Japanese Firms' Inflation Expectations during the Current Inflationary Phase"
1. Introduction
The Japanese economy suffered from a prolonged period of low inflation. For example, the consumer price index (CPI), excluding fresh food and the effect of increased consumptiontax rates, moved between -2 and +2 percent. This trend suddenly diverged at the beginning of 2021. For instance, the rate of increase of CPI soared to +4.1 percent in January 2023and remained at a relatively high level of +2.6 percent in March 2024. Although the recent surge in inflation has caused various problems in the Japanese economy, it has also offerednew frontiers for analyzing Japan’s economy, such as breaking the persistent deflationary trend in the pre-pandemic period.
The Infotainment Research Center published two studies on the inflation expectations (IEs) of Japanese households: “Instability of Japanese households inflation expectations duringthe current inflationary phase” (published in November 2023) and “Japanese households’ inflation perceptions: the formation process and their relationship with the inflationexpectation” (published in February 2024). These studies pointed out that: (1) Japanese households’ short-term IEs significantly exceeded the rate of increase of the CPI during thecurrent inflationary phase; and (2) longer-term IEs also soared compared to those of U.S. and European households.
This study also focuses on IEs, albeit on those of the corporate sector. The main results are summarized as follows:
First, based on Tankan data, Japanese firms' IEs moved around 1 percent before the pandemic. This rose to 3 percent for one-year-ahead and about 2 percent for five-year-aheadexpectations during the current inflationary period. Although larger firms showed lower IEs than smaller firms, no distinct differences were observed between the types of industries.Data on different economic sectors provided supporting evidence for the relationship “IEs of economist and financial market participants < IEs of firms < IEs of households.” Aboutthe half of the firms answered “don’t know” about five-year-ahead IEs due to economic uncertainty; this share increased by additional 5 to 10 percentage points after the pandemic.
Second, firms’ IE mechanisms do not completely follow either rational or adaptive expectations. Rational expectations do not hold because of: (1) a wide dispersion of IEs by firmsize, (2) a clear upward bias of IEs compared to the CPI, and (3) observed correlations between firms’ expected output prices and IEs. In addition, pure adaptive expectations are notsupported because: (1) firms assign higher priority to IEs when actual inflation soars, and (2) IEs are affected by the movement in input costs.
“Rational inattention” (RI) theory became a useful framework for explaining such seemingly inconsistent firms’ behaviors. RI assumes that firms intentionally and rationally ignoreinformation that has relatively small value for their activities. According to the RI theory, firms assign a higher priority to IEs in business operations when inflation rates or input costsrise significantly, whereas firms assign a lower priority to IEs if deflationary conditions last for a significant period.
Third, IEs are not anchored at the 2 percent inflation target (a state of IE moving stably around the central bank’s inflation target). Even longer-term IEs, such as three- and five-year-ahead IEs, have clearly risen during the current inflationary phase. Similar patterns are observed in the U.K., Italy, and other major developed countries.
Data and previous studies on this subject in and outside Japan are limited compared to those on household sectors, economists, and financial markets. Therefore, further researchchallenges include accumulating related data and analyzing the formation mechanisms of IEs.
The remainder of this article is organized as follows. Section 2 summarizes the increased attention towards firms’ IE. Section 3 uses Tankan data to analyze firms’ IEs. Section 4discusses the IE formation mechanism and whether firms’ IEs are anchored around the inflation target. Finally, section 5 presents the conclusions of the study.
2. The purpose of analyzing firms’ IEs
2.1. Increased attention towards firms’ IEs
As mentioned in Section 1, studies on firms’ IEs in and outside Japan are limited. This is mainly due to the following two reasons.
First, data on firms’ IEs are limited. Existing problems in gathering data on firms’ IEs are difficulties in (1) increasing the sample size due to firms’ reluctance to participate, (2)making samples representative of the population in terms of type of industry and size of firms due to sample size limitations, and (3) unifying respondents’ job positions.
Second, academics have focused their attention on the IEs of financial market participants and professional economists. This is because they seem to be better informed about thedevelopment of future inflation than ordinary people. Some studies assumed that every participant had perfect rational expectations and used the IEs of professional economists, evenwhen analyzing the IEs of households and firms. However, in reality, as businesspersons have valuable information on their firms’ input and output prices, they have their ownperspectives on future inflation which are different from those of professional economists.
The tendency to ignore firms’ IEs began to change in the 2010s supported by the following reasons. First, major economic theories began to explicitly treat firms’ price-settingbehaviors in their models. Second, firms’ unique price-setting behaviors attracted economists’ attention when analyzing deflationary economies during the pre-pandemic period.Third, during the current inflationary period, when most developed countries had marked dual-digit inflation rates, economists were curious about how the surge in CPI would affectvarious types of IEs, including households, firms, and financial market participants. Moreover, central bank economists, in addition to academics and professional economists, beganpaying attention to firms’ IEs, since the movements of IEs are one of the important factors in determining the effects of monetary tightening policy under high inflation.
Reflecting such movements, since 2010s, central banks in major developed countries began surveying firms’ IEs (Table 1). The Bank of Japan added questionnaires on firms’ IEs aspart of the Tankan survey that began in March 2014. This effort was ahead of most major developed countries central banks, which started their surveys in the late 2010s or during2020s.
2.2. The relationship between firms’ IEs and the macroeconomy
Before proceeding to the data analysis, this section discusses how firms’ IEs affect the macro economy.
First, firms’ IEs affect the price-setting behaviors of products and services. Final output prices are also affected by the demand and supply conditions in the market. However, as thecurrent inflationary phase shows, firms’ price-setting behaviors played a larger role than usual due to the cost-push inflation caused by price surges in raw materials, depreciating yen,and weakening global supply chains under the influence of the U.S.?China trade conflict.
Second, IEs significantly affect wage-setting behaviors. A typical case was the U.S. economy in the 1970s, when surging inflation increased IEs and caused an upward wage-pricespiral. Once the economy falls into such a condition, containing inflation requires a prolonged tight monetary policy and enduring high unemployment rates, thereby significantlydamaging the economy.
Third, firms’ IEs influence real interest rates. Even if the nominal interest rates are constant, the real interest rates for individual firms are affected by their IEs. Because real interestrates are an important factor in deciding fixed investments, they in turn affect the business cycle.
Fourth, IEs affect monetary policy efficiency. Firms’ IEs are anchored when their long-term IEs move close to the inflation target and are not significantly affected by changes in theactual inflation rate. If this holds, an upward wage-price spiral is unlikely to occur and high inflation rates will swiftly converge to the inflation target (whether Japanese firms’ IEs areanchored is discussed in Section 4.4).
3. Detailed analysis of firms’ IEs
3.1. Japanese firms’ IEs
This section analyzes firms’ IEs using Tankan data. Tankan is the only quarterly survey conducted in Japan on firms’ IEs. The Bank of Japan conducts the survey, covering morethan 10,000 firms located in Japan. The survey askes firms about Diffusion Index (D.I.), such as business conditions, and supply and demand conditions, as well as figures for salesand current profits. It is considered important survey data in Japan to judge business conditions. In 2014, questions about expected inflation (one-, three-, and five-years ahead) andoutput prices (one-, three-, and five-years ahead) (Figures 1 and 2) were added.
Japanese firms’ IEs moved stably at approximately 1 percent for every term during the pre-pandemic period (see Figure 2). They remained positive even when the CPI fell intonegative territory in 2020. Since 2021, IEs began surging and peaked at 3 percent for one-year-ahead IEs and approximately 2 percent for five-year-ahead IEs. One-year-ahead IEscontinued to decline moderately after March 2023, whereas five-year-ahead IEs remained around the peak level. Meanwhile, the three-year-ahead IEs moved closer to the five-year-ahead IEs.
3.2. Relationship between IEs and output prices
In addition to general prices, Tankan asks about firms’ outprice expectations (Figure 2). The one-year-ahead expectations in both figures exhibit similar trends. Expected output pricesfell into negative territory in 2020, similar to the CPI, whereas IEs did not.
Tankan’s figures for the three- and five-year-ahead expected output prices are compiled as the cumulative change rates relative to the current level. Figure 3 modifies the releasedfigures and converts them into annualized change rates from one to five years ahead.
Only the one-year-ahead expected output prices are affected by actual prices. Annualized changes in expected prices for one-to-three years and three-to five years moved stably atapproximately 0.5 percent, despite significant changes in the CPI. Looking closely, in the pre-pandemic period, three-to-five years ahead expectations fell to 0.2 to 0.3 percent. In therecent period, one-to three-year expectations showed slight increases. Overall, while Japanese firms expected a 1 to 2 percent increase in general prices, they assumed a modest 0.5percent annual increase in their output prices.
Outside Japan, expectations of output prices are also available in the survey conducted by the central bank of Italy (Figure 4). The survey began in 2013, a year earlier than in Tankan.Figure 4 shows that expectations of output prices and general prices moved very closely. One difference from the Tankan survey results is that Italian IEs fell significantly faster thantheir output price expectations during the current inflationary phase.
3.3. The relationship between the firms’ inflation perceptions and the IEs
Households' IEs are significantly influenced by how they perceive current inflation rates, or inflation perceptions. Although Tankan does not cover inflation perceptions, surveysconducted in the U.S., U.K., and France do. Although these surveys commonly suffer from relatively short sample periods, the inflation perception of firms moved closely with theCPIs, except in the U.K., where perceptions exceeded the CPI. Since household inflation perceptions are significantly above the actual CPI in most developed countries (Japanesehouseholds shown in Figure 8), this suggests that firms and households have different IEs formation mechanisms.
3.4. IEs by the size of firms
Tankan’s IE survey has a unique advantage over other surveys because of its large sample size, which exceeds 10,000 firms. This enables a detailed analysis of IEs by firm size andindustry type. This section analyzes the firms’ IEs according to their size.
Figure 9 clearly shows that levels of IEs have been stable at “large firms <medium sized firms <small firms” for the whole sample period. Since larger firms have relatively moreinformation on general prices and macroeconomic conditions than smaller firms, larger firms’ upward bias of IEs relative to the CPI is clearly less than that of smaller firms.
Figure 10 shows the expected output prices according to firm size. The output prices of large and medium-sized companies moved similarly, while those of small-sized firms havebeen higher than those of the other two, especially during the current inflationary phase. Because the IEs of small firms are higher than those of large firms, as shown in Figure 9, smallfirms seem to expect a larger increase in their output prices in the future.
