Seasonality is among the most salient features of price changes, but it is notably less analyzed than seasonality of quantities and the business cycle component of price changes. To fill this gap, we use the scanner data of 199 categories of goods in Japan to empirically study the seasonality of price changes from 1990 to 2021. We find that the following four features generally hold for most categories: (1) The frequency of price increases and decreases rises in March and September; (2) Seasonal components of the frequency of price changes are negatively correlated with those of the size of price changes; (3) Seasonal components of the inflation rate track seasonal components of net frequency of price changes; (4) The seasonal pattern of the frequency of price changes is responsive to changes in the category-level annual inflation rate for the year. We use simple state-dependent price models and show seasonal cycles in menu costs play an essential role in generating seasonality of price changes.
It is widely known among both scholars and policymakers that the time series of prices have a sizable degree of seasonality. Figure 1 shows the decomposition of the yearly growth rate of the CPI, for all items and for goods less fresh food and energy, into twelve month-to-month changes within the same year in Japan. It can be seen that there are months in which prices generally increase, such as March and April, and months in which prices generally decrease, such as January and February. Such seasonal patterns have been stable from the 1990s to 2020s.
WP052
Standard New Keynesian models have often neglected temporary sales. In this paper, we ask whether this treatment is appropriate. In the empirical part of the paper, we provide evidence using Japanese scanner data covering the last two decades that the frequency of sales was closely related with macroeconomic developments. Specifically, we find that the frequency of sales and hours worked move in opposite directions in response to technology shocks, producing a negative correlation between the two. We then construct a dynamic stochastic general equilibrium model that takes households’ decisions regarding their allocation of time for work, leisure, and bargain hunting into account. Using this model, we show that the rise in the frequency of sales, which is observed in the data, can be accounted for by the decline in hours worked during Japan’s lost decades. We also find that the real effect of monetary policy shocks weakens by around 40% due to the presence of temporary sales, but monetary policy still matters.
Standard New Keynesian models have often neglected temporary sales, although the frequency of sales is far higher than that of regular price changes, and hence it is not necessarily guaranteed that the assumption of sticky prices holds. Ignoring this fact is justified, however, if retailers’ decision to hold sales is independent of macroeconomic developments. If this is the case, temporary sales do not eliminate the real effect of monetary policy. In fact, Guimaraes and Sheedy (2011, hereafter GS) develop a dynamic stochastic general equilibrium (DSGE) model incorporating sales and show that the real effect of monetary policy remains largely unchanged. Empirical studies such as Kehoe and Midrigan (2010), Eichenbaum, Jaimovich, and Rebelo (2011), and Anderson et al. (2012) argue that retailers’ decision to hold a sale is actually orthogonal to changes in macroeconomic developments.
We study micro price dynamics and their macroeconomic implications using daily scanner data from 1988 to 2013. We provide five facts. First, posted prices in Japan are ten times as flexible as those in the U.S. scanner data. Second, regular prices are almost as flexible as those in the U.S. and Euro area. Third, heterogeneity is large. Fourth, during Japan’s lost decades, temporary sales played an increasingly important role. Fifth, the frequency of upward regular price revisions and the frequency of sales are significantly correlated with the macroeconomic environment like the indicators of labor market.
Since the asset price bubble went bust in the early 1990s, Japan has gone through prolonged stagnation and very low rates of inflation (see Figure 1). To investigate its background, in this paper, we study micro price dynamics at a retail shop and product level. In doing so, we use daily scanner or Point of Sales (POS) data from 1988 to 2013 covering over 6 billion records. From the data, we examine how firms’ price setting changed over these twenty years; report similarities and differences in micro price dynamics between Japan and foreign countries; and draw implications for economic theory as well as policy.
