Do not go breaking your heart: Do economic upturns really increase heart attack mortality?

https://doi.org/10.1016/j.socscimed.2007.04.015Get rights and content

Abstract

Several recent papers in the literature have documented a pro-cyclical effect between business cycles and mortality. In this paper, I explore the relationship between business cycles and incidence, mortality and lethality in acute myocardial infarction (AMI) in Sweden. The sample consists of 21 Swedish regions during the period 1987–2003. Results from the panel data estimations indicate that the business cycle effect is insignificant on overall rates of incidence, mortality and lethality. However, a counter-cyclical and significant effect is found in most specifications for those in prime working age between 20 and 49. Hence, previous recent results from the literature cannot be taken as universal for other countries or settings. It is also shown that a higher share of women, highly educated and non-foreigners decrease incidence and mortality.

Introduction

Several recent papers exploring the association between mortality and business cycles have found that short-term economic upturns increase mortality, a pro-cyclical effect. This is consistent with most of the major causes of mortality, such as cardiovascular mortality in general and acute myocardial infarction (AMI) or heart attack mortality in particular. Early papers on the relationship between the business cycle and mortality tended to use time-series data that indicated a counter-cyclical relationship. The recent papers, on the other hand, have made use of panel data. The panel data approach controls for omitted variable bias due to the possibility of using fixed effects estimation. In the panel data approach different economic developments in different regions are compared to changes in mortality rates in each region, which makes it possible to control for heterogeneity between regions. Hence, the effect of business cycles on the mortality rate is measured by exploring within-unit changes compared to other within-unit changes.

The pro-cyclical effect has been shown on US data by Ruhm (2000), Ruhm (2003), Ruhm (2005), Ruhm (2006), German regional data by Neumayer (2004), Spanish regional data by Tapia Granados (2005), and consistent pro-cyclical results has also been found on data consisting of 23 OECD countries between 1960 and 1997 (Gerdtham & Ruhm, 2006). As an example, Ruhm (2006) shows, focusing only on AMI mortality, that a one percentage point decrease in the unemployment rate is predicted to increase the AMI mortality rate by 1.3%. And he also shows that the risk is largest for those in prime working age (20–44-year olds), where a one percentage point decrease in the unemployment rate increases AMI mortality by 2.37%. One recent paper has reached an opposite conclusion. Gerdtham and Johannesson (2005) explore business cycle effects on the annual probability of individual level mortality in Sweden by using six different proxies for business cycles. They find no effect for women. For men, they find a counter-cyclical relationship for four business cycle indicators on mortality. Regarding cardiovascular mortality they find a significant counter-cyclical relationship for three out of six business cycle indicators for men. However, their results are based on individual level data where the business cycle measures are national in scope, and not at the regional/state level, and the time period studied is different compared to Ruhm (2000), Ruhm (2006) and Neumayer (2004) as well. This makes it hard to directly compare these contrasting results.

There are good arguments both for the case that mortality could increase, as well as decrease, in short-term economic upturns. In economic upturns, negative health effects could arise due to longer working hours which could make it more costly for individuals to invest in health enhancing exercises as well as make it more costly to eat healthier food which might be more time-intensive in preparation (Chou, Grossman, & Saffer, 2004). Ruhm (2005) and Ruhm and Black (2002) have also shown that behaviour such as binge drinking and smoking increase during economic upturns. There is also an indication that economic upturns and the longer working hours which might result from this leads to lower amount of sleep (Biddle & Hamermesh, 1990), which in turn is linked to higher levels of stress and increases in psychological illness (Sparks, Cooper, Fried, & Shirom, 1997). Further, health is in some sense an input in the production function on the labour market. Hence, during short-term economic upturns job-related stress might increase and have negative health effects and increase the risk of heart attacks for individuals with more fragile health status (Pickering, Clemow, Davidson, & Gerin, 2003).

On the other hand, positive health effects of short-term economic upturns and specifically decreases in AMI mortality risk can be due to less psychosocial stress because of reduced risk of unemployment. This given that psychosocial stress is a known risk factor for general cardiovascular diseases (Black & Garbutt, 2002; Brenner & Mooney, 1983; Elisaf, 2001; Fenwick & Tausig, 1994; Krantz & McCeney, 2002). In contrast to the results mentioned above by Ruhm (2005) and Ruhm and Black (2002), Dee (2001) showed using a large individual level dataset over a period of 11 years in the US that binge drinking increases in recessions. Further, in a paper by Novo, Hammarström, and Janlert (2001) it was found that young persons report higher level of somatic and psychological symptoms in recessions. Also, individuals tend to display less life satisfaction and increased unhappiness if becoming unemployed (Clark & Oswald, 1994; Kahneman & Krueger, 2006). Hence, it is obvious that there are contrasting effects, and it becomes an empirical issue to determine which effect is the larger one.

The main purpose of this paper is to use panel data techniques on aggregate regional data in Sweden. More specifically, the empirical analysis devotes interest to how business cycles affect incidence, mortality and lethality from AMI using data on 21 Swedish regions between 1987 and 2003. It is of particular interest to carry out such an analysis using the methods of e.g. Ruhm (2000), Ruhm (2006) and Neumayer (2004) on Swedish data, given that the only recent paper not finding a pro-cyclical relationship between the business cycle and mortality was based on Swedish data (Gerdtham & Johannesson, 2005). But as already mentioned that paper used a different method.

