Policy Update 063
Measuring Productivity in the Service Sector: Current status and challenges
Senior Fellow, RIETI
1. Introduction: Current status of productivity measurement
Productivity is a key economic indicator typically used to monitor domestic economic trends and compare economic growth across countries. At the same time, productivity is a very familiar term used in everyday conversation to describe individuals' job performance, housework, and leisure activities. For instance, when the time required for housework is reduced by introducing a new electric appliance or improving the work process, or when we arrive at a certain destination earlier than expected by successfully beating the traffic jams, we congratulate ourselves for achieving higher productivity (efficiency). Also, when the acquisition of new skills--whether through a training program or by self-teaching--leads to improved work efficiency and higher income, we feel that our productivity has improved.
When we discuss productivity in countries, industries, or companies, we are principally talking about technological progress. We can think of two types of technological progress: one that will reduce the input of materials, cost, and time required for production and the other that is innovative and will enable a sudden, manifold increase in the value (price) of a product and hence contribute greatly to boosting corporate revenues and profits. High expectations are placed on productivity improvement associated with technological developments because it is the source of growth and there would be no sustainable economic growth without it. Here, what comes to mind is the development of various products such as automobiles, home electronic appliances, and personal computers by companies in the manufacturing industry.
In economics, studies on the measurement of productivity have long focused on the manufacturing industry. Japan is no exception. Due partly to its image--held at home and abroad alike--as a country that excels in production technologies led by certain manufacturers, the productivity and technological capabilities of the manufacturing industry have been perceived to be the source of growth.
However, when we look at the composition of the Japanese economy in recent years, the reality is that non-manufacturing industries account for 70% or more of the gross domestic product (GDP). In particular, the share of the service-providing industry (Note 1) in GDP stands above 20% (45% or more when including the transport, wholesale, and retail trade industries), consistently exceeding that of the manufacturing industry since 2000 (Note 2). As evident from these figures, we can no longer explain the overall economic activity by just looking at the manufacturing industry. We must find out the technological structure of non-manufacturing industries and then, by using theoretical and statistical models based thereon, measure the productivity of these industries. The service sector is extremely important in terms of its presence in the overall economic activity and its magnitude as a labor market. Despite this fact, very few studies have focused on this sector of the economy due to the difficulty of obtaining data and because service sector industries in general have failed to become recognized as major industries. As such, it remains unclear how they are structured on both the supply and demand sides.
In Japan, the service sector has been drawing attention in the past several years, ironically, because of its low productivity. In 2005, Japan was ranked 20th in the level of labor productivity among the then 30 member countries of the Organisation for Economic Co-operation and Development (OECD), but placed sixth in that of the manufacturing industry. This led to the perception that the weakness of the service sector is a drag on productivity. A more recent comparison in 2012 also shows that Japan was ranked seventh among the 34 OECD countries--the lowest among the major seven economies--in productivity in all industries but seventh in that of the manufacturing industry (Note 3).
Here, special attention should be given to how we define "productivity" in the service sector. A typical measure of productivity--which has been used in many cross-country comparisons, government reports, and research works--is labor productivity, i.e., sales or value added per worker.
Another measure of productivity, which takes into account the effects of input factors other than labor, is total factor productivity (TFP). Input factors refer to all factors involved in production such as the number of workers, physical capital (machinery, equipment, and facilities), raw materials, and energy. This has been used internationally since the mid-1950s as a measure of productivity for a broad range of industries. In theoretical and empirical research, TFP has been used applied in analyzing the manufacturing industry.
It is also more desirable to use TFP rather than labor productivity for the service sector. However, as aforementioned, economics has long failed to recognize the significance of this sector, and the smaller size of service-sector establishments--as compared to those in the manufacturing industry--makes it difficult to collect detailed data (particularly on machinery, equipment, and facilities). As a result, the development of statistical data for the service sector has been slow to make progress.
As is well known, the service sector is composed of a wide variety of industries, including labor-intensive ones as well as capital-intensive ones such as telecommunications and transport services. Therefore, labor productivity is not necessarily an appropriate measure of productivity.
