense, increase in age indicates (0.003)
higher responsibility so with the consequences of Education 0.820***
socioeconomic crises or natural disasters, age can play (0.163)
a role, but again, in a very minimal manner. Years of Labor Force 0.032**
employment’s results depict that for every unit (0.087)
Household Size 0.033***
increase in work experience, participation in TUPAD
(0.017)
decreases by 0.010. This means they have a Observations 5690
statistically negative relationship. As such, for every Pseudo R-square 0.032**
worker who gains work experience by months, or by Note: Standard errors in parantheses.
Statistical significance at 1%, 5%, and 10% are indicated by *** p<0.01, ** p<0.05, * p<0.1.
years, the less likely it is going to avail the program
since it may secure them a job that may not be 3.3 Propensity Score Matching
strongly affected by displacement or recession. As for Succeeding the multivariate probit
education, it shows a statistically positive relationship regression analysis, PSM was accomplished. In this
with program participation. For every increase in the method, total income of the households is the key
non-continuance of education, there is a 0.820 outcome indicator. Moreover, there is a total of 126 in
increase in the program participation. Moreover, it is the treated group and 5,564 in the controls group. The
important to note that in this variable, the non- two groups were determined by manual data analysis
in which the common characteristics: (a) poor and (b)
continuance was accounted rather than otherwise.
average annual income were factored in. Poor is
Hence, the result. This denotes that when participants classified within the range of minimum wage earners
refrain from pursuing college education, after the to unemployed while average annual income is based
occurrence of natural disasters or socioeconomic on the 2018 minimum wage which is Php475.00. Thus,
crises, they are more likely to avail TUPAD program accounting for the results of the prior regression
due to displacements or a possible recession. Labor analysis and incorporating the total income, ATT
force establishes a positive statistical relationship revealed that once a person participated in the
TUPAD program, their mean outcome in terms of total
with program participation. Similarly, labor force has

5
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

income is 25,942.9994. Correspondingly, with the T- conclusion, the PSM—as supported by sensitivity
stat of 3.16, a value higher than 2.0, this means that analysis—is robust.
there is an actual impact from TUPAD to the
participants’ income. Therefore, this denotes that
Table 5. Sensitivity Analysis (sensatt) Results
availing the TUPAD program can add Php
25,942.9994 to their annual average income of Php No. of No. of ATT Standard t-value
140254.126. Comprehensively, the mean outcome Treatment Controlled Error
from the ATT can be divided into the 30 to 60 day work 126 5,564 0.041 0.014 2.852
program set by the TUPAD administrators while
accounting for the current minimum wage. 4. CONCLUSIONS
Additionally, the PSM results are visualized in Figure
2. The red-dashed curve represents the controls group,
while the blue solid curve represents the treated In a country where there is an apparent
group. The graphical representation illustrates the vulnerability on socioeconomic crises and natural
impact of TUPAD on the total income of the disasters, it is imperative to implement emergency
participants. employment programs to keep the economy afloat.
These kind of programs backs the interim financial
Table 4. PSM Results needs of households while the job market is reviving.
Variable Sample Treated Controls Difference S.E. T-stat
Correspondingly, this study evaluates the impact of
Total Unmatched 140254.126 118478.305 21775.8211 5979.18247 3.64
Income ATT 140254.126 114311.127 25942.9994 8217.4599 3.16
TUPAD, an emergency employment program
Note: S.E. does not take into account that the propensity score is estimated. launched by the government with the goal of providing
temporary jobs to people who are displaced,
disadvantaged, or work poor. As to determining the
impact of TUPAD, total income was established as its
key outcome indicator. The impact of TUPAD on total
income was analyzed through propensity score
matching, a method that identifies the intervention’s
causal impact through matching characteristics of the
participants and the population who are observed to
be non-participants even when the program took place
in apparent. With this, it is necessary to know that one
of the caveats of the evaluation design is the inability
to fully capture detailed explanations of the behavior
and changes occurred during the program. However,
the results were comprehensive and conclusive as to
defining the impacts of TUPAD intervention.
TUPAD program significantly improves the
Fig. 2. K-density Graph of PSM living standards of a household through the additional
interim income stream from the provided or available
The sensitivity analysis—Stata’s sensatt— work set by program administrators. Nevertheless, it
conducted for PSM has resulted into a favorable is important to recognize that education remains the
outcome. Sensatt can prove the robustness of the PSM solid contributor in somehow avoiding the effects
through t-value. T-value must be higher than 2 to brought about by any adverse circumstances. Had the
support the claims of robustness of the study. As government encourage—not just provide free
depicted in Table 5, the t-value is 2.852 under 5,690 schooling—the people to finish their degrees through
observations. Only educind was examined against effective policy on education, displacements and
eeptupadind as sensatt can substantiate its results disadvantages on work can be minimized.
even when working only with one (1) independent Consequently, the size of household may be a factor as
variable and one (1) dependent variable. In

6
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

to why one cannot fully satisfy its financial

obligations. Economically, placing sex education and
family planning into taboo still deters the household
to set their mindset on examining their capabilities
and opportunities against their responsibilities before
resulting into a large family. Once the aforementioned
were established, the rendered work experience can
provide a reliable backing which subsequently
translates into reduction of job displacements because
there is enough year of service that can prove
expertise.
In the last quarter of 2021, the Philippine
government targets to increase the 2022 funding of
TUPAD program by Php 12 billion to aid the
beneficiaries. However, at the end of the day,
indicators and interventions are just signposts of
change.

7
DLSU Research Congress 2022
De La Salle University, Manila, Philippines
July 6 to 8, 2022

5. REFERENCES
Department of Labor and Employment (2014).
Department Order No. 137-14. Series of 2014.
Retrieved August 17, 2021,
https://www.dole.gov.ph/news/department-order-
no-137-14-guidelines-in-the-implementation-of-
the-department-of-labor-and-employment-
integrated-livelihood-and-emergency-
employment-programs-dileep/

Hanlon, J., Barrientos, A., and Hulme, D. (2010).

Just give money to the poor: The development
revolution from the Global South. Kumarian
Press.

Heckman, J. J., LaLonde, R. J., & Smith, J. A.

(1999). The economics and econometrics of active
labor market programs. In Handbook of labor
economics (Vol. 3, pp. 1865-2097). Elsevier.

Smith, J. A., & Todd, P. E. (2001). Reconciling

conflicting evidence on the performance of
propensity-score matching methods. American
Economic Review, 91(2), 112-118.

Tabor, S. (2002). Assisting the poor with cash: Design

and implementation of social transfer programs.
Social Protection Discussion Paper Series No.
0223. World Bank. Retrieved August 20, 2021,
fromhttp://tric.ru/files/2006/01/18/CDRom%20Co
ntent/pdfs/Papers/English/0223.pdf

Urrutia, J. D., Tampis, R. L., & Atienza, J. E. (2017).

An analysis on the unemployment rate in the
Philippines: A time series data approach.
Journal of Physics: Conference Series, 820,
012008. doi:10.1088/1742-6596/820/1/012008