Financial Inclusion in the Philippines

Analyzing financial access, digital payment adoption and financial behavior among Filipino adults using Global Findex 2025 survey data.

---

Key Results at a Glance

Finding	Result
Largest account ownership gap	48.53 pp — Education
Digital payment adoption among account owners	80.37%
Digital activation gap	19.63%
Largest digital adoption gap	44.29 pp — Age
Saving gap by financial engagement	40.35 pp

---

⚙️ Project Type Flags

• Exploratory Data Analysis (EDA)
• Python Analysis
• Data Cleaning / Wrangling
• Statistical Analysis
• Data Visualization
• End to End Analysis

---

Table of Contents

1. Project Overview
2. Objectives
3. Project Scope & Tools
4. Repository Structure
5. Data Workflow
6. Data Model & Schema
7. Analysis & Metrics
8. Key Insights
9. Recommendations
10. Assumptions & Limitations
11. Future Enhancements
12. Deliverables
13. Author

---

1. Project Overview

Context: Financial inclusion extends beyond whether individuals own a financial account. People may have formal financial access but remain digitally inactive, while financially excluded individuals may still participate in financial behaviors such as saving or borrowing.

This project analyzes the Philippines subset of the Global Findex 2025 dataset to examine financial inclusion across three connected dimensions:

Financial Access → Digital Activation → Financial Behavior

Problem Statement: Which groups in the Philippines remain financially excluded and what characteristics are associated with the transition from financial access to active digital financial usage?

Approach: The project uses survey-weighted descriptive analysis, demographic segmentation, behavioral segmentation, interactive visualization, Chi-square tests, and Cramér’s V association measures.

Three research questions guide the analysis:

1. Which demographic groups show the largest gaps in financial account ownership?
2. Among account owners, which groups show the largest differences in digital payment adoption?
3. How do saving and borrowing behaviors differ across financial engagement levels?

Outcome: The analysis found that:

• Education and income show the largest disparities in financial account ownership.

• 80.37% of account owners use digital payments, leaving a 19.63% digital activation gap.

• Age and internet usage are the strongest observed differentiators of digital payment adoption among account owners.

• Financial engagement differentiates saving behavior more strongly than borrowing behavior.

2. Objectives

• Primary Objective: Identify demographic and behavioral characteristics associated with financial exclusion and digital financial adoption among Filipino adults.
• Secondary Objective 1: Quantify financial account ownership using survey-weighted estimates.
• Secondary Objective 2: Identify demographic groups with the largest account ownership gaps.
• Secondary Objective 3: Determine whether account ownership translates into digital payment activity.
• Secondary Objective 4: Identify characteristics associated with digital adoption among account owners.
• Secondary Objective 5: Compare saving and borrowing behavior across financial engagement levels.
• Secondary Objective 6: Statistically validate major descriptive findings.
• Secondary Objective 7: Translate analytical evidence into practical financial inclusion recommendations.

💡 Every analysis decision in this project traces back to one of these objectives.

---

3. Project Scope & Tools

Scope

Dimension	Details
In Scope	Philippine respondents from Global Findex 2025; financial access, digital payment usage, saving, borrowing, internet usage, and demographic factors
Out of Scope	Causal inference, transaction level analysis, credit quality, monetary transaction values, longitudinal trends
Survey Year	2025
Source Dataset	Global Findex 2025 microdata
Granularity	One row per survey respondent
Primary Analysis Level	Respondent level weighted analysis

Tools & Technologies

Category	Tool(s) Used
Data Source	Global Findex 2025 Microdata
Data Access	Google BigQuery
Data Processing	Python
Analysis	Pandas, NumPy
Statistical Testing	SciPy
Visualization	Plotly
Development Environment	Google Colab
Version Control	Git / GitHub
Documentation	Markdown

---

4. Repository Structure

financial-inclusion-philippines/
│
├── README.md
├── notebooks/
│   └── Financial_Inclusion_Philippines_Portfolio.ipynb
├── visuals/
│   ├── financial_inclusion_snapshot.png
│   ├── account_ownership_gap.png
│   ├── digital_adoption_gap.png
│   └── financial_behavior.png
├── data/
│   └── README.md
└── docs/
    └── data_dictionary.md

---

5. Data Workflow

Global Findex 2025
      ↓
Filter Philippines
      ↓
Validate Respondents
      ↓
Select Analysis Variables
      ↓
Check Missing Values & Categories
      ↓
Apply Survey Weights
      ↓
Feature Engineering
      ↓
Weighted Descriptive Analysis
      ↓
Segmentation & Gap Analysis
      ↓
Statistical Validation
      ↓
Visualization
      ↓
Insights & Recommendations

1. Source: [Global Findex 2025 microdata containing approximately 144,090 respondents and 199 variables.]
2. Filtering: The dataset was filtered to respondents whose economy was recorded as the Philippines.