The Bank of Italy conducts a survey of IEs depending on the size of the firms (Figure 11). During the current inflationary phase, although the IEs of small firms exceeded those oflarger firms, the difference seems smaller than those from Tankan. One of the main reasons behind this phenomenon is its unique survey method, in which information on the currentinflation rate is given to respondents just before asking their IEs.
Economists criticize this unique survey method because it significantly reduces the upward bias of IEs relative to the CPI. In fact, if we compare Italian IEs (Figure 12) with those inTankan (Figure 1), the Italian survey’s gaps between long- and short-term IEs are unnaturally smaller than those in Tankan, especially during the pre-pandemic period.
3.5. IEs by type of industries
The Tankan data can be broken down by industry type. Figure 13 shows the IEs of the manufacturing and nonmanufacturing sectors. The difference between the two sectors is only 0.1 percent point, which is relatively small compared to the 0.6 percentage point difference in IEs between large and small firms.
Figure 14 shows the standard deviations (SDs) of 36 industries (19 manufacturing and 17 non-manufacturing industries). In this graph, an increase in the SDs implies that the IEs ofdifferent types of industries are more widely dispersed.
Although the SD level was relatively stable during the pre-pandemic period, it began to surge in tandem with IEs during the current inflationary phase. The SDs peaked in the middle of2022 and fell significantly afterwards, such that the level of SD in March 2024 was almost equal to that at the beginning of 2021. On the other hand, IEs peaked at the beginning of2023 but remained at 2.5 percent even in March 2024. Consequently, the gap between IEs and SDs has widened. It seems that at the beginning of 2021, the SDs were pushed upsignificantly by huge uncertainties created by the pandemic and then peaked and fell sharply as the pandemic was gradually contained and uncertainty diminished.
Figure 15 plots the SD for one, three, and five years ahead for the Tankan firms. This clearly indicates that the sharp increase in the SDs was limited to expectations one year ahead,which showed synchronized movement with a surge in actual inflation and IEs.
No survey outside Japan provides detailed IEs by industry type. However, a survey conducted by the Federal Reserve Bank of Cleveland revealed SDs and IEs (Figure 16). Similar toTankan, the SDs increased significantly during the current inflationary phase.
3.6. Comparing IEs with other economic entities
This section compares firms’ IEs with those of other economic entities, such as households and economists. In case of Japan, households’ IEs are available using the “OpinionSurvey” conducted by the Bank of Japan. Figure 17 shows the IEs of firms and households, along with the CPI. This clearly indicates that the IEs of firms are consistently lower thanthose of households.
The figure also shows that (1) the IEs of firms comoved with those of households and CPI, (2) during the current inflationary phase, the increasing pace of households’ IEs exceededthose of firms; therefore, the peak rate for households reached 10%, while those of firms remained at 3 %, and (3) during the current inflationary phase, the peak period of IEs of firmswas several quarters earlier than that of households.
The Bank of Japan also published a similar graph showing various IEs in the “Outlook for economic activities and prices” (Figure 18). According to the figure, firms’ IEs movedconsistently above those of professionals, including economists, financial market participants, and households, especially during the current inflationary phase (shapes of IEs in Figure18 differ from Figure 17 because the former plots five-year-ahead IEs, while the latter shows those of one-year-ahead).
Although both Figures 17 and 18 depend on the same “Opinion Survey” data for households IEs, the significant differences in the level of IEs were caused by the data themselves andthe calculation methods. The survey asked the respondents to respond to their IEs in two ways. The first directly asked about IEs qualitatively, and the second asked respondents toselect from five choices depending on the magnitude and direction of IEs. Figure 17 uses the median figure obtained from the quantitative answers, whereas figure 18 depends on a five-choice answer and uses an econometric method to quantify the responses. The Bank of Japan explained the reason for choosing qualitative answer rather than quantitative ones bysuggesting the facts that quantitative answers contain many “bias” such as answers concentrated on integers, multiplier of 5, and 0 percent. The Bank concluded that using the medianfigure of quantitative answers could not precisely represent the complex distribution of household IEs.
However, the unique response patterns of qualitative answers are common for household IEs, not only in Japan but also in many major developed countries. Such a pattern isconsidered to be closely related to psychological factors and frequently occurs when respondents are not confident in their answers. Therefore, it is unlikely that a unique patternemerged because of technical problems, such as survey methods or respondents’ irresponsibility. Because respondents are not confident about their responses, it is hopeless toimprove the quality of the outcome of the five-choice responses by mathematically quantifying the data. This mathematical modification seems to have created an unrealistic situation,as shown in figure 18, where households’ IE was as low as that of professionals.
Other countries offer data to compare firms’ IEs to those of other economic entities. First, the Bank of Canada publishes the IEs of households, firms, and economists (Figure 19).This indicates that the level of firms’ IEs is located between those of households and professionals and that the IEs of households surged in the current inflation phase, similar toJapan in Figure 17.
Second, in the U.S., firms’ IEs are also located between households and economists, similar to Japan and Canada. However, in the current inflationary phase, firms’ IEs are closer tothose of households than those of economists (Figure 20).
Third, in Norway, firms’ IEs also lie between economists and households, except in the current inflationary phase(Figure 21). The peak rate of IEs of firms is 6 percent and that ofhouseholds is approximately 4.5 percent.
Fourth, the Bank of France published a research paper in 2021 that presented a table indicating that firms’ IEs lie between those of households and economists (Figure 22).
Other studies also state that firms’ IEs lie between those of households and firms. In addition, a study published in 2022 in the U.S. pointed out that firms’ IE formation mechanismdiffers from those of households and economists. Another study, published in the U.S. by the Federal Reserve Bank of Cleveland, calculated the SDs of firms’ IEs and found thatthey were much smaller than those of households.
This evidence suggests that the in terms of IEs, the relationship “economists<firms<households” holds. Thus, the Bank of Japan’s chart (Figure 18) seems to be an outlier.
4. Formation mechanism of firms’ IEs
4.1. How confident are firm in responding to IEs?
This section describes the formation mechanisms of firms’ IEs.
First, this subsection uses Tankan data to examine how confident are firms in answering about their IEs. Tankan’s questionaries allow a “don’t know” answer if firms are unable tocome up with numerical IEs. Figure 22 plots the share of “don’t’ know” responses. This share increases as the duration of the expectation periods becomes longer. For example,before the pandemic, the share was approximately 15 percent for one-year-ahead, jumped to 30 percent for three-year-ahead, and reached approximately half of respondents for five-year-ahead. Understandably, the longer the expectation period, the more difficult it is to provide a quantified answer about IEs owing to increased uncertainties. Still, it is remarkablethat almost half of the firms have no view about five-year-ahead. In addition, the share of “don’t know” has increased an additional 5 to 10 percentage point since the first half of 2020for every expectation period. At the time of writing this article in March 2024, the share of three- and five-year-ahead remained high. Tankan further asks the reasons for choosing the“don’t know” answer. Figure 23 showed that most respondents picked “uncertainty.” Judging by the timing, the outbreak of the pandemic may have caused the increased uncertaintyamong firms. Next, Figure 24 depicts the share of “don’t know” by manufacturing and non-manufacturing sectors. Before the pandemic, the manufacturing sector’s share was higherthan that of the non-manufacturing sector. This might reflect the fact that (1) the demand and supply conditions of manufacturing goods fluctuate more widely than those of servicegoods and (2) the manufacturing sector faces higher uncertainty due to fluctuations in overseas demand and volatile foreign exchange rates.
Figure 25 shows the share of “don’t know” answer by the firm size. The share of large firms is approximately 10 percent point higher than that of SMEs. This is odd because largefirms generally seem to have a relative advantage in gathering macroeconomic information and developing detailed long-term business plans.
4.2. Business operations and IEs
Using prior studies, this subsection examines how IEs affect firms’ business operations. A U.S. study asked firms about how much general prices affected their business operations(Figure 26). The figure plots the responses for two different periods: 2015, when the inflation rate was modest, and 2022, when the inflation rate increased rapidly during the pandemic.The 2015 responses showed that almost 90 percent of firms chose low effect response answers such as “no effects” or “medium effects.” In contrast, the responses of 2022 shiftedrightward. The share of “no effects” declined while that of “strong effects” increased. This indicates that the information value of general prices for business operations variesaccording to the level of inflation rate.
The same study asked firms about the effects of general prices on product prices (Figure 27). Surprisingly, approximately half of the surveyed firms answered “no effect.” Thus, inputprices, rather than general prices, are the major factors determining product prices.
4.3. Formation process of firms’ IEs
This subsection uses Tankan and other overseas surveys to analyze the formation process of firms’ IEs.
The following three pieces of evidence suggest that firms are not perfectly rational when forming IEs. First, although rational firms should form identical IEs based on the sameinformation, Figure 9 using Tankan data indicates that the level of IEs diverges with firm size. Second, the level of firms’ IEs is consistently higher than the actual CPI or professionalforecasts both inside and outside Japan (Figures 17, 19, 20, and 21). Such upward bias frequently occurs when entities’ inflation perceptions, which provide the basis for IEs, areinaccurate. Third, IEs by industry type in the Tankan survey are significantly affected by their own output prices (R2=0.65). If the firms are perfectly rational, the two variables shouldshow no correlation (see Figure 28).
These observations contradict rational expectation theory, where economic entities grasp all economic conditions perfectly and decide on IEs with full information.
Since firms are not perfectly rational, are they following adaptive expectations theory, where IEs are calculated mechanically depending only on past and present data on inflationrates? This is not the case, since Figure 26 indicates that firms do care about IEs when they face large shocks, such as pandemics or inflation surges. In addition, the Tankan datasuggest that fluctuations in firms’ input costs affect their IEs (R2=0.72), which contradicts the adaptive expectation theory where only past and present inflation rates matter (Figure29). Another survey conducted in New Zealand in 2013 found that firms set priorities in monitoring macroeconomic data. For example, the share of firms closely monitoring GDPreached approximately 80 percent, whereas that of inflation rates was about 50 percent. The survey also pointed out that: (1) firms’ forecast errors for highly prioritized GDP weresmaller than those for inflation rates, (2) firms belonging to highly competitive industries formed accurate IEs, and (3) firms generally collected abundant price information about theirown industries. Such observations also imply that firms do not form IEs through simple calculations, as adaptive expectations theory suggests.
