TY - JOUR
T1 - Health progress and economic growth in the United States
T2 - the mixed frequency VAR analyses
AU - Liu, Yi Hui
AU - Chang, Wei Shiun
AU - Chen, Wen Yi
N1 - Funding Information:
We would like to thank Professor Kaiji Motegi (from Kobe University, Japan) sharing his Matlab codes for us to estimate the LF-VAR and MF-VAR models. The final proof-reading of the study by Lisa Brutcher (at Washington State University, USA) is deeply acknowledged. All errors are mine.
Publisher Copyright:
© 2019, Springer Nature B.V.
PY - 2019/7/15
Y1 - 2019/7/15
N2 - Previous time series analyses on the effect of business cycles on population health suffered from potential aggregation and omitted variable biases stemming from the aggregation of the high frequency data commonly used to measure business cycles (such as quarterly GDP per capita or monthly unemployment rate) into low frequency data (such as annual GDP per capita or unemployment rate). In order to deal with this temporal aggregation issue, this study applies the Mixed Frequency Vector Auto-regressive (MF-VAR) model to investigate the causal relationship between health progress and economic growth during the period of 1948:Q1–2016:Q4 in the US for the first time. Our results based on the forecast error variance decomposition suggest that the MF-VAR model achieves higher explanatory power than the conventional VAR model with single frequency data. Despite a bi-directional causation between health progress and economic growth being identified using the mixed frequency Granger causality tests, the impulse-response analyses under the MF-VAR model found a changing correlation between health progress and economic growth across the four quarters within each 1-year time period. These findings effectively reconcile the controversial results from previous time series analyses on the effect of business cycles on population health.
AB - Previous time series analyses on the effect of business cycles on population health suffered from potential aggregation and omitted variable biases stemming from the aggregation of the high frequency data commonly used to measure business cycles (such as quarterly GDP per capita or monthly unemployment rate) into low frequency data (such as annual GDP per capita or unemployment rate). In order to deal with this temporal aggregation issue, this study applies the Mixed Frequency Vector Auto-regressive (MF-VAR) model to investigate the causal relationship between health progress and economic growth during the period of 1948:Q1–2016:Q4 in the US for the first time. Our results based on the forecast error variance decomposition suggest that the MF-VAR model achieves higher explanatory power than the conventional VAR model with single frequency data. Despite a bi-directional causation between health progress and economic growth being identified using the mixed frequency Granger causality tests, the impulse-response analyses under the MF-VAR model found a changing correlation between health progress and economic growth across the four quarters within each 1-year time period. These findings effectively reconcile the controversial results from previous time series analyses on the effect of business cycles on population health.
UR - http://www.scopus.com/inward/record.url?scp=85062714630&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062714630&partnerID=8YFLogxK
U2 - 10.1007/s11135-019-00847-z
DO - 10.1007/s11135-019-00847-z
M3 - Article
AN - SCOPUS:85062714630
SN - 0033-5177
VL - 53
SP - 1895
EP - 1911
JO - Quality and Quantity
JF - Quality and Quantity
IS - 4
ER -