Academic journal article Demographic Research

Does Income Inequality Really Influence Individual Mortality? Results from a 'Fixed-Effects Analysis' Where Constant Unobserved Municipality Characteristics Are Controlled

Academic journal article Demographic Research

Does Income Inequality Really Influence Individual Mortality? Results from a 'Fixed-Effects Analysis' Where Constant Unobserved Municipality Characteristics Are Controlled

Article excerpt

Abstract

There is still much uncertainty about the impact of income inequality on health and mortality. Some studies have supported the original hypothesis about adverse effects, while others have shown no effects. One problem in these investigations is that there are many factors that may affect both income inequality and individual mortality but that cannot be adequately controlled for. The longitudinal Norwegian register data available for this study allowed municipality dummies to be included in the models to pick up time-invariant unobserved factors at that level. The results were compared with those from similar models without such dummies. The focus was on mortality in men and women aged 30-79 in the years 1980-2002, and the data included about 500000 deaths within 50 million person-years of exposure. While the models without municipality dummies suggested that income inequality in the municipality of residence, as measured by the Gini coefficient, had an adverse effect on mortality net of individual income, the results from the models that included such dummies were more mixed. Adverse effects appeared among the youngest, while among older men, there even seemed to be beneficial effects. In addition to illustrating the potential importance of controlling for unobserved factors by adding community dummies (doing a 'fixed-effects analysis' according to common terminology in econometrics), the findings should add to the scepticism about the existence of harmful health effects of income inequality, at least in the Nordic context.

(ProQuest: ... denotes formulae omitted.)

1. Introduction

The idea that income inequality may weaken people's health and increase mortality has attracted much interest in recent years (see e.g. reviews by Kawachi 2000; Wagstaff and van Doorslaer 2000; Lynch, Davey Smith, Harper et al. 2004; Wilkinson and Pickett, 2006). Some investigators have used an ecological approach to check whether societies (e.g. countries, states, municipalities) with large variation in income fare worse than others in terms of health and mortality, and many of these studies, but far from all, have concluded that there indeed seems to be such a relationship. A particularly robust pattern has been seen within the United States. However, a positive relationship between income inequality and mortality in an ecological analysis may simply reflect diminishing individual health returns to increasing individual income (see elaboration below). A more interesting question is whether individual health and mortality are adversely affected by the income inequality in the community net of individual income (see review of possible reasons below). This calls for a multilevel approach. Unfortunately, the answers to this question have been rather mixed. Whereas some studies have supported the hypothesis about adverse effects of income inequality - and especially American ones where inequality has been measured at the state level - no effects, or in a few cases even beneficial effects, have shown up in other investigations.

When assessing the effect of income inequality, one should of course control for characteristics that are likely to influence both income inequality and health. One simple example of such a potential confounder is whether the place is urban or rural: urban societies typically offer a diversity of jobs, and may therefore also produce large income differences, in addition to showing high mortality for many other reasons, at least in rich countries. While an urban-rural indicator may often be available to the researcher, there are a number of other socioeconomic, political, cultural or environmental factors that may also be confounders, and that may be difficult to measure, or at least not be available in the data at hand. Some of these unobserved factors may be approximately time-invariant, for example because they are somehow linked to the physical characteristics of the communities, and these can be captured by including a 0/1-dummy variable for each community. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.