The RI theory is a useful framework to explain such seemingly inconsistent firms’ behaviors where firms are not perfectly rational nor simply adaptive. RI assumes that firms aregenerally rational and motivated to efficiently allocate limited resources. Hence, they intentionally neglect information that has a relatively small impact on their activities and requires non-negligible costs to make accurate estimations. Thus, RI’s basic idea is that firms are rationally inattentive to low-value information.
According to the RI, firms assign priorities among macroeconomic indicators according to their impact on business activities at the time; hence, priorities can vary. For example, thepriority of IEs increases when an inflation surge or a pandemic occurs. However, the long-term deflationary period seen in the pre-pandemic period lowered the priority of IEs. Thispriority also increases when competitiveness within the industry becomes fierce and fluctuations in input costs increase.
Another study showed that firms in high-inflation countries, such as Uruguay, Ukraine, and Argentina, place higher priority on inflation rates and analyze IEs more carefully than firmsin low-inflation countries, such as the U.S. and New Zealand. Another study conducted in the U.S. reported that firms expected a prolonged inflation surge after the pandemic earlierthan economists. This is because economists regarded the inflation surge as a temporal phenomenon due to supply shortages; meanwhile, firms were aware of the ongoing increasingvulnerability in global supply chains, and therefore, expected prolonged inflation surges.
4.4 Unanchored firms’ IEs
Anchored IEs imply that long-term IEs move stably around the central bank’s inflation target (often set at 2 percent) without being significantly affected by the fluctuations in short-term IEs. During the pre-pandemic period, since inflation rates moved rather stably at low levels in many developed countries, IEs were also relatively stable. Studies focusing on thisperiod concluded that IEs were anchored. The situation changed dramatically after the current inflationary phase, where short-term IEs notably responded to the inflation surge.Furthermore, the long-term IEs, which should be stable regardless of the movements in short-term IEs, moved upward.
This “de-anchoring” of long-term IEs is evident in Japan (Figure 1), Italy (Figure 12), the U.K. (Figure 7), and Canada (figure 19). Most studies also reported that firms’ IEs are notanchored.
First, only 25 percent of firms were aware of a 2 percent inflation target in the U.S. Meanwhile, only one-third knew the target in New Zealand, where the central bank has continued itsinflation targeting policy for 25 years.
Lastly, since Candia et al. (2021) pointed out five conditions for anchored IEs to meet, the question of whether Japanese firms’ IEs are anchored is examined below based on theseconditions:
(1) The average IEs should be close to the inflation target: In Japan, the target is 2 percent. Long-term IEs hovered around 1 percent before the pandemic and rose to 2 percent onlyrecently.
(2) Fluctuations in IEs should be relatively small: The SD of Tankan firms is shown in Figure 15. The SD of one-year-ahead IEs significantly increased during the current inflationphase. An overseas study also showed that the SD of IEs was much larger than that of professionals.)
(3) Firms should have confidence in their IEs: As Figure 22 shows, approximately half of Tankan firms chose “don’t know.” This suggests that firms are not confident in theirforecasts.
(4) IEs, especially those in the long term, should be stable: Figure 1 indicates that the five-year-ahead IEs shifted substantially from the 1 percent level prevalent in the pre-pandemicperiod to a 2 percent level in the current inflationary phase.
(5) Long- and short-term IEs should not exhibit co-movement: Plotting the first difference of long/short IEs showed significant correlations (R2 = 0.77, Figure 30), indicating that bothvariables exhibit co-movement.
Thus, Japanese firms’IEs are not anchored.
5. Conclusion
This study analyzed firms’ IE and their formation process during the current inflation phase. The main results are summarized below.
First, analyzing Tankan data revealed that Japanese firms' IEs moved around 1 percent before the pandemic. This rose to 3 percent for one-year-ahead and about 2 percent for five-year-ahead expectations during the current inflationary period. Although larger firms showed lower IEs than smaller firms, no distinct difference was observed between the types ofindustries. Data from different economic sectors showed that firms' IEs lie between those of professionals and households. About the half of firms answered “don’t know” for five-year-ahead IEs due to economic uncertainty; this share increased an additional 5 to 10 percentage points after the pandemic.
Second, firms’ IEs mechanisms do not follow a completely rational or adaptive expectations model. Rational expectations do not hold because of: (1) a wide dispersion of IEs byfirm size, (2) a clear upward bias of IEs compared to the CPI, and (3) observed correlations between firms’ expected output prices and IEs. Firms do not follow pure adaptiveexpectations either because (1) they assign higher priority to IEs when actual inflation soars and (2) IEs are affected by the movement in input costs.
RI theory became a useful framework for explaining such seemingly inconsistent firms’ behaviors. RI assumes that firms intentionally and rationally ignore information that hasrelatively small value for their activities. According to the RI, firms assign a higher priority to IEs when inflation rates or input costs rise significantly, whereas they assign a lowerpriority to IEs if deflationary conditions last for a significant period.
Third, IEs are not anchored at the 2 percent inflation target (a state of IE moving stably around the central bank’s inflation target). Even longer-term IEs, such as three- and five-year-ahead IEs, clearly rose during the current inflationary phase. Similar patterns are observed in the U.K., Italy, and other major developed countries.
Data and previous studies on this subject in and outside Japan are limited compared to those of households, professionals, and financial markets. This trend gradually began to changein the 2010s, as firms’ pricing behavior during the low-inflation period attracted economists’ interest. Accordingly, much data on firms’IEs, including those in Japan, was now beingcollected, albeit mainly by central banks.
Finally, studies on firms’ IEs have several implications for central bank policy communication. First, central banks need to consider how they can reach medium and small firmswhose IEs contain a larger upward bias than large firms. Second, methods or tools to promote firms’ interest in IEs during the low-inflation period, when firms follow RI and thepriority of inflation is lowered, should be examined. Third, central banks should explore how they can establish repetitive communication tools for firms that can help sustain theeffects of policy communication.
References
Albrizio et al. (2023) “Mining the gap: Extracting firms’ inflation expectations from earnings calls,” S. Albrizio, A. Diziori, P. Simon, IMF-WPs 202/23, October 2023
Bank of Japan (2016) “Firms' inflation expectations and wage-setting behaviors” BOJ-WP series, No. 16-E-10, July 2016
--- (2017) “New facts about firms’ inflation expectation formation: Part 1 revisiting sticky information model,” Y. Uno, H. Naganuma, H. Hara, Bank of Japan WP series No. 17-J-3March 2017 (in Japanese)
--- (2018) “New facts about firms’ inflation expectations: Simple tests for sticky information model,” Y. Uno, S. Naganuma, N. Hara, Bank of Japan WP series, No. 18-E-14, October2018
--- (2019) “The formation of Japanese firms’ inflation expectations: A survey data analysis,” H. Inatsugu, T. Kitamura, T. Matsuda, BOJ WP series, No. 19-E-15, November 2019
Bouche et al. (2021) “Measuring firms’ inflation expectations” P. Bouche, M. Gerardin, E. Gautier, E. Savignac, Bank of France Bulletin, August 2021
Bryan et al. (2015) “The inflation expectations of firms: What do they look like, Are they accurate, and Do they matter?” M. Bryan, B. Meyer, N. Parker, Federal Reserve Bank ofAtlanta Working Paper Series 2014-27a
Candia et al. (2021) “The inflation expectations of U.S. firms: Evidence from a new survey,” B. Candia, O. Coiboin, Y. Gorodnichenko, NBER WP 28836, May 2021
--- (2022) “Macroeconomic expectations of firms,” B. Candia, O. Coiboin,
Y. Gorodnichenko, NBER WP 30042, May 2022
Coibion et al. (2015) “How do firms form their expectations? New survey evidence,” O. Coibion, Y. Gorodnichenko, S. Kamer, NBER WP No. 21092, April 2015
Fukuhara (2024) “Japanese households’ inflation perceptions: the formation process and their relationship with inflation expectations,” T. Fukuhara, Infotainment Research Center,February 2024
Garciga et al. (2023) “The survey of firm’s inflation expectations,” C. Garciga, E. Knotek II, M. Pedemonte, T. Schiroff, Economic Commentary, Federal Reserve Bank ofCleveland, May 2023
Meyer and Sheng (2021) “Unit Cost Expectations and Uncertainty: Firms’ perspective on inflation,” B. Meyer, X. Sheng, Federal Reserve Bank of Atlanta WP Series 2021-12b,Revised March 2024
Richards and Verstraete (2016), “Understanding firms’ inflation expectations using the Bank of Canada’s Business Outlook Survey,” S. Richards, M. Verstraete, Bank of CanadaWP No. 2016-7, February 2016
Savagnac et al. (2021) “Firms’ inflation expectations; New evidence from France,”
F. Savagnac, E. Gautier, Y. Gorodnichenko, O. Coibion, NBER WP No. 29376, October 2021
Weber et al. (2022) “The subjective inflation expectations of households and firms: Measurement, determinants and implications,” M. Weber, F. D’Acunto, Y. Gorodnichenko, O.Coibion, NBER WP No. 30046, May 2022
Japanese Households' Inflation Perceptions: The Formation Process and Their Relationship with Inflation Expectations
This study examined the formation process of household inflation perceptions as well as their relationship with inflation expectations and implications for monetary policyimplementation. Inflation perceptions attracted relatively little attention until recently because they had constantly exceeded the actual CPI and were considered unreliable. This attitudehas changed as inflation perceptions became considered the basis for forming inflation expectations. At the same time, the following reasons are offered to account for the inflationperceptions constantly exceeding the CPI. The first is a lack of households’ basic knowledge regarding the CPI. The second is people’s tendency toward loss aversion. The third ishouseholds’ lack of awareness of the CPI’s quality adjustments. Regarding the relationship between inflation perceptions and expectations, the following evidence was found. First,socioeconomic attributes of consumers affect both inflation perceptions and expectations in the same way. Second, according to household surveys, households heavily rely oninflation perceptions when responding to inflation expectations. Third, crosssectional analysis using individual sample data shows strong correlations between the inflation expectationsand perceptions. Such features of inflation perceptions have monetary policy implications. First, although food and gas prices are excluded from the “core-core CPI,” they should bemonitored closely. Second, differences in inflation expectations caused by socio-economic attributes distort efficient resource allocation. Third, households’ financial literacy shouldbe improved. Future challenges in this field include further data accumulation, which enables detailed studies on the formation process of inflation perceptions
1. Introduction
