Abstract
This paper investigates the causal effect of improvements in health on economic development using a long panel of European countries. Identification is based on the particular timing of the introduction of public health care systems in different countries, which is the random outcome of a political process. We document that the introduction of public health care systems had a significant immediate effect on health dynamics proxied by infant mortality and crude death rates. The findings suggest that health improvements had a positive effect on growth in income per capita and aggregate income.
Similar content being viewed by others
Notes
For the UK, we use own calculations of the mortality rates as weighted average of the mortality rates from England, Wales, Scotland, and North Ireland.
However, the main results do not change when we include these countries.
Missing observations are completed using data from Mitchell (1992) and from OECD Annual National Accounts (Volume 2).
In earlier specifications, we also used a balanced data set for which the number of observations is much smaller. We also used a data set which includes only the observations one period before and after the introduction of universal public health care. The main results are qualitatively and quantitatively almost identical.
The growth rate of variable x i,t is calculated as
$$ \Delta x_{i,t} = \frac{x_{i,t}- x_{i,t-1}}{x_{i,t-1}}, $$where i denotes the country and t the time dimension. The results for log differences are qualitatively equal and quantitatively larger, but the calculation in log differences would potentially overestimate the true values as it delivers negative growth rates exceeding −100 %. The results also do not change when we use average yearly growth rates.
There are some observations of very high growth rates, in particular for Austria, which exhibits growth of up to 283 % during one 20-year window (almost 7 % per year). In robustness checks, we excluded Austria. The results remain robust and qualitatively unchanged.
A historical review of public health insurance for all countries in the sample can be found in Online Appendix A.
Elite democracies are political systems with property requirements for franchise, which exclude large parts of the population from the voting process.
We observe the coverage of public health care for most countries in our data. However, we think that the coverage could have a direct effect on economic development and would therefore invalidate the exclusion restriction. For example, it might have been easier for rich countries to introduce a health insurance system that covers the entire population than for poor countries.
Table 2 contains a list of all countries.
In addition, Finland was immediately affected by the Cold War due to its common border with Russia, while Spain remained under the authoritarian Franco regime after the war.
Only periods which are observed in the unbalanced panel are considered in the following.
Deterministic time patterns, such as trends or period dummies, are coded in synthetic time t but exploit variation in calendar time τ. For example, period dummies are coded as 1 if the respective period t falls in the corresponding decade of calendar time, and zero otherwise. Likewise, period trends are coded according to the respective calendar time period that corresponds to the particular t of a given country–year observation.
Since we condition on X 1, the identifying assumptions are also valid for the interaction between the instrument z and X 1, if z is exogenous conditional on X 1 and X 2 (see e.g., Angrist and Pischke 2008).
Since the instrument is a binary outcome variable, we use a nonlinear Logit specification.
Following Chesnais (1992), we define a country to be post-transitional when the crude death rate is lower than a certain threshold,
$$ \textrm{PostTrans}_{i,t-1} = \left\{ \begin{array}{ll} 1 & \mbox{when } \textrm{cdr}_{i,t-1} \leq q_{50}(\textrm{cdr}), \\ 0 & \mbox{when } \textrm{cdr}_{i,t-1} > q_{50}(\textrm{cdr}), \\ \end{array} \right. $$where \(q_{50}(\textrm{cdr})\) is the median of the crude death rate over all countries and time periods.
The period fixed effects take value 1 if the respective observation falls into a particular period of calender time, and zero otherwise. The periods are 1850–1920, 1921–1935, 1936–1950, 1951–1965, and 1966–2010. If, for example, growth rates for the Netherlands are observed from 1921 to 1941, for this observation, the period dummy for the time period 1936–1950 is equal to one, and all other period dummies are zero.
In unreported specifications, we also included a quadratic year trend and war dummies. Since the general results do not change, we dropped these variables from the final specifications.