The final Philippine sample contains:

• 1,000 respondents
• 1,000 unique respondent IDs
• 0 duplicate respondent IDs Only variables relevant to the research questions were retained for analysis.

3. Cleaning: Data quality and category checks were performed before analysis.

The cleaning process included:

• checking respondent IDs for duplicates
• reviewing missing values in key analytical variables
• checking for missing survey weights
• validating raw category values against the data definitions
• reviewing demographic and financial indicator coding
• checking survey weight values before weighted calculations
• confirming that key binary indicators were consistently coded

These checks helped ensure that the analytical sample and variables were suitable for the subsequent weighted analysis.

4. Transformation: Raw survey variables were transformed into readable categories and analytical segments to support interpretation.

Derived fields included:

• Gender - readable gender categories
• Education - Primary or Less, Secondary, and Tertiary
• Income Quintile - Q1 Poorest through Q5 Richest
• Employment - In Workforce and Out of Workforce
• Residence - Urban and Rural
• Age Group - grouped respondent age bands
• Account Status - Account Owner or No Account
• Digital Payment Status - Digital Payment User or Non-Digital Payment User
• Digital Account Status - Digital Account or No Digital Account
• Internet Status - Internet User or Non Internet User
• Financial Behavior - Saver & Borrower, Saver Only, Borrower Only, or Neither
• Financial Engagement - Financially Excluded, Basic Access, or Digitally Active

Reusable functions were also created to calculate survey-weighted percentages across demographic and behavioral groups.

5. Analysis: The project combines descriptive, visual, and statistical analysis.

Methods used include:

• Survey weighted analysis to estimate population level percentages
• Demographic segmentation to compare financial outcomes across population groups
• Gap analysis to measure percentage point differences between the highest and lowest groups
• Behavioral segmentation to examine saving and borrowing patterns
• Cross tabulation to validate relationships between financial access and digital payment indicators
• Chi-square tests of independence to test categorical associations
• Cramér’s V to compare association strength
• Plotly visualizations to communicate major patterns and disparities

6. Output: The analysis produces:

• weighted KPI summaries
• demographic comparison tables
• percentage point gap analyses
• financial behavior segments
• interactive Plotly visualizations
• statistical validation tables
• key findings and evidence summaries
• prioritized recommendations
• assumptions and limitations
• a documented Google Colab notebook containing the complete analytical workflow.

---

6. Data Model & Schema

Dataset / Table: Global Findex 2025

Analysis Dataset

Field	Type	Description	Example Value
wpid_random	Identifier	Unique respondent identifier	123456789
wgt	Numeric	Survey weight used for population level estimates	0.674757
account	Binary	Whether the respondent has a financial account	1
dig_account	Binary	Whether the respondent has a digitally enabled account	1
anydigpayment	Binary	Whether the respondent made or received a digital payment	1
saved	Binary	Whether the respondent saved money	1
borrowed	Binary	Whether the respondent borrowed money	1
internet_use	Binary	Whether the respondent used the internet	1
female	Categorical	Raw gender variable	1
age	Numeric	Respondent age	32
educ	Categorical	Raw education level code	2
inc_q	Ordinal	Income quintile from 1 (poorest) to 5 (richest)	3
emp_in	Categorical	Raw workforce status variable	1
urbanicity	Categorical	Raw residence classification	2
Gender	Derived	Readable gender category	Female
Education	Derived	Readable education category	Secondary
Income_Quintile	Derived	Readable income quintile category	Q3 - Middle
Employment	Derived	Readable workforce status category	In Workforce
Residence	Derived	Readable residence category	Urban
Age_Group	Derived	Respondent age band	25-34
Account_Status	Derived	Readable financial account status	Account Owner
Digital_Account_Status	Derived	Readable digital account status	Digital Account
Digital_Payment_Status	Derived	Readable digital payment status	Digital Payment User
Internet_Status	Derived	Readable internet use status	Internet User
Saving_Status	Derived	Readable saving behavior status	Saved
Borrowing_Status	Derived	Readable borrowing behavior status	Borrowed
Financial_Behavior	Derived	Combined saving and borrowing behavior	Saver & Borrower
Financial_Engagement	Derived	Analytical financial engagement segment	Digitally Active

Financial Engagement Definition

Segment	Definition	Example
Financially Excluded	Does not own a financial account	account = 0
Basic Access	Owns an account but does not use digital payments	account = 1, anydigpayment = 0
Digitally Active	Owns an account and uses digital payments	account = 1, anydigpayment = 1

Note: These financial engagement segments were created specifically for this analysis and are not official Global Findex classifications.

---

7. Analysis & Metrics

Analytical Approach & Framework

The analysis follows three connected stages:

Financial Access → Digital Activation → Financial Behavior

Research Question 1 - Financial Access

Which demographic groups show the largest gaps in financial account ownership?

