• E31;
  • C43

Imai and Watanabe (2014) meticulously examine possible downward biases in the Japanese consumer price index (CPI). In the past, we have been bombarded by arguments for the existence of upward biases in the Japanese CPI, mostly based on the prejudgment that the Japanese CPI has almost the same bias as its US counterparts. Imai and Watanabe make it clear that the bias in the CPI is not uniform among countries and economies.

To my limited knowledge as a former practitioner of monetary policy, Imai and Watanabe are the first to show that Japan's prices behave differently from those in the USA, in two ways. First, although mostly focused on supermarket items, Imai and Watanabe are able to show that deflation in the CPI for this category in the past several years is in fact associated with quality downgrades, so that the quality-adjusted prices did not fall as much as the official CPI suggested. Although there are methodological differences between this paper (using the geometric average of prices) and the CPI (using a consumption-basket weighted average of “representative” products' prices), Imai and Watanabe's price index “declined by about 16% in 2000–2012, with the rate of deflation per year being 1.3%, which is comparable to the figures for the corresponding items in the official CPI.” In contrast, the corresponding quality-adjusted price index (per-unit price taking account of downsizing) fell far less than the unadjusted price index and the corresponding CPI component. This suggests a downward bias in the official CPI component.

Second, this downward bias is not inconsequential because Japanese consumers base their consumption on quality-adjusted prices, not the quality-unadjusted price tags. Failure to recognize this downward bias may lead to a false assessment of inflationary dynamics. In fact, Japanese consumers are keenly aware that downsizing implies an increasing price, in a sharp contrast with their American counterparts, who tend to be more sensitive to the price tag (quality-unadjusted price) than the per unit price (quality-adjusted price). To put this differently, Japanese consumers are more rational (in an economic sense) than the US consumers studied by Gourville and Koehler (2004). Moreover, the Japanese consumers seem more to be more than simply rational: the majority of Japanese consumers appreciate upsizing and detest downsizing independently of the per-unit price. This may be interpreted as an appreciation of these firms' “sincere efforts” to give more to customers in the case of upsizing, and as detestation of their “cheating” their customers in the case of downsizing. The sense of fairness (“don't cheat”) seems to be found in Japanese customers.

There are several other lessons to be learned from the exercises taken in Imai and Watanabe. First, even though the Statistical Bureau of Japan has an appropriate method to adjust for quality change due to downsizing, the bureau apparently fails to take account of the widespread downsizing at least in these supermarket items. Why were their appropriate methods unable to detect and rectify this problem? Their “representative products” method seems to be the culprit. To identify a few “representative products” and to keep tracking only these items are a cost-saving way to make price indexes. However, this “rigidity” of product choice may cause some bias, if the chosen products become different from other items as time passes by. In this respect, random sampling is better, but very expensive. We need a reasonable compromise here.

Second, related to the first, is the downward bias detected by Imai and Watanabe stable over time? If it is stable, then such bias may be easily removed from the official CPI.

Third, there may be a store-sampling bias as well as a product-sampling bias. The 200+ stores in the Nikkei dataset may not be representative of retail stores in Japan. Moreover, supermarkets, which are the majority of these sample stores, have been in trouble for some time, struggling to stay profitable. The pricing behavior of supermarkets in the Nikkei sample may be different from other retail outlets, especially some profitable franchised convenience stores which are not included in the Nikkei sample.

Finally, although this is a bit of a transcendent comment, the downward bias may be much larger for other items in the Japanese CPI. The most notable candidates include market rents and imputed rents compiled from market rents. Some preliminary studies suggest that failing to take account of age effects on rents may lead to a substantial overestimation of rent deflation. The present procedure of simply averaging the per-square-meter rent in all rented properties in the rotating sample areas does not adjust for the increasing age of these properties. Unfortunately, we do not have studies to examine this issue as rigorously as Imai and Watanabe have done for supermarket items.


  1. Top of page
  2. References
  • Gourville J.T. & Koehler J.J. (2004). Downsizing price increases: A greater sensitivity to price than quantity in consumer markets. Harvard Business School Working Paper no. 04-042.
  • Imai S. & Watanabe T. (2014). Product downsizing and hidden price increases: Evidence from Japan's deflationary period. Asian Economic Policy Review, 9 (1), 6989.