Statistics

Home > Statistics > Outline of Statistics and Statistical Release Schedule > Explanations of Statistics > Explanation of TANKAN > FAQs (Frequently Asked Questions) on the Tankan (Short-Term Economic Survey of Enterprises in Japan)

FAQs (Frequently Asked Questions) on the Tankan (Short-Term Economic Survey of Enterprises in Japan)

February 2019
Bank of Japan
Research and Statistics Department

Contents

List of Questions

1. Q&A regarding the Tankan in general

2. Q&A regarding the judgment survey

3. Q&A regarding the quantitative data

4. Q&A regarding "inflation outlook of enterprises"

5. Q&A regarding the statistical accuracy

6. Q&A regarding the publication of the Tankan

7. Q&A regarding other topics

List of Answers

1. Q&A regarding the Tankan in general

1-1. What is the purpose and what are the survey items of the Tankan?

The Tankan is a nationwide business survey conducted on a quarterly basis, with the aim of providing an accurate picture of business trends of enterprises in Japan, thereby contributing to the appropriate implementation of monetary policy. The Tankan asks various questions about overall corporate activities, including those regarding the enterprise's judgment, such as regarding business conditions both at the time of the survey and three months later, as well as actual results and forecasts of financial statements, and inflation. The Tankan is an abbreviation of "Tanki Keizai Kansoku Chousa (Short-Term Economic Survey of Enterprises in Japan)," and is well-known all over the world.

1-2.Does the Tankan have any legal basis? Is there a legal obligation to respond?

The Tankan is a statistical survey conducted by the Bank of Japan, in accordance with the Statistics Law (Law No. 53 of 2007). Although sample enterprises are not legally required to respond, the Tankan has the cooperation of the vast majority of surveyed enterprises, with almost all responding to each survey.

The Bank is obliged to keep confidential information obtained from respondents under strict security in accordance with the Statistics Law.

1-3. What information does the survey results of the Tankan provide?

With the aim of obtaining an accurate snapshot of overall corporate activities, in addition to quantitative data (annual figures) on items such as sales, fixed investment, and the number of new graduates hired, the Tankan survey also asks questions which draw on the judgment of the respondent, such as their view on their business environment. It also asks for forecasts of these items, such as forecast figures for quantitative data, and on the business conditions three months later for judgment items. The inclusion of both quantitative and judgment data, together with its long history, which provides long-term time series data of survey results, can be counted as strengths of the Tankan.

Utilizing these features, users can perform various analyses on economic conditions and corporate activities for their own various purposes, for example, by combining results of judgment items with those of quantitative data and comparing current conditions with similar ones in the past.

Consequently, the Tankan is used as one of the indicators for the assessment of economic conditions in Japan.

  • The graph of business conditions DI for large manufacturing enterprises and reference dates of business cycle. Compared with the peaks and bottoms of the business cycle, the business conditions DI for large manufacturing enterprises has almost accurately captured the turning points of the economy.
  • The graph of fixed investment in the Tankan of all enterprises in all industries including financial institutions and nominal private nonresidential investment in the National Accounts Statistics of Japan. Year-on-year rates of change of fixed investment in the Tankan is almost consistent with those of nominal private nonresidential investment in the National Accounts Statistics of Japan.

It should be noted that the Tankan aggregates figures from enterprise responses and this does not reflect the Bank's assessment on economic conditions and projections.

1-4. What is the coverage of the Tankan?

The target population of the Tankan is private enterprises (excluding financial institutions) in Japan with capital of 20 million yen or more. The survey population consists of approximately 220,000 enterprises, drawn from the Economic Census, conducted by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry. Sample enterprises (comprising approximately 10,000 enterprises) are selected from this survey population based on industry and size classifications to satisfy established criteria such as statistical accuracy.

Aggregated values are released by industry and size classifications (a total of 93 strata: 31 industries and three capital sizes). note To improve statistical accuracy, the Bank further subdivides enterprises into just less than 400 strata, for the selection of sample enterprises and to calculate estimates of the population total. The Bank revises sample enterprises according to updates of the survey population data. For details of the most recent revision, see "Regular Revision of the Tankan Sample Enterprises" (March 2018).

Financial institutions are also surveyed to supplement the Tankan (see 1-5.).

Note: In the Tankan, manufacturing is comprised of 17 industries and nonmanufacturing is comprised of 14 industries, based on the Japan Standard Industrial Classification released by the Ministry of Internal Affairs and Communications. Enterprise size is defined by amount of capital, in three categories -- large, medium-sized, and small enterprises -- as shown below.

Table : Capital
  Capital
Large enterprises 1 billion yen or more
Medium-sized enterprises 100 million yen to less than 1 billion yen
Small enterprises 20 million yen to less than 100 million yen

1-5. How are financial institutions treated in the Tankan?

Since the March 2004 survey, the Survey of Financial Institutions has been conducted to supplement the Tankan.note

The target population is private financial institutions in Japan with 10 or more employees. The survey population consists of approximately 3,000 institutions, drawn from the Economic Census, conducted by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry.