The Japanese economy has experienced a long period of low inflation. During this period, the Consumer Price Index (CPI), excluding perishables and the effects of increases inconsumption tax, has hovered around the minus-2-to-plus-2-percent range. This low-inflation phase abruptly ended at the beginning of 2021, as the corporate sector began to pass onthe jump in import goods prices to the prices of its own products. In January 2023, the CPI peaked at + 4.1 percent on a year-on-year basis and remained high at +2.8 percent as ofOctober 2023 (Figure 1). Such increases in the inflation rate have caused various economic problems in Japanwhile providing a unique opportunity to analyze households’ inflationperceptions and expectations. The Infotainment Research Center published a study in December 2023 titled “Instability of Japanese Households’ Inflation Expectations During theCurrent Inflationary Phase” (December Study). The study highlighted two findings. First, Japanese households’ inflation expectations significantly overshot the CPI increase rates(CPI: +2.9 percent; one-year-ahead inflation expectation: +10 percent; five-years-ahead inflation expectations: +5 percent). Second, even though the U.S. and European CPI increasedto around +10 percent, their one-year-ahead expectations remained at a maximum of +5 to 6 percent, and the five-years-ahead inflation stayed at +3 to 4 percent. The study concludesfrom the above evidence that Japanese households’ inflation expectations are more unstable than those in the U.S. and Europe.The December Study also showed that the movementof inflation expectations has been closely related to inflation perception, which is households’ view of current price development. Based on the December Study, this study examinesthe formation process of inflation perceptions. Simultaneously, the relationship between inflation perceptions and expectations and the implications for monetary policy implementationare discussed. The main results of this study showed that inflation perceptions have attracted relatively little attention because they constantly exceed the actual CPI and are consideredunreliable. However, their information value has recently become highly appreciated, as inflation perceptions have been considered the basis for forming inflation expectations. In thisprocess, the following explanations are offered for inflation perceptions constantly exceeding the CPI. the first is the lack of basic household knowledge of the CPI. Households relyon the prices of frequently purchased goods such as food and gas as the most reliable sources of inflation perceptions. The synthesized CPI, consisting of food and gas prices, fitsmore closely with inflation perceptions than the overall CPI.Second, people have a tendency toward loss aversion. People are said to be 2.5 times more sensitive to price increases,which lead to a loss in purchasing power, than price decreases. Therefore, price increases cause significant damage to consumer psychology and boost the perception of inflationabove the actual CPI.Third, there is a lack of awareness of households’ CPI quality adjustments. Although such adjustments reduce the price index, consumers simply ignore them,leading to a wider divergence between the CPI and inflation perceptions.Regarding the relationship between inflation perceptions and expectations, the following evidence was found.First, both inflation expectations and perceptions are affected in the same way by differences in consumers’ social attributes, such as gender, income level, and academic background.Second, according to household surveys, households rely heavily on inflation perceptions when responding to inflation xpectations. Third, a cross-sectional analysis using individualdata revealed strong correlations between inflation expectations and perceptions.These findings have implications for monetary policy implementation. First, although food and gasprices are excluded from the “core-core CPI,” their developments should be monitored closely as they are tightly related to the inflation perceptions. Second, differences in inflationexpectations due to socioeconomic attributes also affect real interest rates and resource allocation. Third, households’ financial literacy should be improved as financially literatepersons actively gather macroeconomic information,which, in turn, narrows the divergence between the CPI and inflation perceptions.Because research on inflation perceptionsremains scarce, future challenges in this field include further data accumulation, which enables studies on the formation process of inflation perceptions and their relationship withinflation expectations.The remainder of this paper is organized as follows. Section 2 presents the relationship between perceptions of inflation and expectations. Section 3 highlightsthe factors contributing to higher inflation perceptions than the CPI, including households’lack of knowledge regarding the CPI, reliance on prices of frequently purchased goods, andthe effect of CPI quality adjustments. Section 4 provides evidence of the close relationship between inflation expectations and perceptions. Section 5 discusses the implications ofmonetary policy implementation. Finally, section 6 concludes the study.
2. Inflation Perceptions Exceed Changes in CPI
2.1. Inflation Perceptions
Figure 2 shows the relationship between inflation perceptions and the CPI in Japan. The features of the inflation perceptions are twofold. First, they consistently exceeded the CPI.The divergence amounts to approximately 2?4 percent, even before the current inflationary period. During the current inflationary phase, even though the peak CPI rate remained at +4.1 percent, inflation perceptions reached 10 percent. Second, they closely followed CPI movement. Third, even when the CPI decreased to negative values, inflation perceptionsremained at zero percent. Inflation perceptions exceed the CPI not only in Japan but also in other major developed countries, a phenomenon often called the “inflation perceptionconundrum”. Figure 3 shows the Eurozone case. Although inflation perceptions closely follow CPImovement, their level is 5 to 10 percent higher than that of the CPI. Particularlyduring the current inflationary phase, inflation perceptions reached 20%, whereas the CPI remained at approximately 10 percent.
2.2. The Relationship Between Inflation Perceptions and Expectations
Figure 4 illustrates the relationship between inflation perceptions and expectations in Japan. It illustrates several features. First, both variables move closely. Second, inflationexpectations are more stable than inflation perceptions. Third, during the current inflation phase, inflation expectations increased almost as much as inflation perceptions. Fourth,inflation expectations seem to lead perception by three months.Figure 5 illustrates the situation in U.K. households. It also exhibited several features. First, inflation perceptions andexpectations behave similarly. Second, except for the current inflationary phase, the levels of inflation perceptions and expectations are very close to each other. Third, unlike Japan,during the current inflation phase, inflation expectations remained below 5 percent, while perceptions increased to approximately 10 percent. Fourth, inflation expectations leadperception to a certain degree.Next, Figure 6 shows the case of Eurozone households and shows four features. First, similar to households in Japan and the U.K., inflation perceptionsand expectations behave similarly. Second, before the current inflation phase, the gap between expectations and perceptions was wider than in Japan and the U.K. Third, during thecurrent inflation hase, the gap widened to more than 10 percent point. Fourth, inflation expectations seem to lead perceptions, as in the cases of Japan and the U.K.Based on the abovethree examples, three conclusions can be drawn. First, the two variables moved closely together and followed the movement of the CPI. Second, inflation expectations show moremoderate movements than do perceptions. Third, inflation expectations seem to lead perceptions to a certain degree. As the December Study showed, Japanese households’ inflationexpectations are more unstable than those of the U.K. and the Eurozone.Until recently, most economists in developed countries rarely paid attention to households’ inflationperceptions as they are too high to “justify” or “impractical to use in economic analysis.” Therefore, studies on the gap between the CPI and inflation perceptions are scarce.However, calculating the statistical relationship between inflation expectations and CPI/inflation perceptions revealed that inflation perceptions are much more closely correlated withexpectations than the CPI. In Japan, the correlation between inflation perceptions and one-year-ahead inflation expectations is higher (0.84) than that between the CPI and expectations(0.77). Inflation expectations, on the other hand, are regarded as one of the important macroeconomic variables. Professional forecasts of inflation expectations, such as surveys ofeconomists and people engaged in financial markets, or inflation expectations derived from yields on inflation-linked bonds are often used as typical inflation expectations rather thanthose of households. When the inflation expectations are adapted in macroeconomic models, they often assume rational expectations (people act according to the assumption that theyknow all of the information about the macroeconomic variables) or adaptive expectations (people rely on past inflation rates as a proxy for future inflation rates). Nonetheless, asshown in Figures 4-6, inflation perceptions follow CPI movement, although the level and degree of fluctuation are somewhat different. Thus, economists have paid increasing attentionto inflation perceptions since the early 2000s. In doing so, the first task was to uncover the reasons why inflation perceptions constantly exceed the CPI.
3. The Formation Process of the Households Inflation Perceptions
This section discusses the formation process of inflation perceptions, focusing on the fact that they constantly exceed the CPI. Inflation perceptions are affected by various factorssuch as consumers’ purchasing patterns, information-gathering behavior, and knowledge of the economy. This section identifies five factors that affect inflation perceptions. The firstis the price of frequently purchased goods. Second, we consider food and gas prices. The third is consumers’ tendencies toward loss aversion, the fourthis the CPI’s qualityadjustment, and the fifth is individual experiences with inflation and deflation.
3.1. Correlation between inflation perceptions and the CPI
As Figure 2 shows, inflation perceptions move closely in relation to the CPI. Their correlation is high at 0.76 (observation period: June 2007 to September 2023 on a quarterly basis).In the U.K., the correlation is 0.90 (observation period: November 1999 to September 2003 on a quarterly basis). The Eurozone’s correlation is 0.82 (observation period: March 2004to September 2003 on a quarterly basis). In case of the U.S., for which the observation period is the longest among developed countries, the correlation between the CPI and inflationexpectations using the University of Michigan Survey was 0.91(observation period: January 1978 to September 2003 on a monthly basis).
3.2. Households’ Source of Information in Forming Inflation Perceptions The above subsection describes the strong correlations between inflation perceptions
and the CPIs in major developed countries. The next question is what type of information households rely on when responding to household surveys. First, we examine howknowledgeable households are regarding inflation. The “Financial Literacy Survey,” conducted by the Central Council for Financial Service Information in Japan, contains basicquestions on inflation. The first item is as follows:
“When inflation is high, overall price level, including that of necessities and service, will go up swiftly.”
Respondents are given three choices: “correct,” “incorrect,” and “don’t know.” Survey results shows that 64 percent answered “correct,” while 3 percent chose “incorrect” and 29percent responded “don’t know.” Surprisingly, approximately 40 percent of respondents who chose “don’t know” and “incorrect” do not understand the basic concept of inflation.Some argue that continued deflation in Japan may have affected theseresponses. The next question is more neutral.
“If the inflation rate is 2 percent and you had a deposit interest rate of 1 percent, how much goods and service you can buy one year from now?”