If faster economic growth is associated with worse living conditions at least for part of the population, due to, e.g., urbanization, harsh working conditions in factories, or intensified trade and the faster spread of diseases, this might lead to higher infant mortality and crude death rates. In this case, OLS estimates would underestimate the causal effect. Likewise, the causal effect would be smaller in absolute terms if higher incomes lead to increased fertility with the consequence of higher child mortality. Moreover, the observed changes in infant mortality and crude death rates might be measured with noise, implying attenuation bias. However, in principle, the bias could also go the other way, since faster economic growth might be associated with lower mortality through third factors, such as the absence of epidemics.
Instead, multicollinearity can even help because it increases the predictive power of the model. As sensitivity analysis for the first stage, we applied an outlier analysis. We could not find any outlier with a strong impact on the estimates. The results of the outlier analysis are available upon request.
This would be expected from standard economic models of fertility in which parents derive utility from their surviving offspring and can compensate for child mortality by larger gross fertility and sequential fertility choice, see, e.g., Doepke (2004).
The results are essentially unchanged when adding lags of Δm as additional controls. The respective coefficients are not significantly different from zero.
See Tables B.1 and B.2 in Online Appendix B.
The first stage F-statistic takes often low values in the robustness checks. The reason is that, in many instances, we do not observe the additional control variables in the period after the introduction of a universal public health care system. Therefore, we do not only have a lower number of observations in the sample, but even less observations in the period after the introduction. In light of this, we find it worth noting that the F-statistic is in all specifications still significant at the 5 %-level and the Shea’s R 2 is considerably high.
We calculate a heteroscedasticity robust DWH statistic following Baum and Schaffer (2003).
A delayed placebo makes less sense since the placebo might pick up delayed treatment effects as the spread of health system coverage. Similarly, indirect effects like demographic change and education might become active with a delay.
References
Abbring J, Van den Berg G (2003) The nonparametric identification of treatment effects in duration models. Econometrica 71(5):1491–1517
Acemoglu D, Johnson S (2007) Disease and development: the effect of life expectancy on economic growth. J Polit Econ 115(6):925–985
Acemoglu D, Johnson S (2009) Disease and development: a reply to Bloom, Canning and Fink. Working paper
Acemoglu D, Johnson S, Robinson R (2005) Institutions as the fundamental cause of long-run growth. In: Aghion P, Durlauf S (eds) Handbook of economic growth, vol 1. Elsevier, pp 385–472
Aisa R, Pueyo F, Sanso M (2012) Life expectancy and labor supply of the elderly. J Popul Econ 25(2):545–568
Angeles L (2010) Demographic transitions: analyzing the effects of mortality on fertility. J Popul Econ 23:99–120
Angrist J, Krueger A (1991) Does compulsory schooling attendance affect schooling and earnings. Q J Econ 106:979–1014
Angrist J, Pischke JS (2008) Mostly harmless econometrics. Princeton University Press, Princeton
Barro R (1991) Economic growth in a cross section of countries. Q J Econ 106(2):407–443
Baum C, Schaffer M (2003) Instrumental variables and GMM: estimation and testing. Stata J 3(1):1–31
Bekker P (1994) Alternative approximations to the distributions of instrumental variables estimators. Econometrica 62(3):657–681
Bloom D, Canning D, Fink G (2009) Disease and development revisited. NBER working paper, no 15137
Bound J, Jaeger A, Baker R (1996) Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variable is weak. J Am Stat Assoc 90:443–450
Cervellati M, Sunde U (2011) The effect of life expectancy on economic growth reconsidered. J Econ Growth 16(2):99–133
