Vortragssitzung

Inequality and Health

Talks

Health shocks and political instability – The 1918 Spanish flu pandemic and the collapse of Weimar Germany
Felix Rösel, ifo Institute

Abstract

The Spanish flu killed some 240,000 Germans in autumn 1918. The flu hit Germany in the most unfortunate moment. Millions of soldiers returned from World War I, and hunger and revolutions swept the country. Farm workers fall ill with the flu, exacerbating the much-needed harvest. Germany started with a toxic mix of high flu death tolls and malnutrition into the new era of democracy in 1918. We investigate to which extent the health shock from the 1918 Spanish flu contributed to political instability in Germany’s young democracy after World War One.

Method

We hand-collect and digitize new local data on mortality, crop yields, and election outcomes for all around 210 Bavarian districts in Weimar Germany over the period 1912 to 1933. We estimate local excess mortality in 1918 for all Bavarian districts and link it to local Reichstag election outcomes before and after World War One in difference-in-differences estimations. Our main dependent variable of interest is the vote share for the largest government party SPD which citizens are most likely to blame for mismanagement of the 1918 pandemic.

Results

We find that local excess mortality caused by the 1918 Spanish flu vary substantially across Bavarian districts. Difference-in-differences results show that vote shares for the largest government party SPD decrease in districts where more citizens died from the Spanish flu. We find negative effects of 1918 excess mortality on SPD vote shares in all national elections between 1919 and 1933. By contrast, vote shares for Conservative parties increase. We find no effects on voter turnout. Preliminary evidence suggests that agricultural labor markes play an important role in mediating flu effects and political outcomes.

Conclusion

The 1918 wave of the Spanish flu shocked Germany’s young democracy. We find that regions that were severely hit by the flu turned away from moderate government parties in the very first election in January 1919. Trust could never be restored afterwards. We conclude that there is a direct link between pandemics and political radicalization in unstable democracies.


Authors
Mona Förtsch, ifo Institute
Felix Rösel, ifo Institute
Health-Income Mobility in Germany
Laura Nübler

Abstract

Income-related health mobility and health-related income mobility are measures that exploit additional information in longitudinal data to investigate dynamics in income-related health inequalities. They allow us to distinguish systemic inequalities from transitory ones due to poverty and illness. As life-course developments like work force entry, career progression and retirement may systematically determine the observed health-income dynamics, we suggest age-specific mobility estimators to control for such confounding effects. In our empirical application, we investigate the differences in socioeconomic dynamics between mental and physical health in Germany.

Method

We use a varying concentration index approach based on a non-parametric smoothing technique to derive two age-specific mobility estimators. Income-related health mobility reveals whether changes in health systematically favor the rich or poor, health-related income mobility indicates whether changes in income favor those in good or bad health. We use self-reported measures of mental and physical health-related quality of life from the SF-12, as well as clinical measurements of grip strength and diagnosis of depressive disorders as proxies for physical and mental health. We use longitudinal data from the German Socio-Economic Panel for the years 2004-2016.

Results

For both mental and physical health, preliminary results suggest high mobility among younger and working-age individuals, but little mobility among the retired and elderly, where existing health inequalities mostly persist. For physical health, both mobility estimators show a disequalizing effect that systemically favors the richer and thereby widens income-related health inequalities during the working ages. Conversely, while good mental health does concentrate among the rich overall, changes in mental health were shown to favor the poor, especially for age groups <50 and >70. This results in a potentially equalizing dynamic in income-related mental health inequalities.

Conclusion

The low health-related income mobility observed among the retired, who receive a fixed pension, is unsurprising. However, the lack of systematic income-related health mobility for physical health in this age group lends support to the age-as-leveler hypothesis, which predicts a decline in health inequalities in old age. Furthermore, while physical health inequalities and mobility systematically favor the rich, mental health-related mobility does not. We speculate that success-related stressors among the young and loss of career-related identity among the retired may contribute to the relative decline in mental health among the richer. We then discuss resulting policy implications for the German health care system.


Authors
Laura Nübler
The effect of BMI and type 2 diabetes on socioeconomic status: a two-sample multivariable Mendelian randomization study
Sara Pedron, Helmholtz-Zentrum München

Abstract

The prevalence of obesity and type 2 diabetes is rising globally. Besides being closely biologically interconnected illnesses, both obesity and diabetes can have a substantial influence on socioeconomic status and achievements, leading to high indirect costs. However, estimating the causal impact of these conditions on socioeconomic outcomes is usually complicated by problems of endogeneity. With this study, we aim at assessing the independent causal effect of BMI and type 2 diabetes on socioeconomic outcomes applying novel two-sample multivariable Mendelian randomization (MR) techniques.

Method

We used 69 genome-wide significant single nucleotide polymorphisms (SNPs) for BMI and 42 SNPs for diabetes as instrumental variables. Their independent causal impact on socioeconomic outcomes (household income and regional deprivation) was assessed using summary-level data from the UK Biobank. First, we estimated univariate effects of each exposure on the outcomes and investigated the presence of bias due to overlapping signals. Second, we jointly assessed the effects of both exposures on the outcomes in a multivariate MR approach, explicitly considering overlapping SNPs in the computation of the direct effects.