Only 55% of the respondents answered this question correctly, while 11 percent chose the wrong answer, and 33 percent responded that they did not know the answer. As thisquestion shows, about 40 percent of consumers (“incorrect” plus “don’t know”) lack a basic understanding about the inflation.The responses to the two questions above suggestthat at least 40 percent of consumers lack a basic understanding of inflation and its economic consequences. This implies that these households are incapable of understanding the roleof the CPI or monitoring its actual value. In other words, consumers who can respond regarding inflation perception and expectations based on accurate information related to the CPIaccount for only 60 percent at most, a disappointment for economists. Next, the Bank of Japan’s “Opinion Survey,” conducted in September 2013, directly inquired into sources ofinformation for inflation perceptions. Figure 7 presents the survey results. The most popular answer was “price development of frequently purchased goods,” which included foodand gas prices, followed by “information from TV news and newspapers” and “regular income and payment.” However, each share of the total response is less than half that offrequently purchased goods. Although this question allows multiple responses (up to three responses), approximately half of the respondents relied on the price of frequentlypurchased goods when asked about their inflation perceptions. Similar surveys have been conducted in other developed countries; Figure 8 shows an example of a survey conductedin the U.S. The most popular responses when asked about the inflation expectations is “own shopping,” followed by “family and friends,” which is also related to daily shopping.Mass communications, such as TV, radio, and newspapers, follow, but their share is much lower than the top two.Figure 9 shows the results of a similar survey conducted inGermany. As the survey was conducted in 2023, the top choice was “war in Ukraine,” but the second choice was “fuel,” and the third was “food prices.” “Media about inflation”was fifth, and its share was only about the half that of food prices and gas prices. An European Central Bank (ECB) study segregated consumers into “better performers” and “worseperformers” depending on differences in the formation of inflation perceptions. The former are clever consumers who refer to objective information such as mass media reports whenforming inflation perceptions. The latter are unwise consumers who depend on their own experiences, such as shopping and the price development of frequently purchased goods. InJapan, the share of “worse performers” seems to reach 40 to 50 percent, which may be a reason for the significant divergence between the CPI and inflation perceptions. A survey bythe central bank of Germany showed that even though consumers knew about mass media reports regarding inflation, 89-92 percent of them still relied mainly on information acquiredthrough their own shopping experience. Figure 10 shows the results of the U.K. survey conducted by the Bank of England (BOE). As in other countries, the first to sixth choices wererelated to the prices of frequently purchased goods. Media reports on inflation development and the actual CPI ranked seventh and last, respectively. In another survey, 45 percent ofhouseholds that were asked about the CPI inflation rate answered “don’t know,” and only 10-20 percent answered correctly. Figure 11 shows the distribution of inflation perceptionsin the Eurozone and reflects several features. First, the responses are concentrated at multiples of five such as 0, 5, 10, 15, and 20. Second, even for the range between 0and 5percent, which is close to the actual CPI, responses are concentrated on integers such as 1, 2, and 3. Third, although extremely high responses (such as 30 percent) were observed toa certain degree, responses below zero were rare. These finding align with “rounding responses,” a concept used in psychology, which often emerges when respondents are notconfident with their own responses. Most of the“worse performers” may have given such responses.
3.3 Inflation Perceptions and Price Index Based on Frequently Purchased GoodsI
Figure 13 plots the price index of frequently purchased goods and inflation perceptions. Compared to the overall CPI, shown in Figure 2, the gap between price index and inflationperceptions has narrowed. However, before the current inflationary period, inflation perceptions are much more stable than the frequently purchased goods index.The “consumer priceindex” statistics refer the “family income and expenditure survey” for consumers’ frequency of purchases. According to these statistics, six items were purchased once or more perweek, all of which were food. Goods purchased once or more per month include 80 items, such as food, gas, medical expenses, and daily necessities, and their share of totalexpenditure amounts to slightly more than 10 percent.The fact that the most frequently purchased goods consist of food and gas is consistent with the statistical analysis and anecdotalevidence. In fact, a paper published by the Bank of Japan showed the effect of food and gas prices on inflation perceptionsusing statistical models. Another study by the Central Bankof Germany mentioned food and gas as frequently purchased goods. A study by the Ohio State University conducted a simulated consumer purchasing behavior test and concludedthat food and gas prices are affecting inflation perceptions (they termed this “frequency bias,” and it exists in perception). Furthermore, a stus described in the previous subsection, a“worse performer” relies heavily on the price development of frequently purchased goods rather than macroeconomic information when forming inflation perceptions. This tendencyhas been reported not only in Japan but also in the U.S., the Eurozone, Germany, and Denmark. The Japanese Consumer Price Index publishes a price index according to purchasefrequency (Figure 12). The price development of frequently purchased goods has three features. First, it is more unstable than the overall CPI, or goods purchased less than once permonth. Second, during the entire period shown in the figure; the average rate is 1.4 percent, which is higher than the overall CPI average of 0.6 percent. Third, during the currentinflationary period, although the overall CPI has already peaked, the frequently purchased price index continues to rise. Study by the Central Bank of Denmark stresses the significanteffect of food prices on inflation perceptions. Another study by the Bank of Italy showed that consumers cannot precisely recallinformation on rarely purchased goods such asdurables, and this phenomenon distorts inflation perceptions.
3.4. Inflation Perceptions and Food and Gas Prices
The last subsection reveals that consumers refer to the prices of frequentlypurchased goods, such as food and gas, in forming inflation perceptions. In this subsection, we calculatethe weighted average index based on the food and gas prices in Japan. The weight of food is 26.2%, and that of gas is 1.8 percent, according to the CPI statistics (Figure 14).Comparing the food and gas price index (FGPI) shown in Figure 14 with the overall CPI (Figure 2) and the prices of frequently purchased goods (Figure 13) yielded three features.First, the gap between inflation perceptions and FGPI is smaller than thoseamong the other indices. Second, the FGPI closely follows the surge in inflation perceptions, especiallyduring the current inflationary phase. Third, the correlation between inflation perceptions and the FGPI increases to 0.84, whereas that with the overall CPI is 0.76. These findingsindicate that most households refer to the FGPI when forming inflation perceptions.The CPI is often criticized for its significant divergence from inflation perceptions. The CPI is aweighted average of the prices of individual items using the share of their expenditure as their weights. Although such an approach is reasonable as an official statistic, there is room todevelop a new price index that can follow inflation perceptions more closely than the official CPI by changing the weights from expenditure share to frequency of purchases andfocusing only on frequently purchased goods and services. Some Japanese economists developed a new price index based on this notion. They calculated their own price index basedon (1) limiting items to goods and services purchased more than once a week and (2) changing weights from expenditure to frequency of purchases. This new index amounted to 14.0percent in June 2023, which is much closer to inflation perceptions than the overall CPI.Furthermore, some argue that the gap between the inflation perceptions and overall CPI ispartly due to the fact that the CPI does not cover real estate prices.16 Many consumers have misunderstood that the CPI included real estate prices because they did not understandits definition and coverage. An Italian survey revealed that more than half of the respondents (51.4%) incorrectly reported inclusion of house prices in the CPI definition.Figure 15shows households’ outlook for real estate prices surveyed by the Bank of Japan’s “Opinion Survey.” “Land Prices D.I.” is calculated as deducting the proportion of respondentswho answered “land prices will go down” from the proportion who responded “land prices will go up.” The D.I. increased steadily from the beginning of 2021 until the most recentsurvey, and this upward trend may have contributed to heightened inflation perceptions during the current inflationary phase. This finding implies that real estate price movementsshould also be considered when developing a new price index.
3.5. Inflation Perceptions and Loss Aversion of Consumers
The last subsection discusses the effects of the FGPI as a cause of the divergence between the CPI and inflation perceptions. However, this factor alone cannot explain the overalldivergence between the two indicators. This is because, as Figure 2 indicates, although inflation perceptions constantly exceed the overall CPI, the FGPI is not guaranteed to moveconstantly above the overall CPI. Therefore, additional factors are required to explain this phenomenon. From this point of view, this subsection discusses consumers’ loss aversion,and the next subsection considers the effect of the quality adjustment of CPI statistics as a factor that provides upward bias to inflation perceptions. Loss aversion is a concept oftenused in behavioral economics and psychology. It is a psychological phenomenon in which people are 2.5 times more sensitive to suffering the same amount of loss than gainingpleasure. It also applies to daily purchasing behaviors, where increases in the prices of frequently purchased goods, which is a loss to the household budget, cause much more damageto consumers than the same amount of price decline, which is a gain to the household budget. A study by the Bank of England referred to this phenomenon as “salient events,” and itmay have increased inflation perceptions significantly. As Figure 14 shows, the FGPI is more unstable than the overall CPI, and this, combined with loss aversion, may have increasedinflation perceptions. Such a possibility has been pointed out in many related studies, including those by the Economic and Social Research Institute in Japan, the Central Banks ofCanada, Denmark, and England, the Federal Reserve Board in the U.S., and the European Central Bank. Figure 16-1 shows the share of price-increasing items in Japan. Although rawdata were not available, and statistical analysis could thus not be conducted, the timing of the widening gap between inflation perceptions and the overall CPI in 16-2 and the timing ofthe growing proportion of price-increasing items in 16-1 nearly match. During the current inflationary phase, the gap between inflation perceptions and the CPI widened significantly,which was not seen in the low-inflation period. This may partly reflect consumers’ loss aversion, stimulated by a surge in consumer goods prices and an increasing proportion of price-increasing items. Figure 17 showed the relationship between news heard of price decreases/increases and inflation expectations in the U.S.19 Consumers’ response to pricedevelopment is asymmetrical and much more sensitive to “price increases” than “price decline,” which is consistent with the loss aversion theory. Additionally, one-year-aheadinflation expectations, as a proxy for inflation perceptions, are significantly affected by news about price increases.
3.6. Inflation Perceptions and the Quality Adjustment of the CPI
The CPI statistics adjust each item’s price index according to changes in quality. For example, even if the retail prices of PCs remain unchanged, their price index will decrease as theirperformance improves. On the other hand, consumers focus only on changes in retail prices. Therefore, these differences in price perceptions cause an upward bias in inflationperceptions. Shimizu, 20 the Federal Reserve Board, 21 and the central banks of Denmark, Ireland, and Canada refer to the effect of quality adjustment. The Bank of Canada estimatesthat quality changes lower the CPI by 0.2 percent.
