The impact of technology on socio-economic outcomes has become increasingly salient in policy debates across both developed and developing countries. In developing contexts, digitisation holds promise for narrowing gaps in financial inclusion that persist due to limitations in physical infrastructure. Greater access to digital tools may enable households to save, invest and build resilience against shocks that affect both entire communities and individual households.
This article explores the relationship between digitisation and lower-income status using cross-sectional data from 7,363 respondents across seven Pacific Island countries, collected in 2024. The countries included are Fiji, Papua New Guinea, Samoa, Solomon Islands, Timor-Leste, Tonga and Vanuatu. While the analysis identifies a negative association between digitisation and lower-income status (used here as a proxy for poverty), the nature of the data does not permit causal inferences. The article concludes with a call for stronger data collection efforts to enable causal analysis in Pacific Island settings and other developing country regions.
Methodology
I undertook a regression analysis that accounts for a wide range of factors influencing lower-income status, with a particular focus on digitisation. The dependent variable is a proxy for poverty: lower-income status, defined as earning less than AUD 300 per fortnight. Due to limitations in income classification, more commonly used poverty thresholds, such as earning below USD 2 per day, could not be applied.
The primary variable of interest is a digitisation index measured on a scale from 1 to 18, covering four categories: access to digital devices, digital activities, online activities and digital services. This index was grouped into three categories: low (1–6), medium (7–12) and high (13–18). The analysis focuses on individuals with high digitisation scores.
The regression controls for gender and age to account for demographic effects, as well as occupation categories (such as employed, unemployed and student) to capture employment status. Education and industry sectors are included to reflect differences in human capital and job characteristics that may influence income outcomes.
Given the diversity of the countries in the sample, country fixed effects are included to account for unobservable country-specific factors. Urban–rural residence is also controlled for, recognising its strong influence on access to digital resources, employment opportunities and income levels.
Together, these controls help ensure that the observed relationship between digitisation and lower-income status is not driven by other socio-economic characteristics. Clustered standard errors at the country level are used to account for correlations within countries and improve the robustness of the estimates.
Despite these methodological features, the model cannot establish causality. The cross-sectional nature of the data means it is not possible to observe changes in digitisation and income over time within countries. Panel data, with repeated observations, would allow for more robust causal inference through the construction of treatment and control groups. As such, the findings should be interpreted as correlations rather than causal effects.
Key findings
The regression results indicate a statistically significant negative relationship between digitisation and lower-income status. Relative to within-country average levels, higher digitisation is associated with a reduction in the likelihood of being classified as lower-income of between 7 per cent and 40 per cent, with an average reduction of 21 per cent. This suggests that individuals with higher levels of digitisation are less likely to fall into the lower-income category, even after controlling for other socio-economic factors.
Country-level results show considerable variation relative to the national mean:
- Fiji: Around a 20 per cent reduction
- Papua New Guinea: Approximately a 15 per cent reduction
- Samoa: Roughly a 10 per cent reduction
- Solomon Islands: Around a 15 per cent reduction
- Timor-Leste: Approximately a 30 per cent reduction
- Tonga: Around a 20 per cent reduction
- Vanuatu: Approximately a 25 per cent reduction
Policy implications and recommendations
These findings are not causal and should be interpreted with caution. It is plausible that policies aimed at reducing poverty may also increase digital adoption, rather than digitisation directly reducing lower-income status. For example, improved access to quality education may simultaneously lower poverty and raise digital engagement, particularly when accompanied by digital literacy initiatives.
Given this complexity, drawing firm policy prescriptions from the available data is challenging. However, the variation across countries suggests that policy responses should be tailored to national and regional contexts. There is also clear value in sharing experiences and strengthening intra-regional collaboration across the Pacific.
More broadly, this analysis underscores the urgent need for higher-quality data in the Pacific. As econometric methods increasingly focus on causal inference, governments and development partners should prioritise the collection of data that can support rigorous analysis and evidence-based policymaking. Follow-up work should focus on building panel datasets and carefully designed experimental settings to better understand how digital and development policies affect the lives of people experiencing poverty.
Professor Alberto Posso is Head of the Department of Accounting, Finance and Economics, Griffith Business School.