Key Metric

Weighted Account Ownership Rate

Weighted share of respondents within each demographic group who own a financial account.

Largest Observed Gaps

Factor	Lowest Group	Lowest Rate	Highest Group	Highest Rate	Gap
Education	Primary or Less	33.62%	Tertiary	82.15%	48.53 pp
Income	Q1 – Poorest	33.05%	Q5 – Richest	64.47%	31.42 pp
Age	55–64	34.02%	25–34	55.89%	21.87 pp

Statistical Validation

Factor	Cramér’s V	p-value
Education	0.249	<0.001
Income	0.203	<0.001
Employment	0.151	<0.001
Age	0.109	0.035
Residence	0.056	0.075
Gender	0.055	0.080

Education and income were the strongest observed demographic associations with financial account ownership.

Research Question 2 – Digital Activation

Among account owners, which groups show the largest differences in digital payment adoption?

Digital Activation Rate

Among financial account owners:

• 80.37% use digital payments.
• 19.63% remain digitally inactive.

Largest Observed Digital Adoption Gaps

Factor	Lowest Group	Lowest Rate	Highest Group	Highest Rate	Gap
Age	55–64	50.04%	15–24	94.33%	44.29 pp
Internet	Non-Internet User	48.29%	Internet User	87.52%	39.23 pp
Income	Q1 – Poorest	53.17%	Q2 – Lower Middle	86.42%	33.25 pp

Statistical Validation

Factor	Cramér's V	p-value
Age	0.370	<0.001
Internet Usage	0.354	<0.001
Residence	0.268	<0.001
Income	0.246	<0.001
Education	0.245	<0.001
Gender	0.020	0.651
Employment	0.014	0.755

Age and internet usage showed the strongest observed associations with digital payment adoption among account owners. |

Research Question 3 - Financial Behavior

How do saving and borrowing behaviors differ across financial engagement levels?

Financial Behavior Segments

Behavior	Weighted Share
Saver & Borrower	40.49%
Borrower Only	31.48%
Neither	14.88%
Saver Only	13.14%

Behavior by Financial Engagement

Financial Engagement	Saving Rate	Borrowing Rate
Financially Excluded	36.75%	65.66%
Basic Access	42.95%	66.50%
Digitally Active	77.10%	81.11%

Observed Gaps * Saving gap: 40.35 percentage points * Borrowing gap: 15.45 percentage points

Statistical Validation

Behavior	Cramér’s V	p-value
Saving	0.313	<0.001
Borrowing	0.145	<0.001

Financial engagement showed a substantially stronger association with saving behavior than borrowing behavior.

---

Methods Used

Method	Purpose	Applied To
Survey Weighted Analysis	Estimate percentages while accounting for respondent survey weights	Account ownership, digital payment adoption, saving, and borrowing rates
Descriptive Analysis	Summarize financial and demographic characteristics of respondents	Overall financial inclusion indicators
Demographic Segmentation	Compare financial outcomes across population groups	Age, gender, education, income, employment, and residence
Behavioral Segmentation	Group respondents according to saving, borrowing and financial engagement	Financial Behavior and Financial Engagement segments
Percentage-Point Gap Analysis	Measure the difference between the highest and lowest performing groups	Account ownership and digital payment adoption
Cross Tabulation	Examine relationships between categorical financial indicators	Account ownership, digital usage, saving, and borrowing
Chi-Square Test of Independence	Test whether categorical variables show statistically detectable associations	RQ1, RQ2, and RQ3
Cramér’s V	Measure and compare the strength of categorical associations	Demographic and behavioral factors
Data Visualization	Communicate patterns, gaps, and comparisons visually	KPI summaries and research question findings
Evidence Synthesis	Combine descriptive and statistical evidence into actionable findings	Final insights and recommendations

---

8. Key Insights

Insight 1: Education and income define the largest financial access gaps
Account ownership varied substantially across socioeconomic groups, with a 48.53 percentage point gap by education and a 31.42 percentage point gap by income. Statistical testing also identified education and income as the strongest demographic associations with account ownership, suggesting that financial access remains particularly uneven across education and income levels.

Insight 2: Account ownership does not guarantee digital participation
Although 80.37% of account owners use digital payments, 19.63% remain digitally inactive. This suggests that gaining access to a financial account is only one stage of financial inclusion and that digital activation should be considered separately from account acquisition.

Insight 3: Age and internet usage are key differentiators of digital adoption
Among account owners, digital payment adoption showed a 44.29 percentage point gap across age groups and a 39.23 percentage-point gap between internet and non-internet users. These were also the strongest statistical associations with digital payment adoption, suggesting that digital inclusion efforts may need to address age-related and connectivity-related barriers.

