Endnotes

¹In Korea, the social security balance has been in surplus because the national pension system, which requires a minimum contribution period of twenty years for eligibility, has been accumulating surpluses since it was first introduced in 1988. The balance will, however, deteriorate quickly in the coming years as participants in the national pension system start to retire and become recipients of pension benefits. In fact, at the current contribution rate and the replacement rate, the pension system is expected to go bankrupt within a few decades. Hence, it may be misleading to include the social security account in the overall government budget.

²Of course, other countries may have similar problems: under a more comprehensive definition of the government, the fiscal debt to GDP ratio may substantially increase in other countries as well. However, many people including OK (2007) seem to believe that the size of quasi-fiscal accounts omitted from the official data is particularly large in Korea.

³Korea is not the only country that will witness a sharp increase in the fiscal debt/GDP ratio. In fact, according to IMF (2009b), most OECD economies are expected to see a similar jump in 2009 and 2010.

⁴ASEAN includes Brunei Darussalam, Cambodia, Indonesia, Lao People's Democratic Republic (Lao PDR), Malaysia, Myanmar, Philippines, Singapore, Thailand, and Viet Nam.

⁵Also, they changed the accounting framework for government finance from cash basis to accrual basis around 2004. The switch from cash-basis accounting to accrual-basis accounting was recommended by the Government Finance Manual 2001 of the International Monetary Fund (IMF). The newly constructed government finance data are available only for 1990 and afterward, while the old series are available only up to 2001. The two series are not consistent with each other and thus cannot be combined.

⁶It is clear that we do not need a separate censoring rule for annual series, because a complete cycle (from peak to peak) will always take at least 2 years, which is greater than 5 quarters.

⁷A smoothing parameter of 6.25 is used for an annual series and a parameter of 1600 is used for a quarterly series.

⁸To reduce endogeneity bias, we use instrumental variables estimation for the government consumption equation, with one year lagged value of GDP gap used as the instrument.

⁹The average growth rate during the first year of recovery is not much influenced by the 1979-1981 recession of Brunei Darussalam.

¹⁰I identify and exclude outlier observations before computing the policy measures reported in Table 2. Outliers in each variable are identified using the method developed by Hadi (1994). Stata provides the routine for this procedure.

¹¹Although not reported in the table, we have used quarterly data for OECD economies and found that both fiscal and monetary policies turn expansionary during a recession in OECD.

¹²IMF (2009a) reports the opposite: they use the peak-to-trough amplitude without any transformation and still obtain a positive coefficient. It is not clear where the discrepancy between IMF (2009a) and our result stems from. We have examined quarterly data for OECD countries following IMF (2009a) and still found that a greater fall is associated with a faster recovery.

¹³Similar results are obtained when we first regress the fiscal balance to GDP ratio on logarithm of GDP (detrended using Hodrick-Prescott filtering) and then take differences of the residual. We have also examined residuals from regressing the fiscal balance/GDP ratio on GDP growth and found that the residual is lower in 2009 than in 1998.

¹⁴Our measure of fiscal policy is accumulated changes in government consumption over the recession years, from the peak year to the trough year. Since the duration of a recession can be longer than 1 year, the coefficient on fiscal policy in Table 3 may partly reflect lagged responses in GDP growth to a fiscal spending increase.

¹⁵As mentioned above, we use consumption expenditure, not total expenditure, of the government because of data availability. Government consumption expenditure does not include transfer payments. Transfer payments may have a smaller multiplier effect than government consumption expenditure, because recipients of transfer payments may not increase their spending by the same amount.

Download this Paper [ PDF 354.1KB| 26 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.