Comment on “Micro Price Dynamics during Japan's Lost Decades”


I think that Sudo et al. (2014) present some very interesting features of Japanese micro price dynamics that can be differentiated from the earlier work of Higo and Saito (2007) and Abe and Tonogi (2010). In particular, Sudo et al.'s work would be very valuable from a general perspective when it could uncover the potential systematic effects of deflation on changes in the behavior of disaggregated prices and consumption expenditure. In that regard, my comment begins with Sudo et al.'s five empirical findings. First, posted prices in Japan are ten times as flexible as those in the US scanner data. Second, regular prices are as flexible as those in the USA and Euro area. Third, the frequencies and sizes of price changes substantially differ across different products. Fourth, the importance of temporary sales in the consumption expenditure of households has risen during Japan's lost decades. Fifth, the frequencies of both upward movements in regular prices and sales are significantly correlated with aggregate labor market conditions, whereas other components are not. Given these results, I would like to address several issues that might deserve some discussion.

What happened in 1995 in Japan? Sudo et al.'s (2014) Figures 1, 2, and 5 raise the possibility that 1995 could be the threshold time point for the inflation rate, the number of products sold, and estimates of the price elasticity. If 1995 is an important turning point in the pricing and purchasing behavior of firms and households, it might be worthwhile to see if there is any expectations channel through which private agents can coordinate their beliefs about the future economic environment and reflect these coordinated beliefs into their decision making.

Do we need to distinguish between the source and the propagation mechanism of the deflation period (that might be relevant for micro pricing behaviors of firms)? Sudo et al. (2014) discuss the argument that the onset of the Japanese deflation might be attributed to the decline in productivity or the growth slowdown. However, it might be good to make a distinction between the source and the propagation mechanism of the deflation period in the following sense. Around 1995, households started to be sensitive to prices for their consumption choices, while firms began to use temporary sales that seemed to be more effective because of increases in the elasticity of prices in households' demands. The greater effectiveness of temporary sales gave firms less incentives to change their long-term prices named by reference prices. As the deflation period lengthens, Japanese households and firms got more pessimistic about future macroeconomic conditions. In particular, if such pessimism is on the road of self-confirmation, households become more sensitive to prices for their consumption choices, while firms depend more on temporary sales.

What is the relation between temporary sales and the macroeconomic situation (model-based estimates for the price elasticity of demand and the effectiveness of temporary sales)? An interesting result (based on the existing related literature) is that the frequency of temporary sales is significantly correlated with the macroeconomic environment like the indicators of labor market, while the price elasticity of demand has increased over time. In relation to the estimation of the price elasticity, an alternative approach (not explored in Sudo et al.'s paper) might be to seek a model-based estimate, such as the approach of Hausman et al. (1994), which takes account of a three-stage demand system in estimating demand for differentiated products (used in Levin & Yun, 2010, for a set of US scanner data). In addition, the work of Nakamura and Steinsson also indicates that there is a large increase in the frequency of US sales from 1988 to 2005, especially for processed food and apparel where the frequency of sales doubled. It might be thus interesting to investigate how much of the rising trend in the Japanese sales is due to the Japanese experience of a long-term deflation. In this case, it would be desirable to use a pricing model that can sort out the potentially distinct strategic motivations of temporary sales.