Estimating Japan’s Gross Domestic Income

Based on Taxation Data

1

FEB. 10, 2017

Hiroyuki Fujiwara

Bank of JapanResearch and Statistics Department

The views expressed in this presentation are those of the author’s and do not necessarily reflect those of the Bank of Japan.

2

(1) FY2013

(2) FY2014

・ Consistencies of 3 ways of measuring GDP and net lending / borrowing for households need to be checked.

・ We try to estimate GDP from the income side using taxation data.

Two reasons why I got started on this analysis

・Household saving rate was -1.3 percent(-3.6 trillion yen; It was negativefor the first time since FY1994).

・Currency and deposits of households increased by more than 17 trillion yen.

・Operating surplus and mixed income decreased by 2.4 trillion yen. ・Both corporate profits and corporate tax revenue increased.

１ Introduction

How the Official Series Compiles GDE, GDP and GDI in Japan

２ Estimation for each GDI component

2.1 Compensation of Employees

2.2 Operating Surplus

2.3 Mixed Income

３ Estimated and Official GDI

４ Factors of Divergence between Our Estimated GDI and Official GDE

５ Concluding Remarks

3

Content of Paper

4

１ Introduction

5

・The three aggregates -- GDE, GDP and GDI -- are conceptually the same.

How the Official Series Compiles GDE, GDP and GDI in Japan (1)

・In reality, there may be some discrepancy among them due to differencesin data sources, timing and estimation methods.

In Japan

・In the quarterly estimates, only GDE and compensation of employees are compiled and published.

・In the annual reports, GDE, GDP and GDI are available.

⇒ We focus on the annual reports today.

6

How the Official Series Compiles GDE, GDP and GDI in Japan (2) ・There is a slight discrepancy between GDE and GDP.

・ GDI is equal to GDP.

・Figure 1 : GDE, GDP and GDI

GDE and GDP are estimated independently.

GDI is not estimated independently.

…

(less)Subsidies(3 trillion yen)

…

GDE(486.9 trillion yen)

GDP(487.4 trillion yen)

Industries(428 trillion yen)

GDI(487.4 trillion yen)

Others(130 trillion yen)

Others(60 trillion yen)

Manufacturing（90 trillion yen）

Consumption of fixed capital(104 trillion yen)Service activities

(95 trillion yen)Operating surplus and mixed

income(91 trillion yen)

…

Wholesale and retail trade(69 trillion yen)

Domestic final consumptionexpenditure of households

(288 trillion yen)

Compensation of employees(251 trillion yen)

Taxes on production andimports

(44 trillion yen)

Gross fixed capital formationof non-resi. investment by

private sectors(68 trillion yen)

(CY2014)

Total of GDI matches GDP .

Estimated

A balancing item

7

(1) GDE: Commodity flow approach (2) GDP: Value added approach

How the Official Series Compiles GDE, GDP and GDI in Japan (3)

・Both outputs serve as control totals.

・Using a wide variety of statistical surveys.

However, households and enterprises have recently become more reluctant to respond to statistical surveys.

3) GDI: Its level becomes equal to that of GDP by adjusting operating surplus and mixed income as a balancing item.

(Our novel approach - with reference to the case of the United States)

We try to calculate each component using taxation data and estimate GDI independently.

8

2 Estimation for each GDI component

9

2.1 Compensation of Employees (1)

・National Accounts: Multiplying cash earnings per employee by the number of all employees.Using sample surveys (statistical survey).

・Municipal Tax Survey: Records wages and salaries of all taxpayers who pay local income tax.This is a complete survey (administrative data).

National Accounts Versus Municipal Tax Survey

The difference in wage and salaries has disappeared, whereas the number of employees of the National Accounts has continued to exceed that of the Municipal Tax Survey by about 20%.