Chamberlain G, Imbens G (2004) Random effects estimators with many instrumental variables. Econometrica 72(1):295–306
Chesnais J (1992) The demographic transition: stages, patterns and economic implications. A longitudinal study of sixty-seven countries covering the period 1720–1984. Clarendon Press, Oxford
Corens D (2007) Health system review: Belgium. Health Syst Transit 9(2):1–172
Cutler D, Deaton A, Lleras-Muney A (2006) The determinants of mortality. J Econ Perspect 20:97–120
Doepke M (2004) Accounting for fertility decline during the transition to growth. J Econ Growth 9(3):347–383
Durlauf S, Johnson P, Temple J (2005) Growth econometrics. In: Aghion P, Durlauf S (eds) Handbook of economic growth, vol 1. Elsevier, pp 555–677
Durán A, Lara J, Van Waveren M (2006) Health system review: Spain. Health Syst Transit 8(4):1–208
Flora P (1983) State, economy, and society in Western Europe, 1815–1975, I. St. James, Chicago
Flora P, Kraus F, Pfenning W (1987) State, economy, and society in Western Europe, 1815–1975, II. St. James, Chicago
Flores-Lagunes A (2007) Finite sample evidence of IV estimators under weak instruments. J Appl Econom 22(3):677–694
Fort M, Schneeweis N, Winter-Ebmer R (2011) More schooling, more children: compulsory schooling reforms and fertility in Europe. IZA discussion paper, vol 6015
Garrouste C (2010) 100 years of educational reforms in Europe: a contextual database. European Commission Joint Research Center, Publications Office of the European Union, Luxembourg
Glenngård A, Hjalte F, Svensson M, Anell A, Bankauskaite V (2005) Health system review: Sweden. Health Syst Transit 7(4):1–128
Gurr TR (1974) Persistence and change in political systems, 1800–1971. Am Polit Sci Rev 68(4):1482–1504
Hock H, Weil DN (2012) On the dynamics of the age structure, dependency, and consumption. J Popul Econ 25(3):1019–1043
Hofmarcher M, Rack R (2001) Health system review: Austria. Health Syst Transit 3(5):1–120
Johnsen J (2006) Norway. Health Syst Transit 8(1):1–167
Kunnskapsdepartementet (2007) Education—from kindergarten to adult education. Royal Norwegian Ministry of Education and Research, Norway
Lindert P (2004) Growing public. Cambridge University Press
Lo Scalzo A, Donatini A, Orzella L, Cicchetti A, Profili S, Maresso A (2009) Italy. Health systems review. Health Syst Transit 11(6):1–216
Lorentzen P, McMillan J, Wacziarg R (2008) Death and development. J Econ Growth 13:81–124
Mackenbach J (1996) The contribution of medical care to mortality decline: McKeown revisted. J Clin Epidemiol 49:1207–1213
Maddison A (2006) The world economy, vol 1: a millenial perspective. OECD, Paris
Minder A, Schoenholzer H, Amiet M (2000) Health care systems in transition—Switzerland. European Observatory on Health Systems and Policies, pp 1–82
Mitchell B (1992) European historical statistics, 1750–1988, 3rd edn. Stockston, New York
Persson T, Tabellini G (2003) The economic effects of constitutions. MIT, Cambridge
Robinson R, Dixon A (1999) Health care systems in transition: United Kingdom. European Observatory on Health Care Systems, Copenhagen, pp 1–117
Sandier S, Paris V, Polton D (2004) Health system review: France. Health Syst Transit 6(2):1–145
Schäfer W, Kroneman M, Boerma W, Van den Berg M, Westert G, Devillé W, van Ginneken E (2010) Health system review: the Netherlands. Health Syst Transit 12(1):1–229
Strandberg-Larsen M, Nielsen M, Vallgårda S, Krasnik A, Vrangbæk K (2007) Denmark: health system review. Health Syst Transit 9(6):1–164
Vuorenkoski L (2008) Health system review: Finland. Health Syst Transit 10(4):1–168
Wooldridge J (2010) Econometric analysis of cross section and panel data, 2nd edn. MIT, Cambridge
Author information
Authors and Affiliations
Corresponding author
Additional information
Responsible editor: Junsen Zhang
Electronic Supplementary Material
Rights and permissions
About this article
Cite this article
Strittmatter, A., Sunde, U. Health and economic development—evidence from the introduction of public health care. J Popul Econ 26, 1549–1584 (2013). https://doi.org/10.1007/s00148-012-0450-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00148-012-0450-8