IO-ko na:
Input-Output tables to determine the feasibility of Mindanao's independence
Former President Rodrigo Duterte proposed the idea of an independent Mindanao in response to President Ferdinand Marcos Jr.'s push for constitutional amendments. Highlighting Mindanao's dissatisfaction with the current state of the Philippines, he expressed a desire for independence due to a perceived lack of progress.
In response to Duterte’s statement, economist and professor Emmanel Leyco expressed the potential difficulties if ever Mindanao plans to secede. He asserted the capability of the island region with its substantial agricultural resources, but mentioned its present underdeveloped state that would hamper its aspiration to progress as a separate entity [1].
In support, former House Speaker Alvarez drew parallels between Mindanao's potential independence and Singapore's separation from Malaysia, emphasizing Mindanao's abundant natural resources and potential for development [2].
In Southeast Asia, Singapore was the first to lead the action of secession from its parent country. There are various factors that motivated its separation from Malaysia, but mainly, racial and socio-political sentiments triggered this movement [3]. The separation was a success, and Singapore has since become a developed country with a strong economy. From a general standpoint, Singapore has significantly less land mass than Mindanao, but a lower unemployment rate and higher GDP. However, due to its small agricultural production, it highly relies on imports to sustain its demand for food [4]. Mindanao, on the other hand, is a region whose economy is dominated by the agricultural sector. One-third of its area is dedicated to agriculture, and the island produces 40% of the country’s total food requirements [5].
As such, the team is interested in the economic structure of a separate Mindanao in terms of its production and use of resources in the form of Input-Output Tables [6].
Analyze industry interdependence of Mindanao through existing data science methodologies and compare it against the Philippines as a whole.
Use the national input-output table and the cross-industry local quotient (CLQ) technique, which employs regional GDP, to estimate the regional input-output table of Mindanao.
There is no significant difference in the correlation between agriculture industry inputs and outputs between Mindanao and the Philippines as a whole.
There is a significant difference in the correlation between agriculture industry inputs and outputs between Mindanao aand the Philippines as a whole.
As we are looking into the potential economic model of Mindanao, it is essential to know how the economic model it is a part of looks like. As such, the Philippines and its Southeast Asian neighbors are important models to look into to potentially map how the economic model of Mindanao would look like.
Through our knowledge of data science, we aim to utilize SEA economic data to be able to analyze the interdependence of different industries in the Philippines and whether it correlates to the country’s GDP growth rate.
Analyze and model the national input-output tables of each country to examine industry tightness, specifically which industry the economy depends the most on.
There is no significant correlation between agriculture industry inputs and all industry outputs in the Philippines.
There is a significant correlation between agriculture industry inputs and all industry outputs in the Philippines.
Data collection was done through searching online repositories for publicly available data.
As the data points show, there seems to be a correlation between the input from the agricultural industry to the different industries in a country and the output of those industries. Of note is the Food, beverages, and tobacco industry with its clear linear correlation and relatively large industry size in each country. Another large industry in Indonesia’s case is Construction with also a clear correlation. These first impressions on the data points lead to curiosity on which industries have a strong correlation from the agriculture input.
For research question 1, we first aggregated the industries in the national input-output table from ADB so that it matches the industries in the Mindanao dataset provided by PSA. Next, we used the technique illustrated in this study, described by the diagram below:
We picked CLQ over the other locational quotients since the data available is only enough to create the CLQ as the interregional data needed for the FLQ was not found.
Once the regional input-output table was obtained, Pearson correlation was done between the Input from the Agriculture Industry to the different Industries and the Output of said Industries
Likewise for research question 2, Pearson correlation was done between the Input from the Agriculture Industry to the different Industries and the Output of said Industries for each country.
P-values for each correlation pair (agricultural industry input to industry i to industry i output), from the Pearson correlation test, were combined using Fisher’s method.