Konishi and Nishiyama (2009) (4) estimated labor productivity and TFP by aggregating data on service segments (Note 4) of companies listed on the first section of the Tokyo Stock Exchange (TSE) (Note 5) for each industry type in the service sector, and compared these two measures to examine the validity of labor productivity as a measure of productivity. Table 1 shows the coefficients of the correlation between labor productivity and TFP by industry type. Except for the real estate industry in 2007, the two measures are positively correlated. In industries for which labor productivity and TFP show a high correlation (i.e., the two measures are similar in their average behavior), the effects of physical capital included in TFP can be considered negligible. In other words, as production activities in such industries can be assumed to be less dependent on machinery, equipment, and facilities, labor productivity can be taken as a first-order approximation of productivity. On the other hand, a low correlation between the two measures indicates that labor productivity and TFP are statistically different in their behavior. This can be interpreted as pointing to the need to incorporate the effects of physical capital (such as machinery, equipment, and facilities) in measuring productivity in the service sector as in the manufacturing industry.
|Transport||Wholesale||Retail||Real estate||Accommodations, |
What has been discussed thus far is the existing methods of productivity measurements. In the case of TFP, an increase in total output in a situation where inputs remain constant is interpreted as an increase in productivity or an improvement in the quality of inputs, which includes changes in all factors--technological progress, efficiency, inventions, etc.--other than the quantities of inputs. Also, TFP calculated from data reflects the effects of booms and busts in the economy as well as of unpredictable events such as major accidents and abnormal weather phenomena. It is also subject to the impact of possible demand shocks and various accidents on the firm level and hence invisible to researchers and other outsiders. Thus, it is important to eliminate the effect of the bias resulting from those factors in order to accurately identify productivity (See Konishi and Nishiyama (2013) (5)).
However, even if sufficient data on physical capital in the service sector are collected and the calculation of TFP without any bias becomes possible, it is still uncertain whether it is appropriate to apply this gauge--which has been developed as a measure of productivity in the manufacturing industry--to the service sector as is. In what follows, I will discuss the challenges that may need to be addressed, pointing to some of the characteristics unique to the service sector.
2. Difficulties in measuring productivity in the service sector
The measurement of productivity in the service sector involves addressing the following two potential problems:
1) It may be impossible to calculate TFP because of the lack of sufficient data (particularly on physical capital stock)
2) It may not be appropriate to apply TFP because of the diversity of service-sector industries.
The first problem may be solved by expanding the scope of statistical surveys to include those service-sector industries that are similar to those in the manufacturing industry in structure. However, even in the manufacturing industry, empirical analysis is subject to some limitations as TFP estimates inevitably incorporate factors other than technological progress. One reason behind this is that while in theory TFP only reflects the production activities of companies and assumes a numerical value for each variable accordingly, most data used for measurement in practice are in monetary terms, which include prices determined in the market and thus are demand factors. The non-availability of data on capacity utilization data and inventories may also cause bias. These are the difficulties involved in conducting an empirical analysis of TFP. More specifically, when demand factors are included, there is no telling whether a change in TFP is purely a change in productivity or attributable to demand shocks. This inhibits companies from making proper investment decisions and the government from making and implementing appropriate policies. Obviously, the identification of demand and supply factors poses a critical problem in analyzing the service sector where the provision and consumption of services usually occur simultaneously. In order to solve this problem, Morikawa (2014) (6) measures the productivity of service-sector industries by first estimating TFP based on numerical data for each industry and then controlling for the effect of demand factors faced by firms in the industry (Note 7).
Meanwhile, addressing the second problem requires developing a substitute measure of productivity for those industries where TFP is not appropriate. Since services are typically consumed as provided, service-sector industries are more significantly influenced by demand, as compared to the manufacturing industry. Because of this nature, productivity growth purely attributable to technological innovations cannot be determined by measuring TFP. Furthermore, the great diversity of service-sector industries makes it difficult to identify the sources of value added.
Let's take a look at the freight transport industry. In considering the productivity of this industry, we must first identify what constitutes value added in this particular type of service to transport goods from one place to another. After controlling for such factors as the performance of trucks, the quality of gasoline, and the conditions of roads, we can see that transport service companies seek to improve productivity through the combination of drivers' technical skills (including route selections), time required for loading and unloading, and logistics such as the location of warehouses and distribution centers. Also, unlike in the case of manufacturers, time is a critical input factor for transport service providers as they need to transport and deliver shipments according to timeframes set by shippers (i.e., a factor on the demand side). While we do not care how long it took to produce our living room TV and which is not something we consider in judging the quality or performance of the product, we do care how long it took to receive our parcel. Time is a critical factor in determining the levels of technical skills and customer satisfaction in transport services. In order to measure the productivity of the freight transport industry, it is necessary to formulate a production function for a trucking firm as well as a demand (or utility) function for a shipper (see Konishi, Mun, Nishiyama, and Sung (2013) (7)).
