تئوری نئو ماتریالیستی و رابطه زمانی بین نابرابری درآمد و تغییر طول عمر

This study uses a neo-materialist perspective to develop theoretical predictions regarding temporal ties between income inequality and change in population health. The argument focuses on the relationship between income inequality and adoption of longevity-enhancing innovations. It asserts that longevity change should be influenced by preexisting levels of income inequality and that, consequently, income inequality can cause differential longevity improvement across jurisdictions even if inequality levels remain unchanged. State-level U.S. data from 1970 to 2000 are used to jointly model the effects of initial levels and change in income inequality on 10-year life expectancy change. Results confirm that states with higher levels of inequality experienced less subsequent improvement in life expectancy. Contrary to findings from prior research, analyses also reveal a strong negative association between change in inequality and change in longevity once initial levels of inequality and other state characteristics are controlled. Finally, direct tests of the relationship between income inequality and the adoption of innovations in quality of medical care indicate that the two are highly related and that differences in the average quality of care can account for the negative cross-sectional association between income inequality and life expectancy.

One great demographic shift in economically advanced nations over the past two centuries has been the dramatic growth in life expectancy. Empirical work has attributed longevity gains to improvements in material standard of living and the implementation of health-enhancing innovations in public health and medicine, though scholars have debated the relative importance of each (Catalano and Frank, 2001, Deaton, 2006, Fogel, 2004, McKeown, 1976, Preston and Haines, 1991 and Szreter, 1988). A relatively recent literature has asserted that income inequality also has an important influence on health and mortality. Wilkinson (1992) went so far as to assert that it is the principal cause of health differences among wealthy nations. After an initial flurry of supportive cross-sectional findings, the empirical tide turned against the inequality hypothesis in more recent years (see Lynch, Harper, & Davey Smith, 2003 for a discussion), though supportive results do continue to emerge (e.g., Backlund et al., 2007, Ram, 2005, Subramanian and Kawachi, 2006 and Zimmerman and Bell, 2006). One indirect critique of the inequality hypothesis is that focusing on inequality is unwarranted because it is not a major cause of change in mortality/longevity. Lynch, Davey Smith, Harper, and Hillemeier (2004) find that changes in inequality over time fail to match up in any consistent way with changes in mortality. Although it is readily apparent that inequality, in isolation, is not a principal driver behind life expectancy improvements over time, it does not necessarily follow that income inequality has no important role in influencing changes in longevity or, in particular, differential change across populations. Ultimately, improvements in health outcomes result from adoption of beneficial innovations – whether they be innovations that improve medical care, provide for public goods such as drinking water or waste disposal, increase food production, etc. And one claim of “neo-materialist” theory (Davey Smith, 1996 and Lynch et al., 2000) is that income inequality harms population health because it weakens societies' willingness to make investments that promote the common good. If true, inequality could play an important role in affecting the rate of health change by influencing adoption of the innovations that are proximal determinants of heath improvements. An important issue, particularly when discussing the relationship between inequality and change in health outcomes, is what the temporal ties between inequality and health look like. A number of authors have discussed the question of latency periods between exposure and emergence of observable health outcomes ( Blakely et al., 2000, Lynch et al., 2004, Mellor and Milyo, 2003 and Subramanian and Kawachi, 2003). In this paper I use a neo-materialist perspective to discuss an issue that has received less attention: the temporal relationship between inequality as a distal cause and the adoption of innovations that are proximate determinants of health change. The argument I present asserts a tie between longevity change and preexisting levels of inequality. It does not necessarily include (nor exclude) links between longevity growth and change in inequality. Rather, I argue that within a dynamic technological environment we should observe differential improvement in life expectancy between low and high-inequality populations, even in the absence of any differential change in inequality. One implication of the argument is that results from previous empirical studies of change in inequality and change in population health are likely to be biased by failure to account for the effect of initial levels of inequality on subsequent health change. I test the proposed temporal ties using data on life expectancy and income inequality in the United States for a three-decade period, from 1970 to 2000. The tests include examinations of both the relationship between levels of inequality and subsequent longevity change, and of how observed change-on-change associations are affected by the inclusion of controls for initial levels of inequality and other state characteristics. I also use data on the quality of medical care to directly examine the relationship between income inequality and adoption of health-enhancing innovations. Income inequality, innovation adoption, and longevity improvement A substantial literature is devoted to the causes of longevity improvements in wealthy nations. In influential work, McKeown (1976) cited improved living standards as the predominant contributor. Though not denying the value of factors such as improved nutrition, subsequent work suggests that increased life expectancy in the West was initially attributable primarily to public health improvements such as provision of clean water and sewers and, more recently, to medical innovations such as antibiotics and immunizations (Colgrove, 2002, Cutler and Miller, 2005, Preston and Haines, 1991 and Szreter, 1988). In either case, the gains were largely due to the development and implementation of new knowledge. Of course, providing access to clean water or universal immunizations requires financial resources that are still out of reach for many populations. But even among societies endowed with the necessary resources, adoption of public health-enhancing innovations can still vary widely (e.g., Jencks et al., 2000) depending on social or institutional characteristics. For instance, Evans (1987) describes how intense socioeconomic stratification and concentration of power deterred the installation of a modern water filtration plant in 19th century Hamburg. This failure left the city vulnerable to a deadly cholera epidemic in 1892 that was avoided by other German cities that had pursued sanitary reforms. Szreter and Woolcock (2004) document a similar case in 19th century England, where class divisions blocked the construction of clean water and sewage systems. In a more contemporary example, Boyce, Klemer, Templet, and Willis (1999) find that states in which power is distributed less equally – partially as a function of income inequality – engage in less environmental protection and in turn have lower environmental quality, a correlate of poorer population health. These cases fit neo-materialist assertions that economic inequality affects population health by means of investment in health-enhancing infrastructure.1 “Infrastructure” refers to any number of factors ranging from sewer systems to access to high quality medical care to pollution abatement laws (Lynch et al., 2000). One aspect of the theory that has been largely ignored is its potential implications regarding the temporal ties between inequality and health. I argue below that a neo-materialist perspective suggests temporal relationships that have generally not been reflected in modeling strategies used to analyze the potential influence of income inequality on change in health outcomes. Of course, most empirical work on income inequality and population health does not address change in health outcomes at all, but rather uses cross-sectional analyses with levels of health outcomes at a point in time as the outcome (Macinko, Shi, Starfield, & Wulu, 2003). A smaller body of work does focus on how inequality is related to health change, although the motivation has often resulted less from substantive interest in health change than from concern over heterogeneity bias in cross-sectional analyses. Those studies—using both cross-national data (Beckfield, 2004 and Mellor and Milyo, 2001) and data from U.S. states (Mellor & Milyo, 2001)—have adopted either first-differences or fixed-effects approaches to net out the effects unobserved, temporally stable characteristics of jurisdictions on persisting differences in the levels of population health across those jurisdictions over time. In contrast to much cross-sectional work, those studies find no harmful effects of income inequality on population health. Similarly, in descriptive analyses, Lynch et al. (2004) observe that during the late 1900s the greatest declines in mortality occurred in U.S. regions where inequality grew most. Identification of effects in first-difference and fixed-effect analyses relies on associations between change in inequality and contemporaneous change in mortality. But the motivations presented by the authors of studies that use those approaches are based on methodological concerns, not on substantive theory that emphasizes change-on-change effects. And the preceding neo-materialist illustrations imply a tie between inequality and health that, in fact, requires no change-on-change relationship. Methodological concerns regarding unobserved heterogeneity are, of course, entirely valid. The problem lies in taking recourse to models that rely on sources of identification that exclude mechanisms implied by substantive theory. And if income inequality influences longevity change through its effect on investment in health-enhancing resources, then it is entirely possible that cross-state differences in inequality could produce differential change in longevity even if levels of inequality themselves remained unchanged. Consequently, inequality could affect differential longevity change in ways that would be missed entirely by empirical specifications that only identify change-on-change effects. Since the technological environment is dynamic, societies are continually presented with health-enhancing innovations in which they could invest. Those that invest most aggressively will experience the greatest improvements in longevity. And, according to a neo-materialist perspective, economically egalitarian societies will tend to be more aggressive adopters (Lynch et al., 2000 and Szreter and Woolcock, 2004). Under this conceptualization, the relationship between inequality and change in health within an environment of rapid innovation is dependent primarily on levels of inequality at the time that innovations emerge rather than on change in inequality. To illustrate, suppose that cross-jurisdictional differences in inequality were to remain constant over time. If it is true that jurisdictions characterized by greater income equality are more likely than less equal jurisdictions to adopt health-enhancing innovations, then, during times of health-enhancing innovation, life expectancy will grow more in jurisdictions where levels of inequality are lower, as a result of more aggressive adoption of those innovations. Given that framework, an empirical researcher who was interested in sociopolitical mechanisms tying inequality to health would want to look for associations between levels of inequality and later change, not solely change-on-change effects—though those may also exist, whether for neo-materialist reasons or through other mechanisms. An important implication of this conceptualization is that the effect of inequality on longevity change will be contingent on the rate of health-enhancing innovation. During periods of rapid and accelerating innovation-driven health improvements, differential improvements in health between aggressive adopters and less aggressive adopters will be large. When such change is slower or decelerating, then the differences between aggressive and non-aggressive adopters will be smaller. If inequality is a determinant of innovation adoption, then differences in longevity change between high- and low-inequality states should also vary positively depending on the overall rate of innovation change. Although the preceding framework makes income inequality's effects contingent on an exogenous social condition (the pace of health-enhancing innovation), it is nonetheless a very real causal factor under conventional notions of counterfactual causality. Under the adoption of innovation mechanism, inequality is not a sufficient cause, since the magnitude of its effects depends upon the innovations that are available to be adopted. But as long as we assume that health-enhancing innovation will exist in our society, the argument implies that, all else equal, longevity improvement will be greater where levels of inequality are lower. To recapitulate, the preceding discussion implies two empirical predictions: Hypothesis #1: Improvements in longevity will tend to be greater in states where preexisting levels of income inequality are lower. Hypothesis #2: The strength of the relationship between levels of income inequality and subsequent longevity change will be contingent on the pace of health-enhancing innovation. Those two predictions are silent regarding possible relationships between change in inequality and change in mortality. However, Hypothesis #1 does suggest a source of potential bias in the change-on-change analyses that were discussed earlier. Changes in income inequality in the United States in recent decades appear to have been characterized by regression-to-the-mean, with the smallest increases occurring in the states with the highest initial levels (Barrilleaux & Davis, 2003). The conjunction of higher levels of inequality causing smaller mortality gains (Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996) and high-inequality states also experiencing smaller increases in inequality lead to a third empirical prediction: Hypothesis #3: Failure to control for the effect of initial levels of inequality on subsequent longevity positively biases observed associations between inequality change and longevity change. That is, the omission makes the effects of inequality change appear more beneficial (or less harmful) than they actually are. The following section details the paper's analytic strategy for testing those three