3.7. Consumers’ Past and Present Experiences of Inflation
Consumers’ past and present experiences of inflation may affect inflation perceptions and expectations. For example, the “Great Inflation” in the U.S. in the 1970s and the “Oil PriceShock” in Japan in the same decade have contributed to increased inflation expectations. The December Study showed that inflation expectations as surveyed by the University ofMichigan rose significantly in the 1970s and remained high until the first half of the 1980s. By contrast, economies have suffered from a long period of low inflation or mild deflationsince the late 1990s in Japan and the late 2000s in other developed countries. Because price fluctuations remained modest during this period, consumers paid less attention to pricemovements. Instead, their attention shifted to other, more urgent economic issues such as the employment situation and wage increases. Such changes in attention occur due topsychological mechanisms that efficiently allocate limited recognition resources according to the priority of economic issues. The consumers’ tendency to pay less attention to pricedevelopment during low-inflation periods is called “rational inattention.” This is rational behavior in the sense that when prices are stable, it is rational to allocate limited cognitiveresources to other economic issues and intentionally pay less attention to price development. For example, a study by the Bank of Canada pointed out that the 25 years of low inflationperiod has caused Canadian consumers “rational inattention” to price developments (Figure 18). The Bank of England’s survey showed that the percentage of respondents whochose “inflation is significant problem” was more than 50 percent in the high-inflation period of the 1970s but had declined to less than 10 percent just before the pandemic began in2019. Another study by the Central Bank of Germany argued that because the young generation only experienced a low-inflation period, rational inattention became dominant. 24However, this rational inattention suddenly disappeared at the beginning of the current inflationary period, and people began to pay attention to inflation (Figure 19). Some economistsargue that significant inflation and deflation have long-term effects on inflation perceptions and expectations. For example, the inflation perceptions of those who experienced highinflation remain relatively high for a long period compared to those who did not experience such a situation. However, statistically speaking, the results of testing the relationshipbetween the cohort and inflation perceptions are mixed depending on the surveyed countries and periods; thus, it is too early to determine whether inflation perceptions are affected bycohorts.
4. The Relationship between Inflation Expectations and Inflation Perceptions
As pointed out in Section 2, studies on inflation perceptions have recently become popular because inflation perceptions have been increasingly considered the basis of inflationexpectations. This approach has the advantage of explaining households’ inflation expectations better than the traditional approach.25 The evidence of a close relationship betweeninflation perceptions and expectations is outlined below.First, consumers’ socioeconomic attributes affect both inflation expectations and perceptions similarly. For example, severalsocioeconomic attributes affect inflation perceptions. First, women’s perceptions of inflation were generally higher than those of men. Second, the higher the educational background,the lower is the perception of inflation. Third, people with higher incomes have lower inflation perceptions. Fourth, inflation perception decreases as financial literacy improves.Asurvey by the Federal Reserve Bank of Cleveland in the U.S. yielded several results. First, men’s inflation perception was 4.6 percent from August 1998 to November 2001, whereasthat of women was 6.9 percent. The inflation expectations of men and women were 4.0 percent and 6.4 percent, respectively. Second, as noted in the December Study, householdsrefer to not only inflation perceptions but also other macroeconomic information available to them. This is supported by the time difference correlation analysis showing inflationexpectations, which contains information about the future outlook in addition to inflation perceptions. When households are divided into five income levels, and the lowest-incomegroup responded with inflation perceptions of 9.2 percent, whereas for those in the highest-income group, the percentage was 4.7. In terms of inflation expectations, the lowest-incomegroup responded with 8.4 percent, whereas the highest-income group responded with 4.4 percent. Third, the inflation perception of university graduates was 4.8 percent, and that ofthose whose education level was high school graduate or below was 7.5 percent. Similarly, the inflation expectation of those who had graduated from university was 4.4 percent,whereas that of people who had graduated from high school or less was 6.6 percent.The mechanism behind the relationship between inflation expectations/perceptions andsocioeconomic attrition, especially gender differences, has been widely studied, but no convincing explanations have yet been offered. Influential hypotheses include that, first,because women are still more often engaged in daily shopping than men, women are more aware of changes in the price movements of frequently purchased goods, as described insubsections 3.3. and 3.4. Second, because the portion of women still not engaged in jobs is greater than that of men, they are not interested in inflation development and thus become“worse performers” more often than men. Moreover, household survey results have shown that consumers rely heavily on inflation perceptions when responding to questions oninflation expectations. Figure 20 is a result of the Bank of Japan’s “Opinion Survey” asking information sources of inflation perceptions and expectations. Compared to the responsesfor inflation perceptions, answers for five-years-ahead inflation expectations are tilted toward macroeconomic variables affecting future inflation development, such as exchange rates,stock/land prices, and monetary policy. This tendency is consistent with the results of a December Study, which applied time-difference correlation analysis and found that inflationexpectations contain unique information not included in inflation perceptions. However, more than half of responses were assigned to “backward-looking factors,” implying thathouseholds rely heavily on inflation perceptions (backwardlooking factors) when forming inflation expectations (Figure 21). Similarly, a survey conducted by the Bank of Englandshows that households rely heavily on backward-looking information when forming inflation expectations (Figure 22). The survey compares the important factors in determininginflation expectations for one and five years ahead. In the figure, long-term inflation expectations were determined by relatively more forward-looking factors, such as “expectations offuture economic strength,” “level of interest rates,” and “discussion in the mass media,” than short-term Third, there is a strong correlation between inflation expectations andperceptions when using individual household data. Figure 24 shows an example of an analysis conducted by the Federal Reserve Board. Regrettably, as yet, no individual data havebeen disclosed in Japan. These figures plot the correlation between inflation perceptions on the horizontal axis and inflation expectations on the vertical axis. The figure on the left sideindicates the long-term relationship (5?10 years), and the right side shows the short-term (1 year) relationship. The green line is the result of a linear regression, limiting the X variablesfrom -10 to +20 percent. Another study by the Central Bank of Germany calculated the pass-through from inflation perceptions to expectations.29 A 1 percent increase in inflationperceptions increased inflation expectations by 0.74 percent in the long term and 0.53 percent in the short term. In addition, the pass-through was higher when the actual inflation ratewas lower. A similar statistical analysis was conducted by the Bank of England, which reported a 0.4 percentage point pass-through in five-years-ahead expectations.
5. Implication for Monetary Policy Implementations
As inflation perceptions significantly affect the formation of inflation expectations, they have important implications for monetary policy implementation.
5.1. Demerits of Focusing Only on Core CPI as a Policy Indicator
When discussing monetary policy implementation, often-used price indicators are the “core-core CPI,” which exclude food and gas prices to reduce short-term fluctuation of theoverall CPI. Although it is of the utmost importance to understand correctly the fundamental movement of prices, we should also bear in mind that the “non-core CPI”consisting offood and gas prices significantly affect inflation perceptions.Therefore, both core CPI and non-core CPI should be included when monitoring price development (Figure 25).Additionally, the development of a new type of price index is required. The weight of this index should reflect the purchasing frequency and consumers’ loss aversion (asymmetricweights).
5.2. Macroeconomic Effect of Inflation Perceptions/Expectations
As discussed in section 4, inflation expectations differ significantly depending on socioeconomic attributes such as gender, school career, and income level. For example, women havehigher inflation expectations than men. Because nominal interest rates are equal for both sexes, real interest rates are lower for women than for men. This causes distortions inhousehold behavior, such as expenditure, savings, and investment. In addition, inflation expectations significantly affect wage negotiations. This, in turn, affects the efficiency ofmacroeconomic resource allocation. The central bank of Germany’s latest study stresses the importance of policymakers paying more attention to analyzing and monitoring inflationperceptions. In Japan too, further effort should be devoted to analyzing households’ inflation expectations and perceptions and their application to monetary policy decisions.
5.3. The Central Bank’s Actions to Guide Household Inflation Expectations
Central banks should not only monitor households’ inflation expectations and perceptions but also promote the proper formation of inflation expectations by following practices toimprove the efficiency of monetary policy implementation.First, central bank communication with the general public account for the fact that inflation expectations differ significantlydepending on socioeconomic attributes, such as gender and income level. Second, to reduce the upward bias of inflation expectations compared with the CPI, central banks need topromote financial education for households so that they can actively gather and understand macroeconomic information, including the CPI. The background factor for such efforts isthat the higher the financial literacy, the lower are the inflation expectations. In other words, as the previously mentioned study by the ECB stressed, improving financial literacy wouldincrease the number of “better performers” and decrease that of “worse performers.” Other studies have highlighted similar views.
6. Conclusion
This study analyzed the formation process of inflation perceptions and discussed the relationship between inflation expectations and perceptions as well as the implications formonetary policy. Research on inflation perceptions is scarce because household inflation perceptions constantly exceed the CPI at a significant level. This attitude gradually changedas it was considered to form the basis for inflation expectations. At the same time, the following reasons explain the divergence between CPI and inflation perceptions. First,households lack an understanding of CPI statistics. They rely mostly on the price development of frequently purchased goods when responding to inflation perceptions. In fact, aprice index composed of food and gas prices fits inflation perceptions more closely than does the overall CPI.Second, consumers tend to be loss averse. They are 2.5 times moresensitive to the same amount of loss compared to gain. This sensitivity increases inflation perceptions, particularly when large portions of goods and services are experiencing risingprices. Third, the CPI adjusts the quality improvements for each item by lowering the price index. Households, however, only refer to retail prices, which do not reflect quality changes; thus, the divergence between the CPI and inflation perceptions widens.The following evidence emerged regarding the relationship between inflation expectations andperceptions. First, differences in socioeconomic attributes, such as gender, income level, and educational background, affect inflation expectations and perceptions in the samemanner. Second, according to household surveys, households rely heavily on inflation perceptions when responding to inflation expectations. Third, a crosssectional analysis usingindividual samples showed a strong correlation between inflation expectations and perceptions. The significant role played by inflation perceptions in forming expectations hasimplications for monetary policy implementation. First, food, and gas prices, which are excluded from the “core-core” CPI, are important components of price movements whenassessing households’ inflation perceptions. Second, socioeconomic attributes affect inflation expectations and thus distort real interest rates and efficient resource allocation. Third,households’ financial literacy should be improved through financial education to encourage the formation of inflation perceptions in response to macroeconomic situations.Asresearch on inflation perceptions is scarce, further studies are required to accumulate survey data on the formation mechanism of inflation perceptions and their relationship withinflation expectations. The urgent task is to disclose detailed survey results, such as individual sample data, data sorted by socioeconomic attributes, and distributions of inflationperceptions and expectations.
