Former time-series analyses on the influence of causal relationship between economic growth and military spending may encounter the potential accumulation and neglected variable deviates from accumulating high frequency data generally employed to evaluate business cycles, e.g., monthly macroeconomic and quarterly gross domestic product per capita data, into low frequency data, e.g., yearly military spending data. The mixed frequency vector auto-regressive (MF-VAR) model for solving such temporal aggregation problem in this study, is firstly applied to assess the causal link between economic growth and military spending from the first quarter (Q1) of 1975 to the fourth quarter (Q4) of 2017 in Taiwan. These results via the prediction error variance shows that the MF-VAR model reaches higher explanatory power than the traditional VAR model belonging to single frequency data. Although not only counter-cyclicality but also pro-cyclicality of military spending on economic growth shock in Taiwan is found for the feedback hypothesis via the mixed frequency Granger causality tests, a varying correlation between military spending and economic growth results from the impulse-response analyses via the MF-VAR model. The above-mentioned findings significantly meet the arguable results with respect to former time-series analyses on the business-cycle effect on military spending.