Productivity in the Service Sector: Reforming Japan’s GDP statistics is urgently needed for more accuracy

FUKAO Kyoji
Faculty Fellow, RIETI

Raising productivity in the service sector is the most important challenge for the Japanese economy to exit secular stagnation. However, most activities in Japan's service sector (an amount equivalent to about 40% of the nation's gross domestic product (GDP)) is not accurately measured for real output or productivity, either at the sector or firm level due to data constraints in price statistics and national account statistics. Without correct measurements, it is difficult to conduct precise empirical analyses or devise improvements.

♦  ♦  ♦

The Figure shows changes in labor productivity (real GDP divided by total working hours) for Japan as a whole. Results are shown separately for the manufacturing and primary sectors (agriculture, forestry, and fisheries and mining) and other sectors (referred to below as service sectors, including construction). Increases in service sector labor productivity are significantly lower than those in the manufacturing and primary sectors, and the growth trend started going downward in the 1990s.

Figure. Changes in Labor Productivity (Prices in 2000, Yen/hour)
Figure. Changes in Labor Productivity (Prices in 2000, Yen/hour)
Source: JIP Database 2015

The service sector accounts for about 70%-80% of Japan's overall GDP and employment. It is no exaggeration to say that the chief reason for the sluggishness of the Japanese economy since the 1990s has been stagnant productivity in the service sector. Although much research has been conducted to explore the ways to improve productivity in the service sector, they have faced severe data constraints.

Increases in labor productivity are measured by the increase in real output per hour of work. In the manufacturing and primary sectors, increases in real output can be measured accurately in principle. Taking smartphone production as an example, if its performance improves even when the number of units produced per hour of work and the price per unit do not change over time, it improves consumers' benefits. In that case, it is considered that real output has increased (and thus real GDP also rises by the amount).

In the case of most products in the manufacturing and primary sectors, prices are considered as falling by the level of improvement in performance. As the real output is calculated by deflating the nominal output with the price index adjusting for such performance improvement, improvement in performance is correctly measured as growth in real GDP and labor productivity.

However, accurately measuring the productivity of the output in many service sectors is difficult for two reasons.

♦  ♦  ♦

The first reason is that there are services including public services and most of school education in which sufficient compensation is not paid in a market transaction, and the quality and quantity are hard to measure. Many countries including Japan use the amount of inputs of production factors as a proxy to measure the output of these types of services (output=inputs approach).

Specifically, taking total production costs as nominal output, the weighted average of the prices of factors of production input (wage rates, capital costs, intermediate input prices) as a price index, the factors of production input index calculated by deflating the total production costs with the price index is used as a proxy for the real production index.

In many sectors including manufacturing and many business services in which output can be measured directly, it is widely observed that real output growth is higher than real input growth via innovation and improvements in production efficiency. The difference between the two is called total factor productivity (TFP) growth. When output is measured by the output=inputs approach, TFP growth will be zero by definition. As such, there is a fundamental constraint in measuring productivity in this way.

There are also some problems that are unique to Japan. First, the United States and many European countries create output price indexes for the construction sector taking into account of quality in their price surveys. Real output is measured by deflating nominal output with this price index. Japan and the United Kingdom, on the other hand, have not yet created such price indexes for the construction sector, but instead measure it using the output=inputs approach.

In Japan, economic activity as measured by this approach is equal to about 20% of GDP. At the minimum, switching to the U.S. and European method for the construction sector is desired. I have been conducting joint research as part of a project of the Economic and Social Research Institute, Cabinet Office to create a price index for the construction sector based on such data as real estate transaction prices.

The second problem is that getting a correct measurement of the inputs is vital to the output=inputs approach, but in Japan, not enough consideration is given to this.

With the output=inputs approach, for example, even when total work hours do not change over time, real labor inputs increase by the amount of improvement in labor quality when average school years or skills increase in input labor, thus, it is necessary to regard that real output (as well as GDP) has increased by that amount. In the case of Japan, however, the GDP statistics do not make adjustments for labor quality in construction, private schools, and some social welfare services in Japan. Our research group estimates that insufficient consideration in the construction sector has caused Japan's GDP statistics to underestimate real GDP growth by 1.7 percentage points in total over the period 1973-2012.

Another factor that makes it difficult to obtain accurate productivity measurements relates to the issue of the price index used as the deflator to obtain real output of the wholesale and retail, health care, and other sectors (making up about 20% of GDP). Let's take the wholesale and retail sectors as an example.

Nominal output of the wholesale and retail sectors is calculated as the difference between selling price per unit and purchase price per unit (commercial margin). As for the price index used as the deflator, the Organisation for Economic Co-operation and Development (OECD) and Eurostat suggest that if the quality of wholesale or retail services per unit of traded commodity does not change over time, the margin price per unit (i.e., selling price per unit minus purchase price per unit) would be the appropriate measure of the price of services provided by wholesalers and retailers. The United States and Canada have already switched to this calculation method.

Although the Bank of Japan recently started to use the margin price per unit as the deflator in some commodity categories, Japan still uses the conventional method of using prices of traded commodities as the deflator for the wholesale and retail sectors. If the margin price per unit of traded commodity in the same form of transaction changes, the measurement results obtained by Japan could be very different from those by the United States and Canada. In the explanation that follows, it is assumed that the only factor of production input to wholesale or retail services is labor and the wholesale and retail margin equals labor cost as there is perfect competition among sellers.

The assumptions are that the purchase price and the number of units of traded commodity do not change, labor productivity rises for all shops in the same form of transaction (and with the same quality of services), and labor input for trading one unit of commodity is halved. Under those assumptions, by cutting the labor costs by half, both margin price and wholesale and retail margin per unit of traded commodity will be halved.

With the method recommended by the OECD, as the price of the wholesale and retail services is halved, the real output (wholesale and retail margin/wholesale and retail service price) does not change (and neither does GDP), but labor productivity doubles. With the conventional measurement, in contrast, because prices of traded commodities used as the wholesale and retail service prices only drops slightly, the real output becomes almost half (and GDP also falls), and thus an only slight increase in labor productivity is observed.

This means that the conventional measurement could underestimate productivity growth in wholesale and retail sectors. While the labor productivity in wholesale and retail sectors have been growing in the United States, that in Japan have stagnated. This may be caused by the difference in the GDP statistics employed in Japan and the United States.

♦  ♦  ♦

As such, there are various issues with Japan's service sector statistics. Its economic growth and productivity growth may be underestimated. Recognizing the importance of the service sector, many developed countries are improving their statistics by collaborating with the OECD and others.

Recognizing the shortcomings of the output=inputs approach, the United Kingdom, for example, is trying out a method of combining quantitative indexes (e.g., number of graduates) and qualitative indexes (e.g., average test scores) to measure productivity growth in many services correctly including education and social welfare.

Unfortunately, in Japan, partly because of insufficient staff at statistical offices, there has been little progress in improving service sector statistics. Correct measurement in productivity is inevitable for planning productivity growth policy. Fundamental reforms in service sector statistics is desired in Japan.

>> Original text in Japanese

* Translated by RIETI.

February 15, 2017 Nihon Keizai Shimbun

April 25, 2017