The survey population comprises the following seven sectors: (1) city banks and trust banks; (2) member banks of the Regional Banks Association of Japan and the Second Association of Regional Banks; (3) shinkin banks, etc.; (4) other financial institutions for small businesses; (5) financial products transaction dealers; (6) insurance companies; and (7) non-deposit money corporations. Sample institutions (comprising approximately 200 institutions) are selected from the survey population to satisfy established criteria such as statistical accuracy.

While the sample institutions are categorized into the seven sectors above, to ensure that stable classifications are released in the case of a reorganization across sectors, the aggregated values of financial institutions are released according to the following five classifications: (a) banks ([1] and [2]); (b) financial institutions for cooperative organizations ([3] and [4]); (c) financial products transaction dealers; (d) insurance companies; and (e) non-deposit money corporations. Financial institutions have no size classifications as with large, medium-sized, and small enterprises. As is the case with non-financial enterprises, the Bank of Japan further subdivides financial institutions into approximately 20 strata, for the selection of samples and to calculate estimates of the population total, to improve statistical accuracy. Sample institutions are revised when sample enterprises of the Tankan are revised. For details of the most recent revision, see "Regular Revision of the Tankan Sample Enterprises" (March 2018).

Note: The survey on fixed investment of financial institutions was previously conducted to supplement the Principal Enterprises Tankan from the November 1989 survey (corresponding to the present December survey) until the December 2003 survey. The Survey of Financial Institutions expanded its coverage and survey items in the March 2004 survey, and has been conducted to supplement the Tankan.

1-6. How is the Tankan survey conducted?

The survey is conducted using a paper form or an online form. The paper form or an identification code for the online survey is provided by mail to all sample enterprises. Sample enterprises are required to choose to complete either the paper survey or online survey in advance (it can be changed for each survey).

The online survey started from the March 2011 survey. The share of respondents who choose to complete the online survey is approximately 40 percent as of the June 2018 survey.

1-7. What is the history of the Tankan?

Numerous business surveys are now conducted in Japan and overseas, and the Tankan is one of the oldest. The first business survey in Japan was the Short-Term Survey of the Industries conducted by the former Industrial Bank of Japan, which took the Economic Test -- carried out by the IFO Institute for Economic Research in the former West Germany -- as a model and started in 1951. The Bank took over and revised this survey, and started the Principal Enterprises Tankan in 1957. The Bank then launched the All Enterprises Tankan in 1974, which was substantially revised in 2004 with the termination of the Principal Enterprises Tankan. Throughout this long history the Bank has revised sample enterprises and survey items to adequately reflect actual economic conditions in Japan.

A Brief Chronology of the Tankan
Aug. 1957
The Principal Enterprises Tankan started. The survey, which covered 524 enterprisesnote -- 346 for manufacturing and 178 for nonmanufacturing -- was conducted quarterly.
May 1960
The Business Conditions Survey of Small Enterprises started, which covered 2,666 small manufacturing enterprises and was conducted semiannually.
May 1962
The Business Conditions Survey of Small Enterprises was changed from semiannual to a quarterly basis.
Nov. 1966
The survey title was changed from the Business Conditions Survey of Small Enterprises to the Small Enterprises Tankan.
May 1974
The Small Enterprises Tankan expanded its coverage to include large and medium-sized enterprises in manufacturing and all enterprises in nonmanufacturing, and the survey was titled as the All Enterprises Tankan. The number of sample enterprises was 5,596 (4,243 in manufacturing and 1,353 in nonmanufacturing).
May 1983
The number of samples for the All Enterprises Tankan substantially increased from 5,181 to 7,035 enterprises, mainly in nonmanufacturing (from 1,287 to 3,141).
Nov. 1989
The survey on fixed investment of financial institutions (207 enterprises) started to supplement the Principal Enterprises Tankan.
Nov. 1993
The number of samples for the All Enterprises Tankan substantially increased from 7,376 to 10,011 enterprises, mainly in nonmanufacturing (from 3,386 to 5,822).
Mar. 1997
Survey items of the Principal Enterprises Tankan were integrated into those of the All Enterprises Tankan, causing the number of survey items of the Principal Enterprises Tankan to decrease by half.
Mar. 1999
The number of samples for the All Enterprises Tankan increased from 9,129 to 9,433 enterprises, and release data was expanded. The All Enterprises Tankan was positioned as our main survey, while the Principal Enterprises Tankan was converted to a reference survey.
Mar. 2001
Software investment was added to survey items.
Mar. 2004
In the All Enterprises Tankan, which corresponds to the current Tankan, capital size was adopted for categorizing enterprises into large, medium-sized, and small enterprises instead of the number of employees; the number of samples increased from 8,204 to 10,562 enterprises; and survey items were revised. The Principal Enterprises Tankan was terminated and the Survey of Financial Institutions was intended to be a supplement to the Tankan.
Mar. 2007
Regular revision of sample enterprises of the Tankan (from 9,789 to 11,026 enterprises).
Mar. 2010
Regular revision of sample enterprises of the Tankan (from 10,116 to 11,684 enterprises).
Mar. 2014
Revisions of survey items of the Tankan, including the abolition of some survey items and the introduction of a new survey category regarding the inflation outlook.
Mar. 2015
Regular revision of sample enterprises of the Tankan (from 10,312 to 11,126 enterprises).
Mar. 2017
Research and development (R&D) investment was added to survey items.
Mar. 2018
Regular revision of sample enterprises of the Tankan including the revision of the sample design. The number of sample enterprises of the Tankan was changed from 10,645 to 10,020, and that of the Survey of Financial Institutions was changed from 196 to 210. With the revision of the sample design, the criteria for forming strata for selecting sample enterprises and estimating the population total were changed as follows: in the Tankan, sales was adopted instead of number of employees, as used previously; and in the Survey of Financial Institutions, number of employees was adopted instead of total assets, as used previously.