As another example, consider the case of the hairdressing and beauty salon industry discussed in Konishi and Nishiyama (2010) (8). Value added in this industry is defined as an improvement in the appearance of customers after hairdressing services. When haircutting and other technical skills are improved, time required for such services will be reduced, resulting in an increase in the quality of services per unit hour, greater customer satisfaction, and a greater probability for repeat customers. It also translates into an increase in the productivity of those hairdressers and hence an increase in the salon's sales. In this research, we used a demand- and supply-side model capable of simultaneously explaining customers' behavior and a production function that incorporates standard time required to perform a haircut as an input. Our empirical findings show that the maximum amount of services each hairdresser can provide per unit time (i.e., capacity) increases with the number of years of experience along with individual productivity. This approach is applicable to some other service-sector industries--such as beauty salons, restaurants, medical clinics, law firms, and information technology system development--where time corresponds with the productivity and quality of services or the level of customer satisfaction.
In order to observe productivity and technical capabilities in the service sector, it is important to obtain daily data on the maximum amount of services and standard time spent for performing a haircut. However, what we can usually observe is the lesser of the maximum amount of services available and actual demand for services.
For instance, a hairdresser capable of providing 20 haircuts during the opening hours would not spend two hours per customer even when there are only four customers. However, the fact that there were only four customers is all that is visible to analysts. We would be obviously underestimating the hairdresser's technical capabilities if we interpret this observation as an indication of his or her being able to provide only four haircuts per day. Measurement of productivity unaffected by demand is possible only when we know how long it takes to complete a haircut or how many haircuts can be provided per day. Therefore it is necessary to measure the standard time spent per customer and collect data on the hairdresser's capacity such as the number of haircuts performed on the days when there were more customers than could be handled.
In order to measure the level of technical capabilities and productivity in the service sector, we must first identify and define how and what services and goods are provided in each industry and what sort of value added such services and goods generate. After that, and only then, we should collect data on both the demand and supply sides for empirical analysis.
3. Characteristics of the service sector and lessons for the manufacturing industry
For those service-sector industries that have the characteristics listed below, it is particularly difficult to remove the influence of demand-side factors from measured TFP and hence, as discussed above, a new measure of productivity needs to be developed.
1) Simultaneity: Services are consumed as provided.
2) Inseparability: There is no receiving just part of the service; there is no separating the place of consumption from the place of provision.
3) Intangibility: There is no holding inventories; invisible.
Because of these three characteristics, service-sector companies are mostly domestic players and tend to be locally oriented in their operations. Therefore, the hollowing out of industries--a phenomenon in which production bases such as domestic factories are relocated to overseas destinations resulting in a decline in the competitiveness and production capabilities of domestic industries as seen in the manufacturing industry--is unlikely to happen in the service sector. This is because it is difficult to share, bundle together, or decompose the process of providing services. Also, as the degree of customer satisfaction depends on the subjective preference of individual customers as well as on the varying levels of service quality and technical skills among individual employees, service-sector companies need to invest in market research to learn the preference of customers in their service areas as well as in the education and training of employees.
Because of those factors, it is difficult for service-sector companies to divide and outsource their operations. Thus, it is not that they can relocate anywhere in pursuit of lower production costs. Indeed, as indicated by the fact that we often hear the term "servitization of the manufacturing industry," Japanese manufacturers are increasingly dependent on non-manufacturing functions, such as sales, leasing, and after-sales services, as their main source of value added. How they define this phenomenon--i.e., whether they see it as an increase in the cost of non-core operations or an opportunity that should be taken advantage of not only to decelerate the hollowing out of domestic industries but also to develop a new source of value added, differentiation, and competitiveness--will have a significant impact on their future corporate value.
Placed high on the agenda in the growth strategy of the government of Prime Minister Shinzo Abe, the improvement of service sector productivity has been subject to increasing attention. As discussed at the outset of this article, the reason for the growing attention is that Japan's service sector continues to compare poorly with that of other developed economies. And the reason for an increasing number of studies focusing on the service sector is that many researchers are questioning that finding. They are asking: "Is the productivity of services in Japan truly low? Doesn't it deserve a higher rating given the variety and quality of services provided?"
My research findings point to the importance of using not only supply-side but also demand-side data in measuring service sector productivity. Measuring productivity based solely on business-side (demand-side) data, as has been the case in most existing studies, poses the following problems:
1) The measured level of productivity of a company is subject to change depending on the size of the market or whether the company is located in an urban area with fierce competition or a rural area with a scarce population; and
2) The lack of observation of consumers' purchases from other companies inhibits accurate appraisal of productivity.
Also, the term "productivity" as used in the service sector could be referring to a mixed concept composed of both subjective and objective elements such as technical skills, quality, variety, and customer satisfaction. While it is difficult to identify each and every one of these elements, I believe it is possible to define this complex productivity as a factor that satisfies customers and encourage their return to the store or repurchase of services.