-20

0

20

40

60

80

100

-60

0

60

120

180

240

300

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Divergence rate (X/Y-1) (right scale)

National Accounts

Municipal Tax Survey

(trillion yen) (%)

CY

(X)

(Y)

-20

0

20

40

60

80

100

2,000

3,000

4,000

5,000

6,000

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Divergence rate (X/Y-1) (right scale)

National Accounts

Municipal Tax Survey

(10,000 persons) (%)

CY

(X)

(Y)

Figure 2 Wages and salaries Figure 3 Number of employees

10

2.1 Compensation of Employees (2)

Our estimated wages and salaries= income of taxpayers of local income tax+ income of non-taxpayers of local income tax + income of small differences in coverage and definition

Our estimate and the official series were almost at the same level in 1994.However, as time goes by, our estimate has come to take a larger amountcompared with the official series.

-10

0

10

20

30

40

50

140

160

180

200

220

240

260

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Divergence B-A (right scale)

National Accounts A

Estimate B

(trillion yen) (trillion yen)

CY

(2014CY, trillion yens)

211.913.3-0.4

224.7210.614.2 Divergence (A -B )

Taxpayer of local income tax Non-taxpayer of local income tax Adjusting small differences such as bonuses for directors Estimated total value (A ) Official total value (B )

Estimated and Official Wages and SalariesTable 1 Figure 4 Estimated Wages and Salaries

(2)

(1)

(2) (3)

(1)

(3)

→→

→

Table 1

Figure 4

11

2.2 Operating Surplus (1)

・We attempt to estimate net operating surplus using operating profits (not as a balancing item) from (1) the Financial Statements of Corporations and (2) corporate tax revenue.

・ Corporate income captured by taxation data is quite different from operating surplus in the National Accounts.

(1) In stead of using corporate taxation data directly, we use the Financial Statements of Corporations.

(2) We correct sampling errors associated with the Financial Statements of Corporations by using comprehensive corporate tax revenue data.

・The Financial Statements of Corporations is a sample survey and may contain some sampling errors.

（2 points)

12

2.2 Operating Surplus (2)

National Accounts versus Financial Statements of CorporationsFigure 5 Net operating surplus and operating profits

Figure 6 Consumption of fixed capital and depreciation

Figure 7 Gross fixed capital formation and capital investment

・Figure 5 shows that the gap between net operating surplus and operating profits is relatively small. ・Figure 6 shows that consumption of fixed capital exceeds depreciation by a great deal. → This is because gross fixed capital formation / non-financial assets in the National Accounts exceeds capital investment / non-financial assets in the Financial Statements of Corporations by a great deal (See Figures 7 and 8).

0

10

20

30

40

50

60

70

80

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

National AccountsFinancial Statements of Corporations

(trillion yen)

FY

0

10

20

30

40

50

60

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

National AccountsFinancial Statements of Corporations

(trillion yen)

FY

0

100

200

300

400

500

600

700

800

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

National AccountsFinancial Statements of Corporations

(trillion yen)Figure 8 Non-financial assets excluding land

0102030405060708090

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

National AccountsFinancial Statements of Corporations

(trillion yen)

FY

・We estimate net operating surplus and consumption of fixed capital together (gross operating surplus), not separately. (another key point)

13

2.2 Operating Surplus (3)

・Table 2 is a simple illustrative example to demonstrate how specific figures are treated differently by the National Accounts and the Financial Statements of Corporations.

Table 2 Difference of SNA Account and Corporate Account

・When we compare the sum of net operating surplus and consumption of fixed capital with the sum of operating profits and depreciation, the gap (10 in the example) remains unchanged in both periods 1 and 2. This gap is equal to the amount by which gross fixed capital formation exceeds capital investment.

(2 period example)

Assumption 1: Gross output and corporates' revenue are 100 in both the accounts.Assumption 2: Purchase expenses are 50 in both the accounts.Assumption 3: Depreciation rates are 50% in both the accounts.Assumption 4: The allocation of purchase expense is different between the accounts.