Note: In principle, the Tankan survey covered major companies generally reflecting business trends in each industry among listed companies with capital of 100 million yen or more (excluding financial and insurance companies).

1-8. What is the difference between the Tankan released by the Research and Statistics Department in the Head Office of the Bank of Japan and the Local Branch Tankan released by the Bank's branches?

In addition to the Tankan (released by the Research and Statistics Department in the Bank's Head Office) on a nationwide basis, the Bank's branches also compile and release the Local Branch Tankan, which covers enterprises within each region. Released materials are available on the branch websites (available only in Japanese).

There are two main differences between the Tankan and the Local Branch Tankan:

(1) Coverage

The sample design of the Tankan is based on industry and size classifications on a nationwide basis and therefore does not take industrial structure of each region into consideration. With these facts in mind, the Local Branch Tankan sometimes includes enterprises that the Tankan does not cover, such as local offices of large enterprises.

(2) Calculation method

For aggregation of quantitative data, while the Tankan calculates estimates of the population total, the Local Branch Tankan computes simply aggregated figures.

When using survey results of the Local Branch Tankan, these differences should be noted, and that it is deemed appropriate to use these results as a reference to understand economic conditions of each region. Taking these differences into account, it is also worth noting that a close comparison between levels of diffusion indexes (DIs) or figures published by each branch is not considered appropriate. The same attention should be paid when comparing the Tankan and the Local Branch Tankan.

2. Q&A regarding the judgment survey

2-1. How is the diffusion index (DI) calculated?

For judgment survey items, enterprises choose from three alternatives for each item and their responses are aggregated into the DI as shown below:

DI (percentage points) = Percentage share of the number of respondents choosing Alternative One - Percentage share of the number of respondents choosing Alternative Three

For example, the "Business Conditions DI," which indicates respondents' judgment of general business conditions primarily in light of their profits, asks them to choose the most appropriate answer among three alternatives: (1) Favorable; (2) Not so favorable; and (3) Unfavorable. Based on the responses, the "Business Conditions DI" is calculated by subtracting the percentage share of the number of respondents choosing (3) Unfavorable from that of the number of respondents choosing (1) Favorable.

Specifically, when 30 percent of sample enterprises respond (1) Favorable, 60 percent respond (2) Not so favorable, and 10 percent respond (3) Unfavorable, the "Business Conditions DI" is 20 percentage points (30 percent minus 10 percent).

The degree of change in DIs may sometimes appear inconsistent among release classifications. The following example shows that while the "Business conditions DI" for all industries does not change from the actual result to forecast, that for manufacturing and nonmanufacturing, which form the breakdown of all industries, increases by two points and one point, respectively.

This is because, for calculating DIs, percentage shares of the number of respondents choosing (1) Favorable and (3) Unfavorable are rounded to an integer. This rounding-off is conducted for each classification of manufacturing, nonmanufacturing, and all industries, and consequently aggregated figures are subject to rounding error.

Example: Business Conditions DI (% points)
Table : Business conditions DI (% points)
  Actual Result Forecast Change
(Forecast minus Actual Result)
Manufacturing 20
(30% - 10%)
22
(28% - 6%)
+2
Nonmanufacturing 18
(25% - 7%)
19
(24% - 5%)
+1
All Industries 20
(28% - 8%)
20
(26% - 6%)
0

Moreover, the Bank does not compute weighted DIs by enterprise size, but simply aggregates the obtained responses.

2-2. Why are responses to the judgment survey items aggregated into DIs?

Although percentage shares of enterprises responding to each alternative can be used easily, they might be hard to analyze, for instance, when users are interested in their long-term developments. Therefore, the Bank publishes aggregated DIs so that users can easily understand the developments of these multiple data.

2-3. How can DIs be used?

For judgment survey items, enterprises are asked to respond regarding the conditions both at the time of the survey and three months afterwards. The actual result DIs and forecast DIs are also calculated. The forecast DIs provide an information source on enterprise forecasts for users, while some peculiarities in the behavior of forecast DIs depending on economic conditions and industries should be kept in mind.

