This has been a common thinking among people from provincial areas; but, is this actually true? Our project, Barrioline, aims to assess the economic disparities between rural and urban areas and determine whether one area has a better capacity to alleviate a person from poverty.
Poverty, especially in rural regions, remains a serious problem in the Philippines. With urbanized areas often portrayed as hubs of economic growth and employment opportunities, people from the province have the impression that relocating to an urbanized area will improve their economic status and quality of living. However, with increased migration in the urbanized area, another series of problems arise: overcrowding, illegal settling, and increased competition for jobs, among others; consequently resulting in increased poverty rates in the urban area as well.
Our solution is to utilize data science to analyze various indicators of poverty and economic disparity across the different regions in the Philippines. Comparisons between different regions would be made in order to determine which region, if any, best improves the chances of alleviating poverty.
To look at the different poverty indicators and employment specific variables in order to determine and compare the poverty alleviation capacities and economic opportunities between urban and rural areas in the Philippines.
More than half of the Philippines’ 100 million people live in rural areas, and more than a third of them are poor. [1] The government has set an ambitious target to reduce the national poverty rate to a single digit figure by 2028 as part of the administration’s broader strategy to boost economic growth and tackle socio-economic inequalities. [2]
Recent data from the Philippine Statistics Authority reveals a glimmer of hope. The poverty rate dropped from 23.7% in 2021 to 22.4% in the first half of 2023. [2] Despite this progress, it means that roughly 25.24 million Filipinos continue to live under the poverty line. To add, poverty rates have been far from uniform across the country. Urban regions have a poverty incidence of 13% whereas rural areas have a much higher rate of 36%. This highlights that poverty is concentrated in rural regions. [1, 2] Due to this disparity, a lot of individuals from rural areas are enticed to relocate to urban areas, in search of employment and belief that this is their ticket out of poverty. [3, 4]
How does poverty alleviation differ between urban and rural areas in the Philippines?
Null Hypothesis: Both urban and rural areas are economically stable and equally likely to alleviate poverty.
Alternative Hypothesis: One area is more likely to successfully alleviate poverty compared to the other.
How do employment
opportunities vary across the regions in the country?
Null Hypothesis: There is no significant difference in the employment opportunities across the regions in the Philippines.
Alternative Hypothesis: Employment opportunities are significantly better in urbanized areas.
To address the research questions, our group collected and preprocessed data from the 2022 Annual Poverty Indicators Survey (APIS) and the Labor Force Survey (LFS), which contain the latest data available regarding labor and poverty indicators. In order to manage these large datasets, we sampled 1,000 households out of approximately 43,000 surveyed from both APIS and LFS using tools such as Pandas and Numpy. We also included data regarding labor and employment from the 22nd edition of the Philippine Statistical Yearbook (PSY). All of these datasets are provided by the Philippine Statistics Authority (PSA).
For the first research question, the data from APIS will be used. This dataset contains nationwide indicators of poverty and records of government assistance per household. First, we will determine the economic status of each household by looking at the following variables: region, family size, monthly income and monthly food expenses, and urban/rural status. Once a family’s economic status is determined, the government assistance received per household will be examined to see if it is enough to help alleviate financial issues.
In addressing the second research question, the LFS dataset will be used. This survey is a nationwide survey of households that analyzes employment status. We will be focusing on the following variables: region, urban/rural status, sex, age, and marital status, as well as employment-specific variables such as job status, reasons for not seeking employment, and whether individuals are Overseas Filipino Workers (OFW). These variables will be used to determine the effect that region has on the type of employment Filipinos have.
Aside from those mentioned, basic daily wage pay is another variable that we will be using to answer the second research question. We will be using a table from the PSA OpenSTAT database entitled “Average Daily Basic Pay of Wage and Salary Workers by Major Industry Group, Philippines: 2016 to January 2024 (in peso).” and “Table 9: Employed Persons by Region and Major Industry Group, Philippines: 2018 to 2021” from the Philippine Statistical Yearbook. We will refer to the data from 2018-2021 in order to visualize how the average daily wages and employment per industry now differs from the years before.
Lastly, to show the overall impact and relevance of the study, we used the following data from the Philippine Statistics Authority: “Table 6a: Magnitude of Poor Population with Measures of Precision, by Region, Province, and Highly Urbanized Cities 2018, 2021, and 2023” from the Philippine Statistical Yearbook and “Table A. Total Population, Urban Population, and Percent Urban by Region, Province, and Highly Urbanized City” from the 2020 Census of Population. These were used to visualize the magnitude of the poor population in comparison to the total population per region. The regions were classified as either urban or rural based on their level of urbanization (>50%) stated.
