• D12;
  • D30

Abe and Shiotani (2014) examine cross-sectional variation in prices facing households using Japanese homescan data. Following the seminal work by Aguiar and Hurst (2007) (AH, hereafter), Abe and Shiotani construct a price index at the household level and document that there is substantial heterogeneity in the price index. Interestingly, though the life-cycle profile of the price index is flatter, older households shop at higher prices than younger households in Japan, which is in sharp contrast to the patterns documented by AH for the USA. Furthermore, Abe and Shiotani find that bargain sales explain variations in prices to a large extent, and younger households take advantage of promotions more extensively than older households in Japan. Consistent with AH, however, the household-level price index increases with income and decreases with the shopping frequency in Japan.

This work has important implications for the measurement of household consumption. In most empirical analysis, economists use nondurable expenditure deflated by an appropriate aggregate price index as a measure of consumption. However, if households purchase the same commodities at different prices, the consumption measure may be biased. As Abe and Shiotani point out, the evidence for the heterogeneity in the price index is also important for monetary policy.

I start my discussion with the patterns of the discount use. Abe and Shiotani document that the elderly do not use discounts as extensively as the young. It is a little puzzling because the data suggest that older households shop more frequently, even though the standard deviation is large. Abe and Shiotani show that the store choices of households partially account for the pattern: older households use specialized stores more than younger households, while younger households use pharmacy and home improvement stores to a larger extent. In addition, the following two hypotheses may be worth examining to obtain a better understanding of the pattern. First, note that data on discount use contain membership discounts in AH, while, because of data limitation, Abe and Shiotani's paper focuses on the use of temporary sales measured by the fraction of expenditure spent at the “bargain” prices. If the temporary sales are concentrated when younger households visit stores, for example in the evening, younger households may be able to use sales more extensively than older households. Abe and Shiotani document that the price index slightly increases with age for purchases at supermarkets, which seems to be consistent with this hypothesis.

Second, I wonder if discounts offered by online shops help account for the pattern. It should be noted that the sample periods in AH and Abe and Shiotani are different. The sample period in AH covers 1993–1995, while Abe and Shiotani covers 2004–2006. Between these two periods, online shopping expanded substantially. Since younger households are more likely to shop online than older households, this might help account for the different patterns of the use of discounts. It would be very interesting to examine the patterns of discount use in more detail, if possible, using other data sources collected from retail stores.

Next, I would like to discuss the comparison between Abe and Shiotani and AH. First, these two papers use different samples: The sample in AH includes one-person households (singles), while the sample is restricted to married couples in the dataset that Abe and Shiotani use. It is likely that singles have less shopping time than married couples (especially with one spouse not working), and thus are likely to shop at higher prices. Because the fraction of singles is larger among younger households than older households, the omission of singles may account for the increasing life-cycle profile of the price index.

Second, unlike AH, Abe and Shiotani examine the life-cycle profile of the price index after controlling for income. It is likely that households with higher income (wage) face higher opportunity costs. Therefore, after controlling for income, age effects in Abe and Shiotani may capture factors other than the price of time such as age-specific shopping productivity. I think that this may be the case, for example, because the differential in prices paid by those aged 45–49 and those aged above 60 remains stable at 0.8–1.0% across different specifications with and without shopping frequency.


  1. Top of page
  2. References