Do government spending multipliers depend on the level of government debt? US historical data evidence

Abstract

This article examines state-dependent effects of government debt on government spending multipliers. For the estimation, a new quarterly US historical debt series is constructed. The empirical findings reveal that effects of government spending on the economy as well as the estimated multipliers significantly differ by the level of debt.

Keywords:

• government spending multiplier
• government debt
• local projection
• state dependency

JEL Classification:

• C32
• E62
• H63

Acknowledgements

I am deeply indebted to Ryo Jinnai and Sarah Zubairy for their guidance and helpful comments. All errors are my own.

Notes

¹ Cashell (2010) claims that the debt held by public series is more reasonable to measure the burden of government debt than the gross debt series since all gross debt do not represent past borrowing in credit markets. However, the debt held by public series is not available for entire sample periods. Moreover, the debt limit in the US is determined based on gross debt, and Reinhart and Rogoff (2010) also use annual gross debt in their research. Thus, gross debt is a reasonable measure of the burden of government debt.

² For example, debt-to-GDP in 1942Q1 was 40%, but in 1943Q3 rose 80%. Furthermore, annual data are less relevant to measure multipliers since annual data are not free from the private agent’ anticipation even though a shock is identified by the defence news. See Ramey (2011).

³ The threshold level is not econometrically chosen, but I tested several levels and find that the results with 45% and above are different from the results with levels below 45%. Ilzetzki et al. (2013) use a similar strategy to identify the break point.

⁴ Owyang et al. (2013) point out that spending multipliers calculated by the way in the standard VAR can be biased.

⁵ Results are available upon request.