(References)
Bank of Canada (2020) “Perceived Inflation and Reality: Understanding the Difference” Remarks by Deputy Governor L. Schembri
Bank of Italy (2008) “What’s Behind “Inflation Perceptions? A Survey-Based Analysis of Italian Consumers,” P. D. Giovane, S. Fabiani and R. Sabbatine, Working papers No.655,January 2008
Bank of Japan (2022) “Households’ Perceived Inflation and CPI Inflation: The Case of Japan,” Y. Takahashi, Y. Tamanyu, No.22 E-1, March 2022
--- (2023a) “Outlook of Economic Activity and Prices,” October 2023
--- (2013), (2023b) “Opinion Survey”
BOE (2016) “How Are Households’ Inflation Expectations Formed?” Quarterly Bulletin 2016 Q2, J. Rowe
--- (2019) “Understanding Inflation: Expectation and Reality,” Silvania Tenreyro, July 10, 2019
Bruine de Bruin W et al. (2013) “Measuring Inflation Expectations,” Annual Review of Economics, 2013
CCFSI (2022) “Financial Literacy Survey,” Central Council for Financial Services and Information
Central Bank of Ireland (2022) “What Drives Consumers’ Inflation Perceptions in the Euro Area?” Economic Letter Vol. 2022, No.6
D’Ancuto et al. (2019) “Exposure to Daily Price Changes and Inflation Expectations,” E. D’Ancuto, U. Malmendier, J. Ospia, M. Weber, NBER Working Paper 26273, September2019
Demarks National Bank (2019) “Revisiting the Inflation Perception Conundrum,” K. Ablidgen, A. Kuchier
Deuche Bundes Bank (2023) “The Pass-through from Inflation Perceptions into Inflation Expectation,” Discussion Paper No17/2023, S. Huber, D. Minira, T. Schmidt
ECB (2017) “EU Consumers’ Quantitative Inflation Perceptions and Expectation: An Evaluation,” R. Arioli, C. Bates, H. Dieder, I. Duca, R. Friz, C. Gayer, G.
Kenny, A. Meyler, I. Pavlova, ECB Occasional Paper Series No.187, April 2017
Ehrmann et al. (2017) “Consumers’ Attitudes and Their Inflation Expectations,” M. Ehrmann, D. Pfajfar, E. Stantoro, International Journal of Central Banking February 2017
ESRI (2013) “Disagreement and Biases in Inflation Expectations of Japanese Households,” Y.Ueno and R. Namba, Discussion Paper Series No.300, Economic and Social ResearchInstitute, Cabinet Office of Japan,
--- (2022) ”Micro-shock and Macroeconomic Forecast,” Y.Iizuka, J. Kikuchi, Y. Nakazono, “Keizai-Bunseki,” No. 204 (Japanese only)
FRB (2001) “The Demographics of Inflation Opinion Surveys,” Federal Reserve Bank of Cleveland, October 2001
--- (2016) ” Inflation Perception and Inflation Expectations,” FEDS Notes 2016/12, A. Detmeister, D. Lebow, E. Peneva
--- (2018) “Perceptions and Expectations of Inflation by U.S. Households,” Finance and Economics Discussion Series 2018-073, A. Axerlrod, D. Lebow, E. Peneva
FRB of Cleveland (2001) “The Demographic of Inflation Opinion Surveys,” M. Brayan, G. Venkatu, October 15, 2001
Georgans et al. (2014) “Frequency Bias in Consumers’ Perception of Inflation: An Experimental Study,” European Economic Review 67, S. Georgans, P. Healy, N. Li
Hayo and Neumeier (2022) “Households’ Inflation Perceptions and Expectations: Survey Evidence from New Zealand,” Bearnd Hayo and Florian Neumeier,
International Economics and Economic Policy (2022)
NBER (2020) “Gender Roles and the Gender Expectation Gap,” NBER 26837, March2020
Nihon-Keizai-Shinbun (2023) “Price Development Analysis Needs to Consider Degree of Frequent Purchasing Behavior: Inflation Perspective and the Monetary Policy,” S. Abe andN. Inekura, August 25, 2023 (Available in Japanese only)
Shimizu (2008) “Quantitative Analysis on Recent Inflation Perceptions,” M. Shimizu,”
Instability of Japanese Household Inflation Expectations During the Current Inflationary Phase
This study analyzed Japanese household inflation expectations during the current inflationary phase and concluded that they are relatively unstable compared with those of Europeanand U.S. households. Japanese household inflation expectations show certain unique features. First, they consistently move above the Consumer Price Index (CPI) and closely followthe movement of CPI, especially in the case of one-year-ahead expectations. Second, during the current inflationary phase, inflation expectations suddenly became unstable, since theyrose to +10percent for one-year-ahead expectations and +5 percent for five-years-ahead expectations. Both expectations exceeded the actual CPI at the time. Third, although the peaklevel of CPI was lower in Japan than in the U.S. and European countries,Japanese inflation expectations were significantly higher than other countries during the current inflationaryphase. The contrast is especially remarkable in the case of five-years-ahead expectations, where those of the U.S. and European countries were barely affected by the rises in CPI.Further analysis on the stability of Japanese inflation expectations in different economic circumstances is required, such as an examination of the mechanisms behind householdinflation perceptions, since these are closely related to inflation expectations.
1. Introduction
The Japanese economy has experienced a period of long-term low inflation. During this period, the Consumer Price Index (CPI), excluding perishables and the effects of increases inconsumption tax, has hovered around the minus 2 to plus 2 percent range. This low-inflation phase abruptly ended at the beginning of 2021 as the corporate sector started to pass onthe jump in import goods prices to the prices of their own products. In January 2023, CPI peaked at + 4.1 percent on a year-on-year basis and remained high at +2.8 percent as ofOctober 2023 (Figure 1). The CPI is projected to remain above +2 percent on a year-on-year basis, reflecting the rise in the corporate sector’s cost-push price rises and surges incrude oil prices. The Bank of Japan projects CPI to remain at a relatively high rate of +2.8 percent in both the fiscal years 2023 and 2024.Such increases in the inflation rate havecaused various economic problems in Japan, while providing a unique opportunity to analyze household inflation expectations. This study analyzed Japanese household inflationexpectations and compared them with those of European and U.S. households during the current inflationary phase. The results of our analysis show that Japanese householdexpectations of inflation are unstable. This conclusion is supported by the following observations: First, before the current inflationary phase, they moved around the +2 to +3 percentrange, which was higher but more stable than CPI. In addition, the five-years-ahead expectations were more stable than the one-year-ahead expectations. Second, during the currentinflationary phase, inflation expectations suddenly became unstable as they rose to +10 percent for one-year-ahead expectations and +5 percent for five-years-ahead expectations,much higher than the actual CPIat that time. Third, compared with the U.S. and European countries, Japanese expectations became much more unstable than their expectations,especially for the five-years-ahead inflation expectations. This contrast is astonishing, considering that the peak levels of the U.S. and European CPI (+8 to 10 percent) were muchhigher than those of Japan. The remainder of this paper is organized as follows: Section 2 focuses on inflation expectations, which have consistently moved above CPI and suddenlybecame unstable during the current inflationary phase. Section 3 demonstrates that inflation expectations in Japan are more unstable than in the U.S. and European areas. Section 4tests statistically whether household inflation expectations contain forward-looking information. Section 5 concludes the paper.
2. Household Inflation Perceptions
The Bank of Japan conducts the “Opinion Survey” on a quarterly basis, and it includes household inflation perceptions and expectations. The survey asks about householdperceptionsof inflation in two ways. One is by choosing from five categories. The other asks for direct changes in the price level compared to the previous year. For example, from theSeptember 2023 survey, 68.4 percent of households responded that “present price levels have gone up significantly” and the median value3 of “changes in price levels compared withone year ago” was +10 percent. This value is significantly higher than the actual CPI at the time, which was+2.8 percent. Figure 2 shows the time-series relationship between changesin CPI and inflation perceptions. It exhibits several distinctive characteristics. First, inflation perceptions have been consistently higher than changes in CPI. Second, inflationperceptions closely follow CPI movements. Third, the gap between the two variables widens when CPI increases, including during the current inflationary phase. Fourth, inflationperceptions remain at zero or above, even when CPI falls into negative territories.The first feature of inflation perceptions mentioned above, that is, inflation perceptions have beenconsistently higher than the changes in CPI, caused economists inside and out of Japan to question the reliability of inflation perceptions. Therefore, only small number of studies havebeen conducted in this field. In fact, a paper published by the Bank of Japan in 2022 pointed out that although the reasons behind the gap between CPI and inflation perceptions havenot become apparent, only a few studies have tried to explain the cause of the gap. Figures 3 and 4 show that inflation perceptions have been higher than the actual price index, notonly in Japan, but also in the U.K. and Eurozone.
Both U.K. and Eurozone households share the same features asthose of Japan. First, inflation perceptions constantly exceed CPI (HICP), but closely follow CPI (HICP). Second, inthe Eurozone, the gap between inflation perceptions and HICP widens during inflationary periods. Third, inflation perceptions remain at zero percent or higher even when CPI (HICP)falls into negative territory. The first point mentioned above, where inflation perceptions constantly exceed CPI, is often called the “inflation perception conundrum” and has beenobserved in many developed countries, including Sweden, Italy, and Canada. However, research activities to analyze inflation perceptions are scarce, as studies in this field only beganin the 2000s.