The plot consists of a bar graph depicting the total population per region and a line graph showing the magnitude of poverty per region. The data on the level of urbanization of each region was utilized to classify areas as either urban or rural. As seen in the bar graph, urban regions have a higher overall population. One of the reasons could be due to the influx of individuals moving to cities in search of better employment opportunities. However, urban regions also have a larger number of residents living in poverty. This suggests that for all opportunities urban regions provide, it also comes with major challenges, such as higher cost of living, trapping Filipinos in poverty. On the other hand, rural areas, while having smaller populations, tend to experience less poverty. Overall, this graph shows the need for effective poverty alleviation programs that specifically address the diverse and complex challenges faced in urban and rural settings.
Daily life in the Philippines will be difficult if there is insufficient access to utilities such as electricity and water. Through this graph, we check if there is a difference in the availability of essential amenities between urban and rural areas and find out which area may need more assistance. Using data from the 2022 Annual Poverty Indicators Survey, we discover what percent of households have sufficient access to water, electricity, internet, and public transportation. For water and electricity, we observe that rural and urban areas have high access to such utilities and have little difference between them. For broadband internet, we notice that it is more prevalent in rural households, meaning that there could be a problem with distribution in urban areas and assistance may be needed. For public transportation, the graph suggests that rural areas must give more importance in giving access to public transport in order to close the gap between urban and rural areas in this sector.
From the "2021 Family Income and Expenditure Survey" of the Philippines Statistics Authority, we isolated three columns: URB, TOINC, and TOTEX. URB is whether the household surveyed is in an urban (1) or rural (2) area. TOINC is the total income (in pesos) of the household in 2021, and TOTEX is the total expenditure. Using data from "Highlights of the 2021 Full Year Official Poverty Statistics" of the PSA, we see that the poverty threshold is 12,030 pesos per month, or 144,360 pesos per year. Using this number, we filter out the households to those under the poverty line. Then, we checked whether the household has a higher income than expenditure (meaning that savings are increased) and store the result as a boolean in the table. The counts of this variable are what is tested for independence.
After preprocessing, we get this contingency table:
When we perform the chi-square test of independence on this data, we get a p-value of 2.3759450908147093 x 10-17. This is less than the 0.05 significance level, meaning that we reject the null hypothesis and accept the alternative hypothesis.
This means that urban and rural areas are not on the same level in terms of capability of alleviating poverty. From the contingency table, 77.49% of the urban poor population is saving money compared to 81.07% of the rural poor population. This result may have stemmed from the difference in access to basic necessities. As seen in the data visualization for RQ 1, there is a gap between urban and rural areas in terms of accessibility to some utilities. Note that further testing will be done in order to confirm the cause.
With data taken from the Philippine Statistical Yearbook (2022), we visualized the distribution of employees for each major industry group per region. The regions are further clumped into their respective island groups: Luzon, Visayas, and Mindanao. The top three most populated industries are agriculture, construction, and trade & repair; with Luzon having the most number of employees of the three regions in total. This information is crucial in recognizing the employment opportunities that are available for respective industries in different regions.
We first grouped the regions by their respective island groups Luzon, Visayas, and Mindanao to make the visualization less dense. We also limited our observations to the most recent data available in 2021. The decision was also made to exclude industries that have only one or no employees listed in the table.
For this dataset, we deemed it best to use ANOVA for the analysis. We checked the ANOVA for each individual region, and then checked the summary of the model. The results show an overall model p value of 7.76e-36 (7.76 x 10-36), which is indicative of an extremely low probability of the null hypothesis being true. The value is much lower than the significance level of 0.05, meaning that we reject the null hypothesis and accept the alternative hypothesis.
This follows the typical idea of people moving from the provinces to urban areas to seek better job opportunities,
either because of the abundance of slots or the diversity of the industries. These opportunities are also likely to
provide better pay. From the dataset, the industries with the highest average daily wages are: Information and
Communication, Professional, Scientific and Technical Activities, and Education, with wages ranging from Php 800-900.
Looking at the bar plots, it's easy to tell that these aforementioned industries are largely dominated by employees
within Luzon, concentrated around the urban regions of the island. It's possible that these higher-paying industries
may not have as many branches or institutions in rural regions, and therefore would not have as many slots open for
employees.
It is clear that urbanized areas, particularly the regions located in Luzon, have more employment opportunities available to them. Major industries generally favor these areas because of their higher population, and thus bigger workforce. However, we cannot decisively conclude that more opportunities means better chances of poverty alleviation from this alone. There are far too many factors to consider in this discussion. one of them being the cost of living. The cost can differ wildly depending on the region they live in and the necessities their jobs demand. Further testing is a must before any definitive conclusions can be made.
The dataset to be used is the March 2023 Labor Force Survey from the Philippines Statistics Authority.