Going forward, the utilization of big data on consumers will have a great significance in measuring productivity in the service sector. Meanwhile, since big data held by corporations have been collected for use in their respective business management, not for analytical purposes, some improvements need to be made to the ways of collecting and/or processing data. For instance, it would be useful to have data covering the entire aspects of business operations, such as the time required to provide services, customer traffic, and actual demand for services including demand unfulfilled due to capacity constraints.
For the sake of efficient collection and accumulation of data, researchers and policymakers should work to identify needed data by setting assumptions and developing new theories and approaches. This would provide a clear purpose or usage for big data collection and thereby contribute to analyzing not only productivity but also various economic phenomena.
The establishment of the Nihon Service Award program, which offers the prime minister's and other Cabinet ministers' awards to excellent service providers every year, was necessary to create a common measure to assess services so as to ensure the transparency of the selection process. I hope that this will give momentum to the development of statistical data on service-sector industries and that more researchers will join us to carry out more research works in this field.
January 5, 2016
- ^ In this article, the term "service-providing industry" collectively refers to narrowly-defined service industries that are classified into categories L through R in the Japan Standard Industrial Classification (13th revision in October 2013). Broadly-defined service industries represent the tertiary industry and herein referred to as the "service sector." For the Japan Standard Industrial Classification, see the Ministry of Internal Affairs and Communications' website at the URL shown in Reference (1) below.
- ^ Also, as a share of the labor market, the presence of the service-providing industry has been on the rise since the 1990s to reach approximately 35% in 2012, whereas the share of the manufacturing industry has been on a downward trend since the 1990s to stand at approximately 17% in 2012. The percentage figures were calculated based on data from the National Accounts of Japan listed in Reference (2) below.
- ^ For an international and industry-by-industry comparison of labor productivity across the OECD member countries, see Japan Productivity Center's report listed in Reference (3) below.
- ^ By using data aggregated at the segment level, service divisions of non-service-sector companies—such as retail, real estate, financing divisions of manufacturing companies—can be included in a sample. This enables an analysis of data representing a closer-to-reality industrial structure, not constrained by the conventional industry classification.
- ^ In the case of listed companies, data necessary for calculating TFP can be obtained from their annual reports filed pursuant to Article 24 of the Financial Instruments and Exchange Act.
- ^ Excerpts from Konishi and Nishiyama (2009).
- ^ The book not only analyzes service-sector industries but also includes comparisons with the manufacturing industry to provide a comprehensive set of empirical analyses of productivity in Japan.
- (1) Ministry of Internal Affairs and Communications, Japan Standard Industrial Classification
http://www.soumu.go.jp/toukei_toukatsu/index/seido/sangyo/H25index.htm (in Japanese)
http://www.soumu.go.jp/english/dgpp_ss/seido/sangyo/san13-3.htm (in English)
- (2) Cabinet Office, National Accounts of Japan
- (3) Japan Productivity Center, "Nihon no Seisansei no Doko 2014-nenban [Productivity Trends in Japan 2014]," Chapter III Rodo Seisansei no Kokusai Hikaku [International Comparison of Labor Productivity]
- (4) Konishi, Y. and Y. Nishiyama (2009), "Segumento Deta o Mochiita Saabisu Sangyo no Seisansei no Keisoku [Measuring Productivity in the Service Sector Using Segment Data]," Keizai Ronso Vol. 183, No. 2, pp. 9-22.
- (5) Konishi, Y. and Y. Nishiyama (2013) "A Note on the Identification of Demand and Supply Shocks in Production: Decomposition of TFP," RIETI Discussion Paper, 13-E-99.
- (6) Morikawa, Masayuki (2014) "Productivity in Service Industries: Empirical analyses using microdata," Nippon Hyoronsha Co., Ltd.
- (7) Konishi, Y., S. Mun, Y. Nishiyama and J. Sung (2014) "Measuring the Value of Time in Freight Transportation," RIETI Discussion Paper, 14-E-004.
- (8) Konishi, Y. and Y. Nishiyama (2010) "Productivity of Service Providers: Microeconometric Measurement in the case of Hair Salons," RIETI Discussion Paper, 10-E-51.
- (1) Ministry of Internal Affairs and Communications, Japan Standard Industrial Classification
January 5, 2016
Article(s) by this author
Change in Eating Habits and Avoidance of Cash Becoming Clearer—Consumption Trends Evidenced by Big Data
May 14, 2021［Newspapers & Magazines］
May 21, 2019［Column］
June 2, 2017［Column］
July 25, 2016［RIETI Report］
July 25, 2016［Column］