<レポート017E> PEP DISCUSSION PAPER 2022-1E PARK Seung-Joon, “Eggs first or chickens? A re-examination of Granger causality between increase in government spending and GDP growth”

この前に投稿いたしました、本会共同代表の朴勝俊関西学院大学教授による、デービッド・アトキンソン氏への批判論文「タマゴが先かニワトリが先か? : 政府支出と GDP のグレンジャー因果性に関する検討」を、著者本人が英訳しましたので、ここに掲載いたします。冒頭のAbstractを引用しておきます。

Abstract
Based on data from various countries, a strong correlation was noted between the growth rate of government expenditure and the growth rate of nominal and real GDP, but there was some debate about the direction of causality. In this paper, a simple theoretical simulation model was first constructed to show that scatter plots similar to those actually observed can be drawn in either direction we assume the causality. We then tested Granger causality between general government expenditure, nominal GDP and GDP deflators using data from 1980 to 2021 for 38 OECD countries, and found that the results differed significantly from country to country at different time periods, with many cases suggesting a causality from growth to government expansion. However, as government expenditure statistics such as in the SNA are produced on an accrual basis, the point in time when the amounts are recorded may be later than when the orders are placed, which means that ‘spurious causality’ may be observed.
Therefore, we tried using the lead variable (a variable of later period) of public spending to test for Granger causality, and found that the results could change. To examine this point in more detail, we used quarterly data from the Japanese GDP statistics (1994-2021), namely nominal GDP, government fixed capital formation and government consumption. We found that in Japan since 2008 there was (practically) no causal relationship in neither direction without lead variables. However, with lead variables of government expenditure, one-way Granger causality from public spending to nominal GDP was observed.

ダウンロード↓
economic policy report 017E