(Acknowledgements) The authors are grateful to Charles Yuji Horioka, Yoko Niimi, and participants of the PECC International Workshop on the Social Resilience Project (July 2011, Tokyo) for their helpful comments.

＊） Gakushuin University, Tokyo, Japan

＊＊） Japan Institute for Labour Policy and Training, Tokyo, Japan

1） The default rate on national pension premiums was less than 5% in the 1970s. Data source: Homepage of Pension Finance by Pension Bureau, Japan.

2） Under the Waterfront Strategy, some sorrowful incidents happened. In July 2007, for example, a former welfare recipient of Kitakyushu City was found dead from hunger, leaving a message that “I want to eat rice ball” in their diary. Two other similar deaths were found in Kitakyushu City in January 2005 and May 2006.

3） Description of the Blanchard-Quah decomposition method is from Suzuki and Zhou (2007).

4） Due to the economic growth and the accumulation of national wealth, the public assistance rate has a downward trend in the long run. The unemployment rate, on the other hand, has an upward trend in the long run because mismatch unemployment has increased substantially along with radical technology innovation and industry structure transitions.

5） Public assistance rate=12.12865−0.0043507*trend+3.533062*Oil shock dummy
　　　　(38.43)　(−5.66)　　　　　(11.55)　　　　　 　　　Adj R-squared= 0.5149
　　Unemployment= 0.3017669+0.0074976*trend+0.340172* Oil shock dummy
　　　　　　　　(3.60)　　　　　(36.79)　　　　(4.19)　　Adj R-squared= 0.8192
Notes: (1) t-values are in parentheses. (2) Oil shock dummy equals 1 if before December 1973; 0 otherwise.

6） Estimation result by using 24-month lags is basically the same with the case of using 12-month lags.

7） Suzuki and Zhou (2007) estimate that a much longer period (105 months) is needed for temporary shock to converge to zero. Since our approaches are the same, the gap is likely be formed by the difference in sequence length.

8） Japan’s welfare program permits very little “earned income retained”. According to the estimates of Hashimoto (2006), on average 83-93% of the earned income is balanced out. Put differently, if a welfare receiver earns 100 yen, only 7 to 17 yen will be his or her net income on hand. See Abe (2008) for a detailed discussion about the work disincentive problem of Japan’s welfare program.