3. Inflation Expectations in Japanese Households Are Unstable Compared to Those of the U.S. and European Countries
3.1. Japanese Household Inflation Expectations
While research on inflation perceptions is relatively scarce, inflation expectations have attracted the attention of academics and central bankers. This marked contrast arises becausemovements in inflation expectations greatly influence aggregate demand, such as personal consumption and corporate fixed investments, through changes in real interest rates. Realinterest rates are defined as nominal interest rates minus expected inflation. If this rises, the activities of households and the corporate sector fall, and vice versa. Inflation expectationshave attracted academic attention since the 1980s, and particularly since the global financial crisis. This is because monetary policies in major developed countries maintained nominalinterest rates at zero or in the negative range for a substantial period, and real interest became closely linked to movements in inflation expectations.Since inflation expectations are notdirectly observable variables, professional forecasts of inflation expectations, such as surveys of economists and people engaged in financial markets or inflation expectations derivedfrom yields on inflation-linked bonds, are often used as alternative variables (Figure 5). As of October 2023, these alternative figures ranged between 1 and 1.5 percent. When theinflation expectations are used as a variable in macroeconomic models, they often assume rational expectations (people act according to the assumption that they know all theinformation about the macroeconomic variables) or adaptive expectations (people rely on past inflation rates as a proxy for future inflation rates). As for inflation expectations ofJapanese households, the Bank of Japan’s “Opinion Survey”is conducted on a quarterly basis. The latest survey results (September 2023) show that the most common choice amongfive categories for one-year-ahead inflation expectations was “prices will go up significantly,” and that the median value was +10 percent, which was much higher than the Bank’sprojection of +2.8% for fiscal 2023. In addition, the survey indicated that the median value for five-years-ahead inflation expectations was +5 percent, which was much higher than theprofessional economists’ forecast of +1 to 2 percent. These results have led economists to conclude that the household inflation expectations were too high to “justify” or“impractical to use in economic analysis.”Household inflation expectations were surveyed using the Cabinet Office’s “Consumer Confidence Survey.” It asked about respondents’inflation expectations by offering choices from seven categories (Figure 6-2). The distribution of inflation expectations is similar to that found in the Bank of Japan’s survey. The mostcommon choice was that “prices will go up greater or equal to five percent.” The survey does not directly ask for the numerical values of inflation expectations. However, calculating aweighted mean yielded +4.7 percent, which was lower than that of the Bank of Japan’s survey, but much higher than the professional survey results. The Bank of Japan’s “OpinionSurvey” asks inflation expectations by choosing from five categories. Based on the distributions of the responses, it is possible to calculate a numerical value of inflation expectationunder a certain assumption. However, expectation values generated by such statical procedures are quite often much lower than the value from asking directly about inflationexpectations. This inconsistency suggest that statistical procedures or assumptions need further review. Comparing Figure 7 with Figure 2, movements in inflation expectations aremore stable than those in inflation perceptions. In addition, both variables exceed the CPI increase rates in almost all observation periods. The features of one-year-ahead inflationexpectations are as follows: First, inflation expectations did not increase with inflation perceptions during the 2008 inflation period. Second, before the current inflationary period,inflation expectations were stable, ranging between approximately 2 to 3 percent. Third, during the current inflationary phase, inflation expectations increased 10 percent, similar toinflation perceptions.Next, the five-years-ahead inflation expectations also constantly exceed the CPI increasing rate, except during the consumption tax increase period. Before thecurrent inflationary phase, five-year expectations moved moderately by approximately 2 percent and were more stable than one-year-ahead inflation expectations. However, during thecurrent inflationary phase, five-year expectations increased swiftly to the 5 percent level and exceeded actual CPI. This contradicts the basic feature of five-years-ahead inflationexpectations, which should reflect the long-run average of inflation rates rather than short-term fluctuations in actual inflation. In other words, Japanese household inflation expectationshave not been anchored firmly.
3.2. Comparing Japanese Household Inflation Expectations with Those of U.S. and European Households
In this subsection, the inflation expectations of Japanese households are compared with those of U.S. and European households. Figure 8 shows one- and five-years-ahead inflationexpectations of U.K. households. Even in the U.K., inflation expectations consistently exceed CPI. However, in contrast to Japanese households, five-years-ahead expectations exhibitonly minor responses to changes in the CPI. In addition, the peak level of one-year-ahead expectations was approximately 4 percent, which was about half of the CPI increase. SinceU.K. CPI rose to about 10 percent year-on-year, which was about double that of Japan, the inflation expectations of U.K. households were much more stable than those of Japanesehouseholds. In Figure 7, inflation expectations in 2014 were lower than the actual CPI. This is due to the effect of a consumption tax rate increase at the time. Two different surveysare available regarding the inflation expectations of Eurozone households. A survey conducted by the European Central Bank (ECB) asked for both one- and ive-years-ahead inflationexpectations, although the survey only went back as far as 2020 (Figure 9). Another survey conducted by the European Commission offers a long observation period; however, itlacks long-term inflation expectations (Figure 10). Bearing in mind such statistical limitations, the ECB survey during the current inflationary phase showed that, although HICP movedup to about 10 percent, the increase in the one-year-ahead expectations was much more modest, and their peak level was only about 4 percent. In addition, five-year expectations didnot seem to be significantly influenced by the surge in HICP. Both one- and five-years expectations are much more stable than inflation perceptions, jumping to about 25 percent at thepeak. Figure 10 shows a survey conducted by the European Committee. The one-year-ahead inflation expectations go back to 2004. As with the ECB survey, inflation expectationsconsistently exceed CPI and follow it more closely than inflation perceptions. However, in the current inflationary phase, inflation expectations from the two surveys divergesignificantly. According to by D’Ancuto et al. (2019), until recently, the ECB was reluctant to publish household inflation expectations, since the high expectation results might causedamage to the reliability of the central bank.For example, the peak level of inflation expectations in the ECB survey was 5 percent, whereas that in the European Committee survey was12 percent. Although it is well known that differences in survey methods, such as how questionnaires and answers are provided, affect survey results significantly, the gap between thetwo surveys is too large to justify such factors alone.Regarding U.S. household inflation expectations, surveys conducted by the Federal Reserve Bank of New York (FRBNY) andMichigan University are considered reliable. The FRBNYsurvey, shown in Figure 11, has the following characteristics. First, before the current inflationary phase, even one-year-aheadinflation expectations were quite stable at around 3 percent. Second during the same period, the levels of one- and three-year-ahead inflation expectations were remarkably close to oneanother. Third, during the current inflationary phase, three-year-ahead expectations swiftly returned to a 3 percent level in the latter half of 2023, while one-year-ahead expectations roseto 6 percent. Overall, U.S. household inflation expectations seem to be much more stable than those of Japan. Figure 12 illustrates the results of the Michigan University survey. As thesurvey goes back to 1978 on a monthly basis, it covers the famous “great inflation” period of the latter half of 1970s. During this period, the inflation expectations of U.S. householdsincreased to an astonishingly high level of about 10 percent, while actual CPI rose to about 15 percent at the time. Then, inflation subdued in the first half of the 1980s, as did inflationexpectations, which hovered around 2-3 percent. The exceptions were the pre-Great Recession period and the current inflationary phase. Comparing these two periods, the peak levelof inflation expectations was approximately the same, whereas the actual CPI increased more during the current phase. This implies that inflation expectations were more stable duringthe current period. The differences in survey methods contributed to the fact that peak inflation expectation levels during the current inflation phase were lower than those of theFRBNY. A comparison of the inflation expectations in the three countries and regions strongly suggests that Japanese household inflation expectations show much lower stability thanthose in the other two countries and regions. This conclusion follows from the following observations during the current inflationary phase. First, it was only in Japan and the Eurozone(and only by the ECB survey) that one-year inflation expectations exceeded the actual CPI. Second, only the Japanese long-term inflation expectations overshot changes in CPI. Third,the peak level of longterm expectations in Japan jumped up to 5 percent during the current inflationary phase, which was higher than in the other two countries/regions, where CPI rosesignificantly more than in Japan.
4. Forward-looking Information Contained in Household Inflation Expectations
Recent studies suggest that inflation perceptions are the key determinants of inflation expectations. The Infotainment Research Center conducted another study to examine therelationship between these two factors and the factors determining inflation perceptions. This section examines whether inflation expectations contain forward-looking information byfollowing the ECB’s analysis. The statistical method employed by the ECB calculates the time difference correlations between the CPI and inflation perceptions/expectations andidentifies the peak correlation period for each variable. The ECB’s analysis indicated that, in the Eurozone, although the inflation perceptions lagged behind the movement in CPI bythree months, the lag in inflation expectations was only one month. Judging from the result, the ECB study concludes that inflation expectations contain “limited but useful forward-looking information.”The results of applying the ECB method to the Japanese data are shown in Figure 13. The peak of the inflation expectations correlation had no lag against theCPI, while inflation perceptions lagged behind CPI by three months. This result indicates that similar to the ECB analysis, Japanese household inflation expectations contain limited butuseful forward-looking
information.
5. Conclusion
This study analyzed Japanese household inflation perceptions and expectations and compared them with those of European and U.S. households during the current inflationary phase.From the observations below, we conclude that Japanese household inflation expectations are unstable. First, during the current inflationary phase, inflation expectations jumped to alevel well beyond the CPI’s increasing rate. In particular, both one-year-ahead and five-years-ahead inflation expectations exceeded the CPI rate, diverging significantly from theprevious trend.Second, comparing such developments with those of the U.S. and the Eurozone, although the peak increasing rates of the CPI were higher in those countries, theirinflation expectations were more stable than those of Japan. The contrast is more distinctive in the case of five-yearsahead expectations, as their expectations are barely affected by theCPI hike.Further analysis of the stability of inflation expectations under different economic circumstances and comparisons among countries is required. One area to examine is themechanism behind household inflation perceptions, as inflation expectations seem to be significantly affected by inflation perception movements. Inflation perceptions appear to havesignificantly affected heightened inflation expectations during the current inflation phase. As previously mentioned, the Infotainment Research Center prepared a new study based onthis view.
(References)
Bank of Japan (2008) “About the inflation expectations” BOJ review series 2008 J-15, T. Sekine, K. Yoshimura, C. Wada, December 2008, (available only in Japanese)
--- (2022) “Households’ perceived inflation and CPI inflation: A case of Japan” Y. Takahashi, Y. Tamanyu, Bank of Japan WP series No.22-J-2
--- (2023a) “Outlook of Economic Activity and Prices” October 2023, Bank of Japan
--- (2023b) “Opinion Survey” September 2023, Bank of Japan
Cabinet Office (2013) “Disagreement and Biases in Inflation Expectations of Japanese Households,” ESRI Discussion Paper Series No.300、Y. Ueno, R. Namba
Central Bank of Ireland (2022) “What Drives Consumers’ Inflation Perceptions in the Euro Area?”Economic Letter Vol. 2022, No.6
Consumer Affairs Agency (2019) “An Analysis of inflation Expectation Using “Price Monitoring Survey in Japan” May 2019
D’Ancuto et al. (2019) “Exposure to Daily Price Changes and Inflation Expectations,” E. D’Ancuto, U. Malmendier, J. Ospia, M. Weber, NBER Working Paper 26273, September2019
Demarks National Bank (2019) “Revisiting the Inflation Perception Conundrum,” K. Ablidgen, A. Kuchier
ECB (2016) “EU Consumers’ quantitative inflation perceptions and expectations: An Evaluation” Discussion Paper 38, November 2016
--- (2021) “Making Sense of Consumers’ Inflation Perceptions and Expectations?The Role of (un)certainty” A. Meyler, L. Reiche, ECB Economic Bulletin February 2021
FRB (2001) “The Demographics of Inflation Opinion Surveys” Federal Reserve Bank of Cleveland, October 2001
--- (2016) “Inflation Perceptions and Inflation Expectations” A. Detmister, D. Lebow, E. Peneva, FED Notes December 5, 2016
Statistics Bureau of Japan “Consumer Price Index"