The resulting p-value for the Philippines dataset is 3.2171407277283736e-40, which is lower than the significance level (0.01). Hence, we reject the null hypothesis. Therefore, there is a significant correlation between the input that each industry receives from the agriculture industry and their corresponding outputs. Take a specific look at our methods here.As there is a relationship between the input from the agricultural sector and the output of an industry, it is essential to note the potential ripple effects it could have when a policy to increase agricultural output is issued. Consequently, any environmental or economic event that risks the production of the agricultural industry may also affect industries that rely on or correlate with agricultural output.
Paired with this, it is also important to note that if the agricultural output is increased, if it is not inputted to the other industries, it does not translate into a ripple effect to the other industries.
Even with high correlation values overall, it is evident that there are industries in a specific country that show weak correlation to agriculture input but show substantial correlation in other countries. For instance, the outputs of coke, refined petroleum, and nuclear fuel along with the rubber and plastics industries exhibit a very weak correlation to the agriculture input in the Philippines, yet have a high correlation in other southeast asian countries. Aside from this, there are industries that show positive correlation in the majority of the countries but are negatively correlated in one. That is, apart from the mining, food, and paper product industries, other industries' relationship with agriculture, between SEA countries, has no definite shape - which might imply the varied behavior of their economies.
Nonetheless, because of the strong correlation found between input from the agricultural industry and industry output in Mindanao (in fact even higher than that of the whole country), there is merit in considering the point raised in the news report. However, further research is needed to determine how much the market of the potentially separate Mindanao would depend on the agriculture industry, as well as if this is sustainable.
In an attempt to generalize the trend of the growth in each country, we created and tested models that can predict the future Input-Output Tables of the different countries in this study.
Given that there are categorical variables that are essential to the prediction (Country and Source Industry), we incorporated dummy variables to represent these variables in the linear regression model. Our references for the variables are the Philippines and the Agriculture industry respectively. After organizing the information, fitting the model with the different combinations of features as documented in our methods here, and testing the models, we cross-checked to see the error of the different models in predicting the Input-Output tables of the different countries. This leads us to discover that the error (using mean squared error) is relatively large (around 700 to 850 in the Philippines, Malaysia, and Vietnam, around 1,500 in Thailand, and around 2,300 in Indonesia) given the range of the values in the model (0 to 10,000).
On average, the 2 most accurate models are the ones that have the {Industries, Country, Output by Industry} and {Industries, Country, Output by Industry, Year} as the features, with the latter being more accurate overall. This could be because of the link between the Industry Output and their respective Input-Output tables as the inputs to the other industries and exports are capped at the total output of the source industry. Although this is the case, the relatively large error is still of note and maybe a point of further research. Another thing of note is the larger error when predicting the IO table for Indonesia. This could signify that the economic structure of Indonesia is a potential outlier in the SEA context as can also be inferred in the correlation heatmaps. Thailand is also of note, but not so much as Indonesia.
Find data about the interregional trade within the Philippines to arrive at a more accurate IO table for Mindanao based on the source.
Have the PSA and ADB datasets be coordinated/coherent to contain the same industry classifications.
Do the same correlation methodology for all sectors that can be done (other 6 sectors for correlation with Mindanao, other 34 sectors for 5 countries), so that a more comprehensive and comparative view of the economic structure within a country can be obtained.
Perform methods that best describe the flow between sectors such as network analysis and measures of centrality like closeness or betweenness to effectively compare industry behaviors of each country.
Build on the research to provide stronger evidence of Mindanao’s independence in terms of economy by using other economic indicators.
UPD DCS student, tabletop games and cat enthusiast. Willing to share my collection of cat pics. Contact me at edaldemita@up.edu.ph
UPD DCS student, frozen yogurt and donut enthusiast. Willing to share the best spots for your sweet needs!
UPD DCS student, cat enthusiast (2) and bookworm. Happy to recommend my favorite novels~ Contact me at nbvaldez@up.edu.ph