2008/4 Research & Review

Is Japan's Service Sector Really Low in Productivity? - Analysis based on firm-level data

MORIKAWA Masayuki
Senior Fellow, RIETI
Senior Researcher, Japan Productivity Center for Socio-Economic Development (JPC-SED)

Micro-level analysis of service industries needed

Improving the productivity of service industries is seen as a top priority for achieving sustainable growth in the Japanese economy at a time when the nation's population is set to decline. Service Productivity & Innovation for Growth (SPRING), a business-government-academia forum established last year, has been undertaking various activities designed to increase the productivity of service industries.i

The productivity of Japanese service industries is widely believed to be low. Indeed, on an aggregate level, Japan strongly contrasts the United States, where rapidly increasing productivity has been observed in a wide range of service industries from the mid-1990s onward. However, attention must be paid to the high degree of heterogeneity among companies within the same industry. Also, there are many ways to look at productivity; in terms of labor productivity (LP) or total factor productivity (TFP), in the level or in the growth rate, or in comparison with the U.S. service sector or with Japan's manufacturing sector. Accurate "diagnosis" of the actual situation is a crucial prerequisite to formulating and implementing appropriate "remedies" to help increase the productivity of service industries. This involves more than just viewing average figures aggregated at an industry level; meticulous micro-level analysis, for instance at a firm or establishment level, is needed.

I would like to summarize the key findings from my analysis of firm-level data, conducted for the purpose of shedding light on such basic facts as whether the productivity of Japanese companies in the service industries is really low, and if so, why, and in which types of companies.ii

Is service sector productivity really low?

Analysis of productivity in service industries using microdata lags far behind that of the manufacturing industry, for which solid and reliable data are available. Even in the U.S. and Europe, such analysis has until now been conducted primarily on the manufacturing industry, and only recently begun to be applied to retail and other service industries. Consequently, the structure and dynamics of service sector productivity are now being found to differ substantially from those of the manufacturing sector.

In recent years, the Japanese government substantially expanded the coverage of service industries subject to the Basic Survey of Japanese Business Structure and Activities, enabling us to perform a panel analysis. Our analysis used the data for 2001-2004 obtained in this survey and we observed the following:

1. Service companies' productivity levels are not necessarily low relative to those of manufacturing companies. Many service companies have high productivity levels. Of the companies belonging to the (narrowly-defined) service sector, nearly 70% exceed the median for manufacturing companies in their TFP levels (see figure).

2. In terms of TFP growth rate, more than 60% of service companies exceed the median for manufacturing companies. However, the aggregate TFP growth rate for the entire service sector, weighted based on the size of companies, is substantially lower than the corresponding figure for the manufacturing industry. The sample does not include very small companies so definite conclusions cannot be drawn from this. However, it is fair to say that the relatively low rate of productivity growth for the entire service sector is more attributable to the low number of large blue chip companies capable of driving up productivity growth for the whole industry, rather than to the high number of small- and medium-sized enterprises (SMEs).

3. Dispersion among service companies in productivity levels is larger, and particularly notable in terms of TFP, than that in among manufacturing companies. Most of the dispersion of productivity level and growth rate among firms is intra-industry rather than inter-industry. The large dispersion of productivity among service companies is not attributable to the variety of types of businesses in the sector.

4. In the service sector, the redistribution effect of an increased share of high-productivity companies and a decreased share of low-productivity companies negatively contributes to sector-wide productivity, resulting in an increase in the shares of companies with relatively low productivity.

Share of service companies outperforming median manufacturers in productivity

What kinds of companies have high productivity?

Using the same data set, we also analyzed the relationships between TFP and typical company characteristics - e.g. research and development (R&D) intensity, use of ICT networks, foreign ownership, and corporate age - and made some findings.

1. R&D intensity works to increase the growth rate of TFP in the manufacturing sector, but such a relationship between R&D and TFP growth cannot be seen in the non-manufacturing sector.

2. Companies widely using ICT networks tend to have high productivity levels. This tendency is particularly strong in the service sector. However, after controlling for firm fixed-effects, increased use of ICT networks cannot be confirmed as directly contributing to an increased level and growth rate of TFP. It can be inferred that some company characteristics - the vitality and flexibility of each organization, managerial quality, and employee capabilities - that cannot be captured by statistics are more fundamental determinants of productivity.

3. Increased foreign ownership increases the level of TFP in the manufacturing sector, but this effect is not observed in the retail and other service industries. Inward direct investment has been expected to serve as a catalyst for boosting service sector productivity, but no such effect has been observed to date.

4. The younger a company, the higher is its level of TFP, and this relationship is more pronounced in the retail and other service industries. This indicates the possibility that start-ups and their growth may play an important role in improving sector-wide productivity.

Service sector productivity related to Japan's entire socioeconomic system

While the service sector has a number of highly productive blue-chip companies, substantial dispersion exists among companies' productivity levels and growth rates. This indicates a high potential for increased sector-wide productivity. The findings discussed above point out the importance of facilitating changes in organization capital, managerial quality, and other corporate characteristics as well as of creating an environment that promotes the start-up and growth of new companies with excellent managerial quality. Despite large productivity gaps among companies, attention must be paid to the malfunctioning of the mechanism for increasing productivity through the entry and exit of companies.

Service sector productivity is related to Japan's entire socioeconomic system which includes corporate governance, the functioning of the labor and financial markets, and the geographical distribution of economic activities.

These analyses are intended to stimulate debate and need to be further developed. RIETI is undertaking various research projects related to service sector productivity. I would like to invite readers of this column to revisit our findings in the future.

>> Original text in Japanese

Footnote(s)

i For details of SPRING activities, refer to its website (http://www.service-js.jp/).

ii For details of the analysis results, refer to "Is Productivity in the Service Industries Low? An Analysis Using Firm-Level Data on the Dispersion and the Dynamics of Productivity" (RIETI Discussion Paper 07-J-048) and "What Kinds of Company Have High Productivity? Company Characteristics and TFP" (RIETI Discussion Paper 07-J-049).

April 23, 2008