Users can also analyze economic conditions and corporate activities for various purposes, for example, by combining DIs and relevant quantitative data -- such as (1) "Business Conditions DI" and current profits or the ratio of current profits to sales, and (2) "Production Capacity DI" and fixed investment -- and by comparing current DIs with those during similar economic phases in the past.

2-4. What should we keep in mind when using individual DIs?

The Tankan has the following four kinds of DI regarding supply and demand conditions and inventory levels: (1) "domestic supply and demand conditions for products and services" ("excess demand" minus "excess supply"); (2) "overseas supply and demand conditions for products" ("excess demand" minus "excess supply"); (3) "inventory level of finished goods and merchandise" ("excessive or somewhat excessive" minus "insufficient or somewhat insufficient"); and (4) "wholesalers' inventory level" ("excessive or somewhat excessive" minus "insufficient or somewhat insufficient").

Among them, while item (3) asks enterprises to judge their own inventory levels of finished goods and merchandise, the other items (1), (2), and (4) ask them to judge supply and demand conditions or inventory levels in the industry of responding enterprises. Therefore, when using these data, this difference in definitions should be noted.

In another instance, enterprises are requested to judge their changes in conditions, instead of their levels, for the following three items: (a) "change in interest rate on loans" ("rise" minus "fall"); (b) "change in output prices" ("rise" minus "fall"); and (c) "change in input prices" ("rise" minus "fall"). Although most of the DIs indicate judgment on the level of conditions, it should be noted that these three DIs indicate that of change in conditions.

In addition, there are discontinuities due to changes of calculation method and so on in some data series. For details, see "Explanation of the Short-Term Economic Survey of Enterprises in Japan (Tankan). [PDF 248KB] " Please note that these data series are treated as continuous in the "BOJ Time-Series Data Search," for the convenience of users.

3. Q&A regarding the quantitative data

3-1. What are the estimates of the population total? How are they calculated?

The main purpose of the Tankan is to provide an accurate picture of business trends of enterprises in Japan, thereby contributing to the appropriate implementation of monetary policy. However, conducting a survey of all enterprises in Japan places the burden on all respondents, and entails considerable costs for the Bank to keep the confidential information obtained from respondents under strict security. It also takes substantial time for the Bank to compile and publish aggregated figures. As a result, it can be difficult to capture actual economic conditions in a timely manner.

Consequently, the Bank develops a sample survey framework. Specifically, sample enterprises are selected from a survey population of private enterprises with capital of 20 million yen or more -- based on the Economic Census conducted by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry -- to satisfy established criteria such as that of statistical accuracy. Total values of the target population (population total) are estimated by amplifying the figures obtained from sample enterprises based on sampling ratios.

Example: Calculating Estimates of Population Total

When the simply aggregated amount of sales of 50 sample enterprises -- which are selected from 200 survey population enterprises -- is 150, the estimate of the population total is computed as shown below:

Estimate of the population total
= average sales of sample enterprises * number of survey population enterprises
= (150/50) * 200 = 600

For quantitative survey items, estimates of the population total are calculated by strata based on industry and size classifications (total of slightly less than 400 strata), and these figures are aggregated to obtain the population total.

3-2. How are figures aggregated when enterprises do not respond with figures? What is the missing value imputation method?

Since the March 2004 survey, the Tankan has introduced the missing value imputation for annual projections of quantitative survey items. When enterprises do not respond with figures, their most recent response is individually substituted for aggregation.

When figures of the new fiscal year (FY X+1) cannot be obtained, the most recent figures (FY X) are substituted for aggregation.
The diagram showing the missing value imputation method when figures of the new fiscal year cannot be obtained.The details are shown in the main text.

From the June survey, when figures of the current fiscal year (FY X) cannot be obtained, figures of the same fiscal year obtained in the previous survey are substituted.
The diagram showing the missing value imputation method when figures of the current fiscal year cannot be obtained.The details are shown in the main text.

Until the December 2003 survey, estimates of the population total had been calculated excluding non-responding enterprises. This means that the average figures of each stratum for population estimation, to which the non-responding enterprises belong, had been substituted for missing figures.

In this case, estimates of the population total could deviate from true aggregated values if true values of the missing figures deviate from the average values of the stratum.

With this in mind, the Bank discussed alternative methods and found that the method of substituting an enterprise's most recent response which included figures likely yields more accurate aggregated values for annual projections. Based on this analysis, the Bank has revised the missing value imputation since the March 2004 survey. For details, see "Methodology for Handling Missing Values In TANKAN" (June 2002).

3-3. How are figures aggregated when the response figures significantly change? What is the treatment of outliers?

With respect to "Sales," "Current Profits," "Net Income," "Fixed Investment," and "Software Investment," the Tankan has introduced the following treatment of outliers since the December 2010 survey. Under this method, when figures reported from enterprises have a substantial impact on year-on-year rates of change and revision rates of estimates of the population total so that they are regarded as outliers, these figures are treated to remove their effects for aggregation because they do not represent the survey population.

