Digital wage payments have been shown to be a driver of both financial inclusion and financial health. While it has become the norm for people in developed economies to receive their wages digitally, this is not the prevailing situation in the majority of developing economies. In Africa, where employment is mainly in an informal sector that largely relies on cash payments, the adoption of digital wage payments holds immense potential for improving financial inclusion and health. In this post, we explore how machine learning predicts the likelihood of individuals in Africa receiving their wages digitally.

The Landscape of Wage Payments in Africa 

According to the United Nations Department of Economic and Social Affairs, Africa is experiencing remarkable growth in its labor population, with an estimated 2 million persons in the working age population entering the workforce each month. However, employment opportunities predominantly lie within the informal sector, which poses unique challenges. The informal sector, characterized by its lack of formal contracts and social protection, mostly relies on cash transactions. This reliance on cash often hinders financial inclusion and limits access to essential financial services for those employed in the informal sector. For instance, cash payments in the informal sector typically lack the formal structures and protections associated with formal employment. Workers in the informal sector may miss out on benefits such as health insurance, retirement plans and social security; which perpetuates economic vulnerability and limits opportunities for long-term financial stability. 

The Digital Frontier: Empowering Workers 

Research done in the area of digital payments has demonstrated digital wage payments to be more beneficial to workers and their respective enterprises as compared to cash-based wage payments. By embracing digital wage payments, individuals enjoy heightened security as the risks associated with carrying and storing physical cash are mitigated. Moreover, the convenience of digital payments eliminates the need for in-person transactions, saving time and effort for individuals and productivity hours for enterprises. Moreover, digital wage payments open doors to financial empowerment, granting individuals access to a broader range of financial services such as savings, loans, and insurance 

The Goals of Our Study

Given the importance of digital wage payments, this study aimed to build a model that could predict the likelihood of an individual in Africa to be paid digitally. Furthermore, the study sought to establish the most significant factors in Africa that predict an individual’s likelihood of receiving digital wage payments.

The Predictive Model: How Machine Learning Works 

To achieve our goals, we utilized machine learning methods. Machine learning holds immense promise in terms of how we analyze data and make informed decisions. This is more so in tasks whose main goal is making predictions. Fundamentally. machine learning involves the development of algorithms that enable computers to learn from and analyze large volumes of data, identify patterns that may not be obvious to humans, and make predictions on new data based on learned information. In our case, by leveraging wage payment data, demographic insights, and other relevant factors from the individual-level data for the year 2021 (obtained from the Global Findex Database), we built a predictive model that enabled us to estimate the likelihood of receiving digital wage payments for individuals from 25 African countries. Moreover, by demystifying the principles behind our machine learning model, we were able to grasp some of the important predictors of receiving digital wage payments and how they impact the predictions. 

Deciphering the Most Important Factors  

What Significantly Predicts One's Digital Wage Prospects 

When it comes to digital wage payments in Africa, a multitude of factors come into play and diversely affect the likelihood of individuals receiving their wages digitally. Of the factors that were considered in this study, account ownership was the most significant factor in predicting whether an individual received digital wages or not. The second most important factor was the economy from which an individual comes from, with individuals from South Africa, Namibia, Mozambique, and Kenya more likely to be paid digitally as compared to their counterparts. This finding suggests that broader country-related factors such as infrastructural development levels and regulations may be even more relevant than most individual-related factors such as one's gender or economic status. 

How do Some of these Factors Affect One's Digital Wage Prospects 

1. Account Ownership 

• As per our predictive model, those who own an account (bank or mobile money) are more likely to receive their wages digitally than those without one. Moreover, those who do not own an account have amongst the lowest chances of digital wages. 

• The above highlight that there is a high correlation between financial access (and more generally financial inclusion) and digital wage payments.

• Owning an account is normally the most fundamental pre-requisite for receiving digital wage payments and our results are therefore quite intuitive and expected. 

2. Income Group 

• Our predictive model suggests a monotonic increasing relationship between one's income level and their likelihood of being paid digitally with one's chances of receiving digital wages rising as their income level rises. 

• Amongst the demographic factors, income was the most significant predictor. 

• The interaction effects between income and other variables were quite stronger than the effect of income alone on the prediction. This was more so the case for the upper-income quintiles. For instance, those with the highest education level were more concentrated amongst the richest 20% and thus had a relatively higher likelihood of receiving digital wages than other subgroups. 

• Even though our predictive model is driven by correlational relationships, it still supports inferences from causal studies that suggest for the poorest individuals to be targeted for policy action. 

3. Wage Provider 

• Our model reveals that those in the public sector are more likely to be paid digitally than those in the private sector. 

• The proportion of those without an account amongst private sector workers is significantly larger than those without an account in the public sector.

• This may be attributable to the informal sector (which dominates the private sector) which operates outside the reach of government and relies mostly on cash-based transactions to make payments of wages.

4. Mobile Phone Ownership and Usage for Financial Transactions

• Both ownership and frequent usage result in higher chances of receiving digital wages as per our model.

• Amongst all the categories of the variables included in the model, the usage of mobile phones for financial transactions had the greatest positive impact on one's likelihood of receiving digital wages.

• These findings are consistent with initiatives already in place across the African continent meant to promote mobile phone ownership and usage, and inadvertently financial inclusion; examples of which include Kenya's Safaricom-backed Lipa Mdogo Mdogo, Rwanda's marathons as well as M-KOPA Solar.