There are good arguments for restricting the study to explore the association between business cycles and AMI mortality. First of all, the results in the literature are primarily driven by the association between AMI mortality and the business cycle (Ruhm, 2006). And between 1987 and 2003 AMI contributes to between 20% and 30% of all deaths in Sweden and the other biggest source of death, cancer, is not dependent on short-term economic changes (Ruhm, 2000). Further, AMI mortality is measured with good precision (Socialstyrelsen, 2000) and the determinants of heart attacks are well known and they are certainly affected by macroeconomic and environmental conditions (Ruhm, 2006). As an example of how AMI is affected by short-term environmental/economic factors, there are also weekdays effects on AMI mortality, where there is a significant increase during Mondays (Willich, Lowel, Lewis, Arntz, & Keil, 1994). New in this paper compared to previous literature is also the more detailed data on AMI, where I also estimate the business cycle effects on incidence and lethality. Other than business cycle effects, the paper also tests the relationship between the share of men, foreigners and highly educated individuals on AMI incidence, mortality and lethality.

The paper is structured as follows. In the second part, the empirical method and data sources are presented. The results presented, using the model which gives the best fit, in the third part indicate that the business cycle effect is insignificant for total AMI mortality but show a counter-cyclical and significant effect on the mortality among individuals aged 20–49. The results implies that the near consensus view in the literature that short-term economic upturns are bad for your health is not universal and cannot be taken for granted to be true in different countries or time periods. A very crude estimate of the economic impact of a one percentage point decrease in the unemployment rate based on the results from the best fitting model shows that the extra economic benefit from reduced deaths among those aged 20–49 is 1,800,000–1,900,000 US dollars using the value of one life year and 49,500,000–52,250,000 US dollars using the value of a statistical life. Other than the business cycle proxies, it is found that an increasing share of men in each region is associated with, as expected, higher incidence and mortality. Also lethality, the share of fatal outcomes from an incidence, increases when the share of men increases. A higher share of foreign born individuals in each region is also associated with higher incidence and mortality. And a higher share of highly educated is associated with a lower incidence and mortality.

Section snippets

Data and method

In estimating the short-term impact of business cycles on AMI mortality, incidence and lethality a sample of 21 Swedish regions, which also are independent providers of health care in each region, is used. Data covers these 21 regions between 1987 and 2003. Hence, a balanced sample consisting of 21 cross-section units followed for 17 years, giving a total number of 357 observations. Three broad different dependent variables will be used in the paper. The first is the incidence rate of AMI that

Results

Table 2 below shows the results for the dependent variables incidence and mortality. To control for heteroskedasticity all standard errors are robust standard errors as estimated in Stata (StataCorp, 2001).

Model 1 is the most basic approach, not controlling for time effects nor fixed effects. The coefficient of the unemployment rate is negative, but this rather reflects a general trend of increasing unemployment that has taken place in Sweden and decreasing incidence and mortality due to better

Discussion and conclusions

The consensus that is beginning to emerge in the literature using panel data approaches that economic upturns increases cardiovascular and AMI mortality (pro-cyclical relationship) is not consistent with the results presented here. On the contrary, the results here on Swedish data reveal that short-term economic upturns has an insignificant overall effect on AMI mortality using the models with best fit. The business cycle effect is however significant for individuals aged 20–49 using the

References (35)

  • J.E. Biddle et al.

    Sleep and the allocation of time

    Journal of Political Economy

    (1990)
  • H.M. Brenner et al.

    Unemployment and health in the context of economic change

    Social Science & Medicine

    (1983)
  • P. Böckerman et al.

    Does a slump really make you thinner? Finnish micro-level evidence 1978–2002

    Health Economics Letters

    (2006)
  • A.E. Clark et al.

    Unhappiness and Unemployment Economic Journal

    (1994)
  • T. Dee

    Alcohol abuse and economic conditions: Evidence from repeated cross-sections of individual-level data

    Health Economics

    (2001)
  • M. Elisaf

    The treatment of coronary heart disease: an update—Part 1: An overview of the risk factors for cardiovascular disease

    Current Medical Research and Opinion

    (2001)
  • R. Fenwick et al.

    The Macroeconomic Context of Job Stress Journal of Health and Social Behavior

    (1994)
  • Cited by (67)

    • The effect of the economic collapse in Iceland on the probability of cardiovascular events

      2020, Economics and Human Biology
      Citation Excerpt :

      Although dating further back, the literature on business cycles and health received renewed impetus when Ruhm (2000) found that mortality increases in good economic times. Other studies followed, using different data, methods, and health outcomes – most with results consistent with those of Ruhm’s (Ariizumi and Schirle, 2012; Gerdtham and Ruhm, 2006; Tapia Granados, 2005; Tapia Granados and Ionides, 2017; van den Berg et al., 2017), but others with conflicting findings of a counter-cyclical relationship or no relationship (Economou et al., 2008; Gerdtham and Johannesson, 2005; Stuckler et al., 2009; Svensson, 2007, 2010). There is evidence that the pro-cyclicality of mortality has fallen in recent years, at least in the United States (Ruhm, 2015), and limited recent analysis raises the possibility that economic crises have effects that differ in relative size or direction from those of smaller shocks (Ruhm, 2016).

    View all citing articles on Scopus
    View full text