For example, when figures of the new fiscal year (FY X+1) are substantially larger than those of the previous year (FY X), and the figures of the new fiscal year are detected as outliers, the most recent figures (FY X) are substituted for aggregation.
The diagram showing the treatment of outliers when figures of the new fiscal year are substantially larger than those of the previous year, and they are detected as outliers.The details are shown in the main text.

From the June survey, when figures of the current year (FY X) are significantly different from those of the previous survey, and they are detected as outliers, figures obtained at the previous survey of the same fiscal year are substituted.
The diagram showing the treatment of outliers when figures of the current year are significantly different from those of the previous survey, and they are detected as outliers.The details are shown in the main text.

Please note that the treatment of outliers is a statistical method to further improve the statistical accuracy of the Tankan. Figures are rarely regarded as outliers.

For details of the treatment of outliers, see "Explanation of the Short-Term Economic Survey of Enterprises in Japan (Tankan) [PDF 248KB]" and "Treatment of Outliers in Business Surveys: The Case of Short-term Economic Survey of Enterprises in Japan (Tankan)" (September 2010).

3-4. What should we keep in mind when using the year-on-year rates of change and revision rates of figures in the Tankan?

The Tankan takes a survey on figures of each fiscal year six times in total, from the March survey of the previous fiscal year to the June survey of the following fiscal year. Figures (estimates of the population total) of each fiscal year are finalized in the June survey of the following fiscal year.

For instance, as shown in the table below, figures of FY X-1 are finalized in the June X survey. At the same time, the year-on-year rates of change for FY X-1 as well as denominators for computing the year-on-year rates of change for FY X are also determined.

Therefore, when using the year-on-year rates of change for FY X, it should be noted that denominators (figures of FY X-1) in the March X survey are different from those in other surveys (the June X survey to the June X+1 survey).

Numerators and Denominators for Calculating the Year-on-Year Rates of Change for FYs X-1 and X

The diagram showing numerators and denominators for calculating the year-on-year rates of change. The details are shown in the main text.

Note:Figures in parentheses indicate the survey number for figures of each fiscal year.

Revision rates are calculated by comparing estimates of the population total of the previous survey and those of the following (current) survey as shown below:

Revision rates = (Estimates of the population total of the current survey - Estimates of the population total of the previous survey) / Estimates of the population total of the previous survey * 100

It is important to note that the "year-on-year rates of change in the previous survey" plus the "revision rates in the current survey" nearly equals the "year-on-year rates of change in the current survey" relationship does not necessarily hold. This is because, as explained above, the denominators for calculating the year-on-year rates of change in the March survey are different from those in other surveys, which sometimes results in seemingly inconsistent cases. For example, while the year-on-year rates of increase in the June survey are higher (lower) than those in the March survey, revision rates of the June survey are negative (positive).

Example 1: Fixed Investment of Large Manufacturing Enterprises (FY X)
Figure Example 1: Fixed Investment of Large Manufacturing Enterprises (FY X)
Mar.X survey June X survey
Year-on-Year Rate of Change(%) +2.3 Right Arrow +7.7
Revision Rate(%) n.a. -0.5
Example 2: Current Profits of Large Manufacturing Enterprises (FY X)
Figure Example 2: Current Profits of Large Manufacturing Enterprises (FY X)
Mar.X survey June X survey
Year-on-Year Rate of Change(%) +3.6 Right Arrow -0.3
Revision Rate(%) n.a. +1.6
Numerical Example for Example 1:

Year-on-year rate of change in the March survey = 10,595.8 billion yen (figures of FY X in the March survey) / 10,361.5 billion yen (figures of FY X-1 in the March survey) * 100 - 100 = 2.3%

Year-on-year rate of change in the June survey = 10,547.2 billion yen (figures of FY X in the June survey) / 9,790.2 billion yen (figures of FY X-1 in the June survey) * 100 - 100 = 7.7%

Revision rate in the June survey = [10,547.2 billion yen (figures of FY X in the June survey) - 10,595.8 billion yen (figures of FY X in the March survey)] / 10,595.8 billion yen (figures of FY X in the March survey) * 100 = -0.5%

3-5. How are "predicted exchange rates (the U.S. dollar-yen rates)" calculated?

Predicted exchange rates are calculated based on the figures of "exchange rates for exports" obtained from survey responses. The Bank computes the weighted averages using the survey response figures of the amount of exports rather than their simple averages as is the case with DI. Only figures of those enterprises which have provided an answer to both "Exports" and "Exchange Rates for Exports" are used for calculation.

Specifically, predicted exchange rates are calculated as shown below.

The formula for calculating the predicted exchange rate in each release classification. The expected exchange rate equals to the summation of the estimate of the population total of exchange rate of individual enterprises multiplied by their amount of exports in each stratum divided by the summation of the estimate of the population total of the amount of exports in each stratum.

3-6. What should we bear in mind when using "fixed investment" data?

Looking at average revision patterns of past surveys for annual projections of fixed investment including land purchase expenses, fixed investment of large enterprises tends to be revised upward from the first March survey to the June survey, with investment plans crystallized and those from the previous fiscal year delayed to the fiscal current year. From the December survey to the actual result survey (the June survey of the next year), fixed investment of large enterprises tends to be revised downward due to delay in construction and the deferment of plans to the following fiscal year.