GDP statistics (SNA account) period 1 period 2Gross output (a1 ) 100 100Intermediate input (a2 ) 30 30Consumpiton of fixed capital (a3 ) 0 10Net operating surplus (a4 = a1 -a2 -a3 ) 70 60Gross fixed capital formation 20 20Fixed assets (end of the period) 20 30

Gross operating surplus (a5 = a3 +a4 ) 70 70

Corporate AccountRevenue (b1 ) 100 100Current expense (b2 ) 40 40Depreciation (b3 ) 0 5Operating profits (b4 = b1 -b2 -b3 ) 60 55Capital investment 10 10Fixed assets (end of the period) 10 15

Operating profits + Depreciation (b5 = b3 +b4 ) 60 60

14

2.2 Operating Surplus (4)

・We estimate the operating surplus according to the following formula.

GOS : gross operating surplus, πF : operating profits, δF : depreciation, IS : gross fixed capital formation , IF : capital investment, NOS : net operating surplus, δS : consumption of fixed capital

Equation (1) GOS = πF + δF + IS－IF

Equation (2) NOS = GOS － δS

Equation (3) π = πF ×τ

/ τF

Prior to corporate tax adjustment

After corporate tax adjustment

・Assuming that the deviation is largely due to the sampling error of the Financial Statements of Corporations and that the error is proportional to the sampling error of the tax.

We substitute π to πF in equation (1).

π : adjusted operating profits, τ : actual corporate tax revenue from the government τ ∶ corporate tax from the Financial Statements of Corporations

15

2.2 Operating Surplus (5)

・Our estimate do not differ much from that of the National Accounts statistics.The average divergence between these two series for the entire sample period is as

small as -0.5 trillion yen or -1.2% . However, 2014 observes a significantly wider gap.

Figure 9 Estimated and Official Net Operating Surplus

12.6

0

10

20

30

40

50

60

70

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

National Accounts A

Estimate (after tax adjustment) B

(trillion yen)

FY

-15

-10

-5

0

5

10

15

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

B-A

(trillion yen)

FY

Difference

16

2.3 Mixed Income

・We estimate mixed income of households using self-assessment income taxation data of private unincorporated enterprises from the National Tax Agency.

Figure 10 Estimated and Official Mixed Income of Households

0

2

4

6

8

10

12

14

16

18

08 09 10 11 12 13 14

National Accounts

Self-assessment income taxation data (before adjustment)

Self-assessment income taxation data (after adjustment)

(trillion yen)

CY

・We subtract any rent from land that households receive and add the amount of an income deduction for income earners.

Our estimate is only about 10 trillion yen and does not differ much from that of the National Accounts.

17

３ Estimated GDI and Official GDE

18

Estimated GDI and Official GDE

513

474

482

490

528

484

495

519

440

450

460

470

480

490

500

510

520

530

540

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Official GDE Estimated GDI

(trillion yen)

FY

Figure 11 Estimated GDI and Official GDE(a) Level

(2) is not so large (Statistical discrepancy in Figure 11 [b]).

2.5

14.4

12.6

12.9 14.514.69.7

12.3

29.5

-20

-15

-10

-5

0

5

10

15

20

25

30

35

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Operating surplus Compensation of employees

Statistical discrepancy Total

(trillion yen)

FY

- 0.9

(discrepancy between official GDI and GDE )

- 14.8

(b) Difference between estimated GDI and official GDE

・Our estimated GDI is larger than the Official GDE.

(1) mainly owes to compensation of employees. As for 2014, operating surplus also made a large contribution.

・The difference between these two series is divided into (1) that between our estimated GDI and the official GDI (= GDP), and (2) that between the official GDI ( = GDP) and GDE.

19

Estimated Real GDI and Official Real GDE

525 520

530 525

540 530

543 556

440

460

480

500

520

540

560

580

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Official GDE

Estimated GDI

(trillion yen)

FY

2.0

-1.0

2.5 2.4

-6

-4

-2

0

2

4

6

8

95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Official GDE

Estimated GDI

(y/y % chg.)