Insight 4: Financial engagement differentiates saving more strongly than borrowing
Saving showed a 40.35 percentage point gap across financial engagement levels, compared with only 15.45 percentage points for borrowing. Borrowing remained relatively common even among financially excluded respondents, suggesting that financial exclusion does not necessarily mean financial inactivity and that inclusion should be evaluated across multiple financial behaviors.

---

9. Recommendations

Priority	Recommendation	Based On	Suggested Owner
High	Prioritize simplified and assisted financial access initiatives for lower education and lower-income groups	Insight 1: Education and income show the largest financial access gaps	Financial Inclusion / Retail Banking
High	Track digital activation separately from account acquisition and provide post opening support for digitally inactive account owners	Insight 2: 19.63% of account owners remain digitally inactive	Digital Banking / Customer Engagement
High	Strengthen assisted digital onboarding and digital literacy support for older and non internet users	Insight 3: Age and internet usage are key differentiators of digital adoption	Digital Banking / Branch Operations
Medium	Develop accessible saving initiatives for financially excluded and basic access groups	Insight 4: Saving differs more strongly across financial engagement levels than borrowing	Deposits / Product Strategy
Low	Explore broader financial inclusion measures that combine account ownership, digital activity, saving, and borrowing rather than relying on a single indicator	Insight 4: Financial exclusion does not necessarily mean financial inactivity	Financial Inclusion / Consumer Research

---

10. Assumptions & Limitations

Assumptions

• Survey weights represent the target population: The provided survey weight (wgt) was treated as the appropriate weighting variable for calculating population-level descriptive percentages.

• Source variable definitions are valid: Variables such as account ownership, digital-payment usage, saving, borrowing, education, employment, and internet usage were interpreted according to the definitions and coding provided in the Global Findex dataset.

• Binary indicators represent reported participation: Variables such as account, anydigpayment, saved, and borrowed were treated as indicators of whether a respondent reported the corresponding financial behavior.

• Financial engagement was simplified into three analytical segments: Respondents were classified as Financially Excluded, Basic Access, or Digitally Active to support the analysis. These categories were created specifically for this project and are not official Global Findex classifications.

• Demographic categories were grouped for analytical clarity: Variables such as age, education, income, employment and residence were converted into readable categories or grouped bands to make comparisons easier to interpret.

• Statistical significance was evaluated at the 5% level: A p-value below 0.05 was used as the threshold for identifying statistically detectable associations in the sample.

Limitations

• The analysis is based on a cross-sectional survey: The data captures respondents at a particular point in time. Therefore, the findings identify patterns and associations but cannot establish causal relationships.

• The Philippine sample contains 1,000 respondents: While survey weights were used for descriptive estimates, some subgroup results may be based on relatively small numbers of respondents and should be interpreted cautiously.

• Financial behaviors are self-reported: Responses related to saving, borrowing, account ownership, and digital payment usage may be affected by recall error, misunderstanding, or response bias.

• Behavioral indicators provide limited detail: Variables such as saved, borrowed, and anydigpayment indicate whether an activity occurred but do not capture its frequency, monetary value, duration, intensity, or financial outcome.

• The statistical tests do not implement a full complex-survey design: Chi-square tests and Cramér’s V were used as supplementary sample level association measures. A more rigorous inferential analysis would account for the complete survey design where the necessary design variables are available.

• Potential confounding factors were not controlled simultaneously: The analysis evaluates demographic and behavioral relationships primarily through segmentation and categorical association tests. Multivariable modeling could determine whether associations remain after controlling for other characteristics.

• Country comparisons and time trends are outside the project scope: The analysis focuses only on the Philippines and does not determine whether the observed patterns are unique relative to other countries or have changed over time.

• Association should not be interpreted as causation: For example, higher digital payment adoption among internet users does not demonstrate that internet usage directly causes digital payment adoption. Other demographic, socioeconomic or behavioral factors may contribute to the observed relationship.

---

11. Future Enhancements

• Apply survey design aware statistical inference if full survey design variables become available.
• Build multivariable models for account ownership and digital payment adoption.
• Compare the Philippines with selected ASEAN economies.

---

12. Deliverables

Deliverable	Description	Location
Analysis Notebook	Complete Python analysis, transformations, statistical tests and visualizations	notebooks/Financial_Inclusion_Philippines_Portfolio.ipynb
GitHub README	Executive case study and key findings	README.md
Visualizations	Portfolio-ready charts	visuals/
Data Documentation	Dataset source and field definitions	data/README.md / docs/data_dictionary.md

---

13. Author

Clinton Taguba

Data Analyst | Aspiring Data Scientist & Data Engineer

• 🔗 https://www.linkedin.com/in/clintontaguba/
• 💼 https://clinttaguba-lab.github.io/
• 📧 tagubaclinton@gmail.com

---

Last updated: August 2026