Physical isolation and loneliness: Evidence from COVID lock-downs in Australia

Abstract

Loneliness contributes to mortality risk, comparable to the risks associated with smoking and obesity, although the causal determinants of loneliness remain less clear. This paper leverages mandatory stay-at-home orders in Australia as a natural experiment and employs data from a panel study to investigate the causal link between physical isolation and loneliness. By analyzing variations in the number of lockdown days experienced by respondents up until their interview dates, and utilizing difference-in-differences analyses with individual, region, and year fixed-effects estimations, we find, contrary to expectations, that the number of days in lockdown does not significantly impact loneliness. Our study examines cumulative, concurrent, and non-linear effects, and assesses external validity through community morale and peer effects during lockdowns using spatial analysis. Additionally, we delve into heterogeneous effects across various factors, such as income, age, personality, living arrangements, and remoteness, finding statistically and empirically insignificant effects. However, for extroverts and young people, we observe weak statistical significance. We investigate exclusion restrictions by analyzing factors including social contacts, internet access, job industry, and household characteristics in relation to loneliness; as well as time use and relationship satisfaction to better understand the underlying mechanism. Our study challenges the notion that 'being alone' and 'being lonely' are interchangeable concepts, providing the first empirical causal evidence of no links between the two. Furthermore, our findings refine earlier understandings of social isolation, highlighting that it likely encompasses factors beyond physical isolation.