FY

Figure 12 Estimated Real GDI and Official Real GDE

(a) Level (b) Growth rate

・The reading of 2014 exceeded the previous peak of 2007 by a wide margin, whereas real GDE of the National Accounts did not surpass the 2007 peak.・In our calculation, real GDI grew by 2.4% in 2014, which is a sharp contrast to a contraction by -1.0% in GDE.

20

(1) When using our estimate for compensation of employees, discrepancy between non-financial and financial transactions accounts for households narrows for FY2013.

However, we need to make other adjustments in order to decrease the remaining discrepancy between the two transactions accounts.

(2) Our estimate of net operating surplus increases in FY2014.This is the same result as corporate profits in many statistical surveys and corporate

tax revenue also increased in FY2014.

-10

0

10

20

30

40

50

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14

Non-financial transactionsModified non-financial transactionsFinancial transactions

(trillion yen)

FY

Financial surplus

Financial deficit

Figure 13 Net lending/borrowing for households

Results for the initial question

21

Introduction of the SNA2008 (1)

・In December 2016, the National Accounts had a comprehensive revision including the introduction of the SNA2008.

・The revision pushed up GDE to a large extent.Contribution of difference before and after revision is attributable to the following two factors.(1) Changes in definitions from SNA1993 to SNA2008

(2) Others including changes in data sources

Figure 14 Difference between annual report for 2014 (SNA1993 base) and that for 2015 (SNA2008 base) (official GDE)

-5

0

5

10

15

20

25

30

01 02 03 04 05 06 07 08 09 10 11 12 13 14

Others

Contribution of changes in definitions from SNA1993 to SNA2008

Total

(trillion yen)

FY

22

(Source) http://www.nowcast.co.jp/2017/01/08snagdp-14gdp21.html

(y/y % chg.)Figure 15 Real growth rate

Official GDE ( SNA2008 base)Estimated GDI ( SNA2008 base) by NOWCAST, Inc.Our Estimated GDI ( SNA1993 base)

Introduction of the SNA2008 (2)

・NOWCAST, Inc. (private company) started to compile and publish the estimated GDI of SNA2008 base using our estimation method.

23

4 Factors of Divergence between Our Estimated GDI and Official GDE

(1) Between FYs 1994 and 2013, we can see a gradual increase in the difference between the estimated GDI and the official GDE.

(2) Then, in FY2014, we can see a rapid increase.

24-10

-5

0

5

10

15

20

25

30

35

94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14FY

(1)

(2)

(trillion yen)

Figure 16 Difference between estimated GDI and official GDE

Difference between Estimated GDI and Official GDE

The difference between Economic Census (statistical survey) and Company Sample Survey(taxation data survey) expands gradually in the long run.

Economic Census・・・ 1.75 million Company Sample Survey ・・・ 2.62 million

Figure 17 Number of Corporations

(1) Number of corporations

25

Factors of Divergence between Estimated GDI and Official GDE (1)

(Data of 2014)

0

50

100

150

200

250

300

80 85 90 95 00 05 10 14

Company Sample Survey

Economic census (Establishment and Enterprise Census until 2006)

(10,000 corporations)

Economic Census (Establishment and Enterprise Census until 2006)

26

・ In source surveys of the National Accounts, companies are supposed to report their nominal values with the consumption tax included and the National Accounts are then compiled as such.

(2) Effects of the consumption tax hike in 2014

・ If a certain number of those companies had excluded consumption tax when reporting, there could have been a disruption in the data series when the consumption tax rate was raised from 5% to 8% in 2014. This would have resulted in the growth rate of the corresponding year appearing lower than the actual growth rate.

Factors of Divergence between Estimated GDI and Official GDE (2)

27

５ Concluding Remarks

28

Concluding Remarks

・ Further development of our GDI estimation based on taxation data will require the following two practical issues to be resolved.

(1)The long time lag in the availability of taxation data needs to be overcome. Taxation data cannot be used for the annual report of Japanese National Accounts, which is usually published in December of the following year.

(2)We need to explore how to estimate GDI on a quarterly basis.

Thank you for your attention

29