Going “beyond the GDP” in the digital economy: exploring the relationship between internet use and well-being in Spain

The study investigates how internet use relates to subjective well-being (SWB) in the context of ongoing digital transformation and persistent digital inequalities. Prior work shows ICT diffusion is uneven across socioeconomic groups and that well-being is shaped by personal and contextual factors. The paper’s goal is to disentangle the relationship between internet use (decision and intensity) and multiple dimensions of well-being in Spain, explicitly accounting for socioeconomic determinants of digital use. It emphasizes a beyond-GDP perspective in policy, framing well-being outcomes as central targets of the digital decade and assessing whether digital engagement enhances or displaces aspects of well-being.

Early research on the social impacts of internet use proposed competing hypotheses: enhancement (reinforcing offline ties and social capital) versus displacement (substituting offline activities and potentially increasing isolation). Initial evidence suggested negative effects on well-being (Kraut et al. 1998), which largely diminished in later follow-ups (Kraut et al. 2002). Subsequent studies report mixed findings—positive, negative, and null effects—across varied well-being outcomes (happiness, life satisfaction, mental health, social participation) and different technologies (general internet use, social media, device ownership). Heterogeneity also stems from diverse populations (general adult samples versus specific groups like teenagers or older adults) and methods, with relatively few experimental or quasi-experimental designs (e.g., Facebook deactivation increasing happiness and life satisfaction; Allcott et al. 2020). A key gap is treating digital use as exogenous; literature on digital inequalities shows use is rooted in socioeconomic characteristics that also shape well-being. This paper addresses endogeneity concerns by explicitly modeling the socioeconomic origins of internet adoption and intensity before estimating effects on SWB.

Data: Microdata for Spain from the European Social Survey (ESS) Rounds 8 and 9 (2016 and 2018), yielding approximately 3,614 observations. Key constructs: (1) Internet use: Daily Internet (=1 if daily or mostly daily use) and Internet time (minutes per typical day for daily users); Internet Speed (regional average maximum Mbps at NUTS II). (2) Subjective well-being (SWB): Happy (0–10), Satlife (0–10), Meetings (1–7, frequency of meeting friends/relatives/colleagues), Discuss (0–6, number of people to discuss intimate/personal matters), Sactiv (1–5, social activity participation vs. peers). (3) Socioeconomic covariates: Female, Employed, Partner, Bad Health, Hampered (hampered in daily activities by illness/disability/infirmity/mental health problem), Age (continuous), Age>60, Domicile (5 categories), Rural, subjective income Hincfel (4 categories), Income (below/above 5th decile or DK/DA), Education (ISCED 1–7, treated as continuous), and Round 9 indicator. Model: A system of three simultaneous equations estimated separately for each SWB dimension (five systems total): (i) probability of Daily Internet use (full sample), (ii) Internet time for daily users (intensity equation), and (iii) SWB outcome as a function of Internet time and covariates (daily users). Internet Speed is included in the first two equations to capture infrastructure quality. Interaction terms assess heterogeneous effects of Internet time for three groups of policy interest: Hampered (disability/health-related limitations), Female, and Age>60. To ensure convergence, some categorical variables were recoded or treated as continuous. Internet time enters linearly; a squared term was tested but not retained due to insignificance. Descriptive statistics indicate 70% daily users and an average of 224 minutes per day among users. Pairwise correlations show weak but positive intercorrelations among SWB measures; Internet time correlates negatively with happiness and life satisfaction and positively with meetings, discuss, and social activity, with only the correlation with Meetings reaching statistical significance. Speed was not significant in well-being equations in checks.