On the other hand, fixed investment of small enterprises tends to be revised upward in a persistent way from the first March survey to the actual result survey (the June survey of the next year), because many small enterprises do not create investment plans in advance and consequently their fixed investment tends to be recorded as annual projections along with execution.

These revision patterns are past averages, and vastly different cases can be seen from year to year. Nevertheless, when assessing forecast figures of fixed investment, attention should be paid to the peculiarities in revision patterns.

(1) Large Enterprises
  • Manufacturing. The graph of revision patterns of past surveys for annual projections of fixed investment of large manufacturing enterprises. The details are shown in the main text.
  • Nonmanufacturing. The graph of revision patterns of past surveys for annual projections of fixed investment of large nonmanufacturing enterprises. The details are shown in the main text.
  • All industries.The graph of revision patterns of past surveys for annual projections of fixed investment of large enterprises in all industries. The details are shown in the main text.
(2) Small Enterprises
  • Manufacturing. The graph of revision patterns of past surveys for annual projections of fixed investment of small manufacturing enterprises. The details are shown in the main text.
  • Nonmanufacturing. The graph of revision patterns of past surveys for annual projections of fixed investment of small nonmanufacturing enterprises. The details are shown in the main text.
  • All industries. The graph of revision patterns of past surveys for annual projections of fixed investment of small enterprises in all industries. The details are shown in the main text.

The Tankan asks enterprises to list "amounts of newly listed tangible fixed assets (including new land purchasing expenses and lease assets)" as "fixed investment." It is worth noting that fixed investment in the Tankan includes secondhand items which are newly listed in tangible fixed assets and differs from private nonresidential investment in the National Accounts Statistics of Japan.

3-7.What should we keep in mind when using "software investment" data?

The Tankan asks enterprises to list "amounts of newly listed intangible assets (including lease assets) included in software investment" as "software investment." With information technology (IT) investment spreading in Japan, the survey of investment including software such as computer programming can help capture the actual economic conditions more accurately.

It is worth noting that software investment in the Tankan differs from computer software in the National Accounts Statistics of Japan.

3-8. What should we keep in mind when using "R&D investment" data?

With domestic offices of Japanese enterprises taking on an increasingly significant role in research and development (R&D), the Tankan added R&D investment to survey items from the March 2017 survey. The Bank asks enterprises to list "R&D expenses listed on the income statement" as "R&D investment."

The National Accounts Statistics of Japan, conforming to the new international statistical standard (2008 SNA), estimate R&D investment mainly based on R&D expenses. However, strictly speaking, this is different from data of the Tankan.

For details of the survey on R&D investment, see "Planned Revisions of the Tankan" (December 2016).

4. Q&A regarding "inflation outlook of enterprises"

4-1.How is the consumption tax treated for "inflation outlook of enterprises"?

The survey form, guidelines for filling in the form, and example answer sheet of the Tankan clearly ask enterprises to exclude the effects of institutional changes such as changes in the consumption tax rate.

4-2. What is "the average of enterprises' inflation outlook" shown in the Summary of "Inflation Outlook of Enterprises"?

"The average of enterprises' inflation outlook" is the weighted average of the rates of price change drawn from survey responses excluding "Don't know" and "Don't have clear views on general prices." The survey response rates for price change are based on the following assumption: for example, "around +15%" and "around +6% or higher" indicate +15% and +6%, respectively.

4-3. What should we keep in mind when using "inflation outlook of enterprises" data?

For the two survey items, "outlook for output prices" and "outlook for general prices," enterprises are asked to select the multiple choice response closest to their expectations for price developments 1 year ahead, 3 years ahead, and 5 years ahead. It should be noted that while the former asks expectations of rates of change compared to current levels, the latter asks expectations of year-on-year rates of change.

5. Q&A regarding the statistical accuracy

5-1. What kinds of statistical error could the Tankan contain?

Sample surveys including the Tankan contain two kinds of errors, sampling errors and non-sampling errors.

Sampling error is an error caused by surveying part of a population (sample) instead of a whole population. For example, with samples which are biased toward larger enterprises rather than average-sized enterprises in the survey population, estimates of the population total such as that of sales and fixed investment are larger than true values (population values obtained from censuses). This type of error -- caused by sample bias -- is called sampling error.

When sample enterprises provide no response or an incorrect response, estimates are likely to deviate from true values. All errors including this type of error, excluding sampling errors, are called non-sampling errors. Thus, not only sample surveys but also censuses could include non-sampling errors.

For example, non-sampling errors can arise from:

  • Non-response;
  • Respondent misunderstanding;
  • Coverage errors resulting from gaps between the target population and the survey population;
  • Mistakes in data processing; and
  • Other factors such as the design of a survey form, the order in which questions are asked, and the number and type of options offered.