Using the demographic variables (age, sex, marital status, household size, urban/rural, highest educational attainment), we
will use k-means clustering to group similar people and see how the different groups compare in terms
of the job related variables. This will hopefully give a sense of how the demographic variables
are related to the job related ones.
For this data we are only considering working age Filipinos (15 years old and over) who are not OFWs,
so we filter out those who do not fit this criteria.
The highest educational attainment is split into 11 categories, the major occupation group is split into 10 categories,
and the major industry group is split into 22 categories according to the dictionary included with the data. These
categories will be listed as the results are shown.
For the k-means clustering, the features will be scaled using scikit-learn's StandardScaler so that they have an equal effect on the clusters.
After initial preprocessing, we do tests to determine what number of clusters would be used. The first test is checking the inertia of each possible number of clusters. This is called the inertia, and a lower value is better for clustering. We will also use the silhouette scores of each k to determine the best number of clusters. This is a measure of how similar members of a cluster are to each other, and a higher score indicates less overlap between clusters.
As can be seen in the graphs, 6 seems to be the optimal number of clusters. It is located at a slight elbow in the inertia graph, and it is at a peak in silhouette score. Therefore, I will use six clusters for the k-means clustering.
For the initial dataset, the number of people who were of working age and not an OFW was 32955. Most of those people lived in NCR (14.30%), and had an average age of 39.50, with a standard deviation of 17.91 years. The percentage of people living in urban areas was 49.34%, and the percentage of males was 50.55%. For employment, 62.36% of the population had work.
When the dataset is split into 6 clusters, here are how the demographic features look like:
Most common region per cluster:
The purpose of the clustering was to check if the clustering people by their demographics could give us insights on work-related variables, and here is how they look like for our current clusters.
Most common major occupational group per cluster (percentage of working population):
Many things can affect what job a person gets. For example, a person may want to get a job early if they
have a very large household and support their family. Elderly people may retire and not have work anymore.
The purpose of using machine learning on this dataset is to help see connections between these factors on
the availability and type of work in the Philippines.
The results of the clustering may present a clearer picture of what demographic factors influence work in
the Philippines. We can separate the clusters by their demographics, and I noted the most obvious characteristics
of the clusters.
In our study, we discovered that urban and rural regions differ when it comes to their access to utilities, financial practices, and employment trends. While both types of region have access to basic utilities like water and electricity, rural areas unexpectedly have better broadband internet access, suggesting distribution issues in urban areas. But there is still a gap between these regions. These utilities are important for poverty alleviation because they are needed in activities and opportunities that support economic and social development.
Financial habits also differ, with rural areas displaying a slightly higher tendency to save money. Effective money management practices help in ensuring long-term financial stability and growth, which may lead to better poverty alleviation efforts. Upon performing hypothesis testing using the difference in saving and spending habits, we see that one area is more likely to successfully alleviate poverty compared to the other. This means that urban and rural areas are not on the same level in terms of capability of alleviating poverty. This result may have stemmed from the difference in access to basic necessities which additional research can answer.
Employment patterns, as seen through machine learning, show that rural areas predominantly engage in agriculture, limiting economic diversity compared to urban regions which boast a wider range of job sectors. This lack of diversity could restrict economic growth, thereby undermining poverty alleviation efforts.
Additionally, our research indicates a significant difference in employment opportunities between these regions, with urban areas generally offering more job openings. We can say that employment opportunities are significantly better in urbanized areas, in terms of available job openings.This explains why individuals from rural provinces migrate to cities in search of better jobs. However, it remains unclear whether this increase in opportunities directly correlates with improved poverty alleviation.
As mentioned earlier, we suggest doing additional research in order to find the possible reason as to why one area is more likely to successfully alleviate poverty. Additional domain knowledge may be helpful in determining which factors to consider in order to determine the reason why poverty alleviation is different in urban and rural areas.
To deepen our understanding regarding the job opportunities among the regions, future research should also examine additional variables such as the average wages across different industries and the cost of living in each region. These factors will provide a more comprehensive view of the economic landscape and help tailor more effective poverty reduction strategies.
Lastly, in applying the findings of this research to poverty alleviation, both urban and rural areas need tailored approaches. Policies targeting urban areas should concentrate on improving internet connectivity distribution to enhance job opportunities and economic participation. While initiatives in rural areas should focus on enhancing the reliability of utilities and expanding the variety of job sectors beyond predominantly agricultural work. Moving forward, future research should investigate the causal factors behind these regional disparities and implement targeted interventions. This approach will allow for the refinement of strategies aimed at effectively reducing poverty across both urban and rural regions in the Philippines.
abcajandab@up.edu.ph
IV - BS Computer Science Student
acgo8@up.edu.ph
II - BS Computer Science Student
mrlaviste@up.edu.ph
II - BS Computer Science Student
plmontecillo@up.edu.ph
IV - BS Computer Science Student