Fiscal stimulus effectiveness in Japan: evidence from recent policies

ABSTRACT

This article examines the effects of Japanese fiscal policy after the 2008 global financial crisis using a mixed vector autoregression/event study approach. We focus on the effects of stimulus packages with environmental benefits. The empirical results show that a tax break and subsidy programme designed to promote the adoption of eco-friendly cars helped stimulate automobile production, while a similar programme intended to promote the purchase of energy-efficient appliances had no effect on appliance production.

KEYWORDS:

• Fiscal policy effectiveness in Japan
• environmental stimulus package
• eco subsidy
• a mixed vector autoregression/event study approach

JEL CLASSIFICATION:

• E23
• E62
• H30

Acknowledgements

The author would like to thank two anonymous referees, Prateek Goorha, Shigeyuki Hamori, Yuji Kawano, Akihiro Kawase, Etsuro Shioji, Kenji Takeuchi, and Masao Tsuri for their insightful comments and suggestions. The delegates of 12th international conference of the Japan Economic Policy Association, the 51th annual meeting of the Public Choice Society, and the 14th meeting of Association of Public Economic Theory are also acknowledged.

Disclosure statement

No potential conflict of interest was reported by the author.

Supplementary

Supplemental data for this article can be accessed here.

Notes

¹ The application to eco-car subsidy was ended on 21 September 2012 because it had reached its upper budgetary limit. For details, please see http://www.cev-pc.or.jp/ECO/.

² Although the eco-car tax break was initially planned to be end in February 2011, it was extended until March 2011.

³ For information on the fiscal policy response after the global financial crisis in Japan, please see Iwaisako (2010) and Asako (2012). For the details of the eco-car subsidy programme, please see Alhulail and Takeuchi (2014).

⁴ We also estimate our VAR model using the index of industrial production, which denotes the production of all mining and manufacturing industries, instead of $y_t^{El}$ or $y_t^{Auto}$. However, the impulse response functions are found to be insignificant.

⁵ This follows the arguments in Blanchard and Perotti (2002) and Miyazaki (2009).

⁶ Two fiscal expansions, eco-car subsidy programme and eco-car tax break, affect current automobile industry production. However, it is plausible that there is no feedback from the eco-car tax breaks to the eco-car subsidy programme. We thus assume the following relationship between structural form residuals, ${\epsilon }_t$, and reduced form residuals, $u_t$ ($A{u}_t={\epsilon }_t$), and implement the over-identification restriction test:

$\left(\begin{array}{l}100\\ 010\\ -a_{31}-a_{32}1\end{array}\right)\left(\begin{array}{l}{u}_t^{SUB}\\ u_t^{TAX}\\ u_t^{y^{Auto}}\end{array}\right)=\left(\begin{array}{l}{\epsilon }_t^{SUB}\\ {\epsilon }_t^{TAX}\\ {\epsilon }_t^{y^{Auto}}\end{array}\right).$

⁷ To capture the effects of each policy, we would ideally focus on production of the items mostly affected by the policies, such as gas-electric hybrid cars, air conditioners, and refrigerators. However, METI data are not available at such a detailed level.

⁸ The eco-points programme also targeted LCD televisions capable of receiving digital terrestrial broadcast signals, which are included in the ‘electric machines for home use’ data category (minsei-yo denshi kikai in Japanese). LCD televisions are the only items in this category targeted by the policy, whereas $y_t^{El}$ covers a broader set of target items. Further, the Japanese government encouraged households to purchase this type of television leading up to July 2011, when all of Japan’s TV programming was switched from analogue broadcasting to digital terrestrial broadcasting. If we examine the effects on electric machines for home use, it is clear that we cannot disentangle the policy effects of the eco-points programme from the increase in the production of LCD televisions before this transition. As such, we use $y_t^{El}$.

⁹ For details, please see: http://www.meti.go.jp/statistics/tyo/iip/pdf/b2010_mechanism_iipj.pdf.

¹⁰ Another option would be to end the sample period in February 2011 in order to eliminate the influence of the Great East Japan Earthquake. When this is done, however, the impulse response functions are not estimated to be significant.

¹¹ Other methods, such as Monte Carlo integration and bootstrap replications, could also be considered to calculate the confidence intervals of the impulse response functions. However, since there is no established method for calculation, we report the results based on asymptotic distributions. For confirmation, we also calculate the confidence intervals with 500 bootstrap replications, but the results and implications are not fundamentally changed.

¹² For more detail on the fuel-efficient and low-emissions cars covered by the eco-car tax break in our sample periods, please see http://www.jama.or.jp/tax/exemption201205/tax_list_old.html.

¹³ There are no monthly datasets that cover only products that benefited from the eco-points programme. Therefore, we have no choice but to use the production indicator as an approximation when estimating the macroeconomic effects of the programme.

¹⁴ The lag lengths used for the additional tests in this subsection are also chosen using the likelihood ratio test of Sims (1980).

Additional information

Funding

This work was supported by a Grant-in-Aid for Scientific Research (Young Scientist) and financial aid from the Zengin Foundation for Studies on Economics and Finance.