For every survey, the Bank releases standard error ratios of sales (see 5-2.) by industry and enterprise size, one of the criteria for statistical accuracy, in the "Comprehensive Data Set." The Bank also analyzes the relative errors of sales and fixed investment in the Tankan. For details, see "Revision of the Sample Design of the Tankan Using the Economic Census of Japan" (September 2016).

5-2. What measures are taken in the Tankan to reduce sampling errors?

Sample enterprises of the Tankan are revised according to updates of the survey population data, drawn from the Economic Census conducted by the Ministry of Internal Affairs and Communications and the Ministry of Economy, Trade and Industry. Most recently, they were revised in March 2018.

When selecting sample enterprises, the Bank establishes the following criteria to reduce sampling errors and maintain statistical accuracy. The Bank also sets different criteria for the Survey of Financial Institutions.

  1. (1) For the six main categories -- two industries (manufacturing and nonmanufacturing) times three capital sizes (large, medium-sized, and small enterprises) --, standard error ratios of sales are within the target range (3 percent for manufacturing and 5 percent for nonmanufacturing).note
  2. (2) For individual industry classifications, standard error ratios of sales are generally within the target range of about 10 percent.
  3. (3) Sampling ratios of each stratum for population estimation achieve the target of 0.5 percent or more.
  4. (4) Sample distributions of capital size and sales could be deemed not to deviate from survey population distributions.

In regular revisions of sample enterprises, the Bank selects samples to satisfy the targets above. For details, see "Explanation of the Short-Term Economic Survey of Enterprises in Japan (Tankan). [PDF 248KB]"

In spite of these revisions, a decrease in the number of sample enterprises -- due to bankruptcies, mergers, and other factors -- could cause an increase in sampling errors prior to the next revision. Therefore, the Bank examines the statistical accuracy stated above once a year in principle and, if necessary, adds new samples in the March survey.

Note:The standard error ratio is a coefficient of variation for sample estimates, calculated by dividing the standard deviation of sample mean by the survey population mean. It represents a relative size of deviation between estimates obtained from sample surveys and true values (survey population values).

5-3. What measures are taken in the Tankan to reduce non-sampling errors?

When a survey form is not collected from a sample enterprise or a sample enterprise gives no responses to all or part of the survey items in a collected survey form, the Bank attempts to increase the response rate as follows:

  • Around the survey deadline (see 7-2.), the Bank encourages non-respondents to complete and send back forms by sending a postcard reminding the enterprise of the deadline if necessary;
  • After the deadline, the Bank calls non-respondents and encourages them to complete and send back survey forms;
  • After collecting survey forms, the Bank checks whether a response is given to every survey item. If not so, the Bank calls the respondents and asks for their responses.

The Bank also makes efforts to obtain responses, for example, by allowing enterprises to respond with: (1) approximate figures based on internal targets or estimation at the time of the survey, when there are no authorized forecasts: and (2) figures based on accounting standards that may differ from Japanese standards.

There are sometimes response errors by respondents due to misunderstandings and mistakes when filling out the form. For the purpose of reducing these types of errors, the Bank provides guidelines for completing the form and an example answer sheet. Furthermore, the Bank makes an effort to prevent erroneous responses, for example, by calling respondents, for confirmation of responses where necessary.

In cases where responses are not available despite the efforts described above, the Bank conducts the statistical method of the missing value imputation to gain more accurate aggregated figures (see 3-2.).

5-4. How are mergers and spin-offs involving sample enterprises treated in the Tankan?

In principle, the Bank takes the following procedures for mergers and spin-offs involving sample enterprises of the Tankan.

Mergers

When the core enterprise -- that with the largest sales volume -- is a sample enterprise prior to the merger, the merged enterprise is included in the sample. Specifically, in cases where there is a merger between a sample enterprise and another sample enterprise or a non-sample enterprise with a smaller sales volume, the merged enterprise is included in the sample. On the other hand, when there is a merger between a sample enterprise and a non-sample enterprise with a larger sales volume, the merged enterprise is excluded from the sample.

Spin-offs

Only the core enterprise -- that with the largest sales volume -- is included in the sample after the spin-off. However, all enterprises -- the core enterprise as well as other spin-offs -- are included in the sample when (1) the core enterprise and other spin-offs are grouped in the same stratum for population estimation and (2) the impact on estimated sales or fixed investment turns out to be small with the inclusion of these enterprises into the sample.

The Bank has introduced two exceptional rules applied when the impact of mergers and spin-offs on the estimates is above a certain level, such as large-scale mergers and spin-offs. These rules are the distribution method and the subsequent inclusion of spin-offs.

For details of treatment of mergers and spin-offs, see "Explanation of the Short-Term Economic Survey of Enterprises in Japan (Tankan). [PDF 248KB]"

6. Q&A regarding the publication of the Tankan

6-1. What is the main workflow until release?

For every survey, the Bank provides survey forms to sample enterprises (currently about 10,000) by mail or online about one month prior to scheduled release dates (in principle, at the beginning of April, July, October, and mid-December).