Results

Available SNPs for BMI and diabetes overlapped at two distinct signals at the TCF7L2 and FTO loci. The univariate analysis indicated that higher BMI is related with lower household income (marginal effect of 1-SD increase in BMI (β)=-0.092 [95% CI -0.138, -0.047]) and higher regional deprivation (β=0.051 [95% CI 0.022, 0.079]). Results of the multivariate MR approach indicated that the effect of BMI controlling for diabetes is slightly lower than in the univariate analysis for both income (β=-0.089 [95% CI -0.13, -0.048]) and deprivation (β=0.049 [95% CI 0.023, 0.075]). Diabetes was neither in the univariate nor in the multivariate analyses significantly associated with the socioeconomic outcomes considered.

Conclusion

High BMI leads to lower socioeconomic status. The causal effect of diabetes on socio-economic outcomes should be object of further research in order to guide evidence-based intervention.


Authors
Sara Pedron, Helmholtz-Zentrum München
Christoph F. Kurz, LMU München
Lars Schwettmann, Helmholtz-Zentrum München
Michael Laxy, Helmholtz-Zentrum München
Messung von Finanzierungsgerechtigkeit und das Problem mit Konsumentenentscheidungen: Ein pragmatischer Ansatz
Martin Siegel, TU Berlin

Abstract

Ein verbreitetes Kriterium zur Bewertung der finanziellen Gerechtigkeit im Gesundheitswesen ist eine proportionale oder leicht progressive Verteilung der Ausgaben. Bei der Betrachtung direkter Gesundheitsausgaben wäre eine Trennung von medizinisch notwendigen und zusätzlichen Ausgaben wichtig, da Entscheidungen für zusätzliche Leistungen oder höherwertigere Hilfsmittel (z.B. Brillen, Hörgeräte) zu deutlich höheren Ausgaben unter Haushalten mit einer höheren Zahlungsfähigkeit führen können. Als Folge würden die Ausgaben weniger regressiv als bei einer reinen Betrachtung notwendiger Ausgaben erscheinen. Da diese Trennung mit Befragungs- oder Abrechnungsdaten kaum möglich ist, schlagen wir eine separate Betrachtung von Ausgabenkategorien je nach Abdeckungsgrad durch die Krankenversicherung und Spielraum für Konsumentenentscheidungen vor. Wir argumentieren, dass ein solcher Ansatz Hinweise auf mögliche Ungerechtigkeiten bei Ausgaben für medizinisch notwendige Basisversorgung liefern kann.

Method

Wir verwenden Daten der Einkommens- und Verbrauchsstichprobe aus den Jahren 2008, 2013 und 2018. Wir übernehmen die Untergliederung in 11 verschiedene Ausgabenarten (Medikamente mit/ohne Verordnung, Heilmittel mit/ohne Verordnung, Hilfsmittel, ambulante Versorgung, Zahnarztbehandlungen und Materialien, Ausgaben für Labore und stationäre Versorgung) für eine Bewertung des Ausmaßes des Versicherungsschutzes und möglicher Konsumentenentscheidungen anhand der Zuzahlungsregeln für GKV-Versicherte. Wir berechnen Kakwani-Indexe für die einzelnen Kategorien und für die gesamten direkten Gesundheitsausgaben anhand des Haushaltsnettoeinkommens und zeigen, welche Kategorien wie stark zum Verteilungseffekt der gesamten direkten Gesundheitsausgaben beitragen.

Results

Direkte Gesundheitsausgaben sind in allen Jahren signifikant regressiv, sodass ärmere Haushalte überproportional zahlen. Die einzelnen Kategorien zeigen ein differenziertes Bild: Ambulante Versorgung, wo notwendige und effektive Versorgung ohne weitere Zuzahlungen erfolgt, zeigt eine stark progressive Wirkung, während etwa zahnärztliche und stationäre Versorgung keinen signifikanten Verteilungseffekt zeigen. Deutlich regressive Wirkungen zeigen Medikamente sowie Heil-/Hilfsmittel, die den größten Teil der Regressivität direkter Gesundheitsausgaben erklären.

Conclusion

In Kategorien mit einem breiten Spektrum an Wahlmöglichkeiten und einer geringen Abdeckung lassen sich signifikant progressive Effekte beobachten, während sich starke regressive Verteilungswirkungen in Kategorien mit klar begrenzten Wahlmöglichkeiten zeigen. Unsere Ergebnisse lassen vermuten, dass das Zuzahlungssystem der deutschen GKV bei einer reinen Betrachtung der notwendigen Basisversorgung noch deutlich regressiver erscheinen würde.


Authors
Martin Siegel, TU Berlin
Jonathan Cylus, WHO Regional Office for Europe
Laura Nübler, TU Berlin