- Digital use determinants: Both Daily Internet use and Internet time are strongly patterned by socioeconomic factors. Higher education and income increase the likelihood of daily use and time spent; age reduces both. Urban residence and higher regional Internet Speed predict greater use and more time. Daily use rose between 2016 and 2018 (positive Round 9 effect). - Effects on well-being depend on dimension: Greater Internet time is associated with lower happiness and life satisfaction and fewer in-person meetings, but with more people to discuss intimate matters and greater social participation relative to age peers. - Approximate coefficients (direction and significance): - Happiness: Internet time ≈ −0.0012 (p<0.01) - Life satisfaction: Internet time ≈ −0.0010 (p<0.01) - Meetings: Internet time ≈ −0.0005 (p<0.01) - Discuss: Internet time ≈ +0.0015 (p<0.01) - Social activity (Sactiv): Internet time ≈ +0.0008 (p<0.01) - Heterogeneity: - Hampered: The negative association between Internet time and Meetings is stronger for hampered individuals; no other consistent significant interactions for other SWB outcomes. - Gender: No differential associations by gender in the Internet time–SWB links. - Older adults (Age>60): The negative association between Internet time and happiness is attenuated (interaction positive). The positive association between Internet time and social participation is stronger for older users. - Socioeconomic covariates and SWB: - Bad Health negatively associated with all SWB outcomes; Hampered negatively with happiness. - Partner positively associated with happiness and life satisfaction, but negatively with Meetings and Discuss. - Employed positively associated with life satisfaction; negatively associated with Meetings. - Domicile: Town/small city residence positively related to life satisfaction and social life variables; country village/countryside negatively related to happiness and social life. - Subjective income hardship (higher Hincfel category numbers) is negatively related to all SWB measures. - Descriptives: 70% daily users; average Internet time among users is 224 minutes/day.

The findings demonstrate that once socioeconomic determinants of digital engagement are accounted for, the intensity of internet use exhibits mixed associations with well-being across dimensions. Negative links with happiness, life satisfaction, and face-to-face meetings alongside positive links with personal connections and social participation suggest a substitution of offline social interactions with online or digitally mediated forms, consistent with the displacement hypothesis in certain domains while indicating enhancement of network breadth and participation in others. Digital inequalities continue to structure who uses the internet and how intensively, thus indirectly influencing well-being outcomes. For vulnerable groups, hampered individuals experience greater reductions in face-to-face meetings as time online increases, whereas older adults see attenuated negative happiness effects and amplified gains in social participation. Policy efforts aimed at improving societal well-being through digitalization must therefore recognize the multidimensional nature of well-being and the possibility that aggregate net effects may be close to zero as positive and negative impacts offset each other across dimensions.

The paper contributes by modeling internet use as an endogenous, socioeconomically driven process and by evaluating its effects across multiple well-being dimensions using ESS microdata for Spain (2016, 2018). It shows that more time online is associated with lower happiness and life satisfaction and fewer face-to-face meetings, but more intimate discussion partners and greater social participation compared with peers, indicating partial substitution of in-person interactions with digital ones. Effects vary across groups: hampered individuals show stronger displacement in meetings, while older adults experience weaker negative effects on happiness and stronger positive effects on participation. Policy implications include tailoring digital transition policies to specific well-being targets and acknowledging that access alone may not yield well-being gains—especially for vulnerable groups. Future research should leverage richer, activity-level internet use data to distinguish effects by online service type and address data access asymmetries between researchers and private platforms to better assess causal links between digital behaviors and well-being.

- Measurement: Well-being outcomes are subjective self-reports and may involve comparative judgment biases (e.g., Sactiv compares to peers). - Digital behavior granularity: ESS provides coarse measures (daily use, minutes), lacking detail on types of online activities, which may differentially affect well-being. - Causality: Despite modeling endogeneity of use via a three-equation system, observational data limit causal inference; unobserved confounding may remain. - Sample/estimation: Convergence constraints necessitated recoding and treating some categorical variables as continuous, potentially smoothing heterogeneity. - Infrastructure: Internet Speed was not significant in well-being equations, suggesting potential omitted-context factors; regional speed may proxy broader infrastructural or socioeconomic conditions imperfectly.