1. Hitotsubashi University and RIETI Hitotsubashi University
Why Did Japan’s TFP Growth Slow Down in the Lost Decade? An Empirical Analysis Based on Firm-Level Data of Manufacturing Firms
Kyoji Fukao
Hitotsubashi University and RIETI
Hyeog Ug Kwon
Hitotsubashi University
November 2005

2. １. Motivation
Recent studies suggest that the slowdown in Japan’s TFP growth in the 1990s was concentrated in the manufacturing sector.
However: Few studies have examined the causes of the TFP growth slowdown in the manufacturing sector.
For example, the “Zombie” hypothesis mainly focuses on non-manufacturing sector
(Ahearne and Shinada, 2004; Caballero et al., 2004).

3. The Structure of the Paper
1. Did TFP Growth Really Slow Down in the 1990s?
2. Decomposition Analysis of TFP Growth in the Manufacturing Sector
3. Inter-firm Differences in TFP

4. ２．Did TFP Growth Really Slow Down in the 1990s
２．Did TFP Growth Really Slow Down in the 1990s? A Survey of Studies on Japan’s TFP Growth
Different studies on Japan’s TFP growth produce different results.
But many studies find:
TFP growth slowed down in the 1990s.
The slowdown in TFP growth was more pronounced in the manufacturing sector.

6. 3. Decomposition Analysis of TFP Growth in the Manufacturing Sector
We use the firm-level panel data underlying the Basic Survey of Japanese Business Structure and Activities ( ).
Compared with Nishimura et al. (2003), our data cover a more recent period.
We use the method of Baily, Hulten and Campbell (1992) and Forster, Haltiwanger and Krizan (1998).

7. Measurement of firm f’s TFP level in year t

8. Decomposition of TFP growth at the industry level

9. Caveats
The Basic Survey only covers firms with at least 50 employees and 30 million yen of paid-in capital.
Our data on firms that “exited” include firms which shrunk or changed their main business from the manufacturing sector to another sector. Firms, which were merged and became part of another firm, are treated as “exited.”
We do not take account of the quality of labor.
We assume that all firms in an industry sell their products at the same price.

13. Results of Our Decomposition Analysis
In “high tech” industries, such as pharmaceutical products and electronics, TFP growth rates are high.
Compared with the U.S. and South Korea, the reallocation and entry effects in Japan are small.
In Japan, the net-exit effect was negative.

15. 4. Inter-firm Differences in TFP
Did inter-firm TFP gaps widen?
How do high-TFP firms differ from low-TFP firms?
Are low-TFP firms either catching up or exiting? Or are they continuing to stay in business and stay behind?
Preceding studies:
Morikawa (2004): Standard deviation of profit rate did not increase. But S.D. of sales increased.
Shinada (2003): Inter-firm differences in productivity widened in electrical machinery and transportation machinery. The analysis is based on the DBJ database.

20. Did Inter-firm TFP Gaps Widen?
The TFP gap widened mainly in industries with a high R&D intensity and where the internationalization of firms is more advanced
we measure internationalization by outward direct investment, the introduction of foreign capital, and procurements from abroad.

22. How do high-TFP firms differ from low-TFP firms?
The top-quartile firms show a higher R&D intensity, higher outward FDI ratio, and higher foreign procurement ratio.
They are bigger and have a lower liability/asset ratio.

23. Are low-TFP firms either catching up or exiting
Are low-TFP firms either catching up or exiting? Or are they continuing to stay in business and stay behind?
More than one-half of firms which originally ranked in the bottom three deciles in 1994 remained in the same three deciles in Similarly more than one-half of firms which originally ranked in the top three deciles in 1994 remained in the same three deciles in 2001.
Low-TFP firms did not shrink the number of workers by much more than high-TFP firms.
The likely reason is that high-TFP firms are in the process of restructuring their activities and many of them are relocating production abroad.
This finding suggests that the “metabolism” – the expansion of employment and output by high-TFP firms and the contraction or exit of low-TFP firms – is not working well in Japan’s manufacturing sector.

26. Main Result 1. Low “Metabolism”
According to our decomposition analysis, the exit effect was negative. The entry and reallocation effects were small.
We also found that the expansion of employment and output by high-TFP firms and the contraction or exit of low-TFP firms were small in Japan’s manufacturing sector.
Hollowing-out or zombie firm effects?

27. Main Result 2. Widening of TFP GAPs
We found that the TFP gap between the 75 percentile firm and the 25 percentile firm is widening in many industries where R&D intensity is high and the internationalization of firms is more advanced.
The TFP gap seems to be widening because high-TFP firms tend to have a higher R&D intensity, a higher degree of internationalization, are larger, and have a lower liability-asset ratio, and these characteristics enhance their productivity further.

28. Ongoing project JIP Database: 100 industries EU KLEMS project
JIP Micro-database
Establishment-level data of Manufacturing Survey