Completed survey forms are sent back to the Bank's Head Office (Research and Statistics Department) and responses are entered into the Bank's system. The Bank's staff at the Head Office or branches then scrutinize responses, calling respondents for confirmation where necessary.

At the Head Office, a final check of responses is made, and then survey results are aggregated.

With a view to ensuring security control, the aggregated results are kept confidential and inaccessible to anyone (including the Bank's staff) until the day of release (see 7-1.).

6-2. What kinds of publication does the Tankan provide as survey results?

The Bank provides the following materials as results of the Tankan.

Survey Results
  1. (1) Summary (main figures of survey results)
  2. (2) Outline (selected figures from the "Summary")
  3. (3) Figures by Industry (main figures by industry)
  4. (4) Summary of "Inflation Outlook of Enterprises" (main figures of "Inflation Outlook of Enterprises")
  5. (5) Comprehensive Data Set (detailed release data)
  6. (6) Long-Term Time Series Data (detailed long-term time series data)
Release Media
  1. (a) The Bank's website (for (1) to (6))
  2. (b) Information room at the Bank's Head Office (for (1) to (4))
  3. (c) (5) is also available as a statistical periodical (for a cost). For details, see List of Publications.

6-3. Are there long-term time series data of the Tankan?

Long-term time series data of the Tankan are available at the "BOJ Time-Series Data Search," which is linked from the "Statistics" section on the Bank's website.

6-4. What other information is published in addition to survey results?

The Tankan provides the following relevant information in addition to survey results, to enhance user convenience and transparency for the public. All of the following are available on the Bank's website.

  1. (1) Explanation of the Tankan
  2. (2) Sample form of the Tankan
  3. (3) Guidelines for filling in the form and example answer sheet (sent to enterprises, together with survey forms [available only in Japanese])
  4. (4) Publications regarding revisions of the Tankan

6-5. Are the release dates and times of the Tankan decided beforehand?

The Bank announces the Tankan release dates and times beforehand.

Specifically, release dates of the Tankan for the next 12 months are announced on the Bank's website around the end of June and December, together with other major statistics compiled by the Bank. The release times are always at 8:50 a.m. Japan Standard Time.

6-6. Why is the Tankan released at 8:50 a.m.?

The survey results of the Tankan are released just before 9:00 a.m., when Japanese financial and capital markets open for trade or market transactions become active.

6-7. Why do the survey results of the Tankan provide only aggregated figures and exclude any explanations or judgments?

The Bank believes that the survey results should be interpreted by users not producers of statistics, and consequently any interpretations or policy judgments are excluded.

The Bank will continue its efforts to provide useful information about the methodology of the Tankan, including these FAQs, so as to enable users to understand survey results more easily.

The Bank's views on statistics including the Tankan and the assessment of financial and economic conditions based on these statistics are shown in the Outlook for Economic Activity and Prices, which reflects discussions made by the Bank's Policy Board at Monetary Policy Meetings.

6-8. What measures are taken when published survey results are found to include incorrect figures?

When released data need a correction, corrected data, in principle, will be compiled and released as swiftly as possible. However, data would not be corrected in some cases, for example, where responding enterprises revise their figures after our release.

6-9. Whom should we contact to make inquiries about the Tankan?

The contact information for inquiries regarding the Tankan is as follows:

  1. (1) Public Relations Department (Tel: +81-3-3279-1111; e-mail: prdmail@boj.or.jp)
  2. (2) Business Survey Group, Economic Statistics Division, Research and Statistics Department (Tel: +81-3-3279-1111, ext.4023; e-mail: post.rsd5@boj.or.jp)

If you have comments on these FAQs or requests for new topics to be added, please send e-mails to: post.rsd5@boj.or.jp.

7. Q&A regarding other topics

7-1. How does the Bank of Japan manage confidential information?

Security control in the statistics production process as a whole is essential in order to ensure the reliability of statistics and prevent illicit use of statistics. The Bank is required by the Statistics Law to protect confidential information provided by enterprises under strict control.

With this in mind, the Bank keeps confidential information regarding the Tankan, such as survey results prior to release and individual responses obtained from enterprises, under strict security according to internal rules.

Specifically, during each survey period, the workspace is kept in isolation and only the staff related to the survey are allowed to have access to this area. Confidential information is handled by a limited number of staff as well as being kept strictly under lock and key for paper-based information and secured by password protection for electronically based information.

In addition, no-one has access to aggregated results of the Tankan before the day of release, and even on that day, only a limited number of staff are able to handle survey results before the time of release (8:50 a.m.).

7-2. Does the Bank of Japan set any deadline for enterprises to complete and send back survey forms?

The Bank sets a deadline. For every Tankan survey, the Bank provides survey forms to sample enterprises and publish survey results over about a one month survey period. The deadline by which enterprises complete and send back forms is usually set at approximately two weeks after the dispatch of survey forms, allowing an adequate period for enterprises to complete survey forms and for the Bank to go through the responses.

Consequently, responses from enterprises tend to be collected intensively during the few days around this deadline.

The Bank continues to collect responses until one business day prior to the day of release.