Rationality and information advantage of the Japanese government’s construction forecasts over economic development

Figures & data

Figure 1. Construction investment and economic growth.

Construction investment exhibits similar patterns to economic growth.

Figure 2. Forecast and actual values.

Forecast data exhibit similar patterns to those of actual data for each panel. They show a downward trend from the early sample period to the end of the 2000s, and a slight increase in the 2010s.

Table 1. Descriptive statistics.

	Actual data				Forecast data					Actual data				Forecast data
	Mean	St. dev	Max	Min	Mean	St. dev	Max	Min		Mean	St. dev	Max	Min	Mean	St. dev	Max	Min
	Total									Building
Full sample	6.1	10.6	33.6	−10.8	5.7	7.1	20.0	−5.5	Full sample	6.4	12.6	41.5	−20.2	6.3	6.7	19.3	−5.7
1961-1969	19.8	7.3	33.3	8.7	13.2	4.0	20.0	6.0	1961-1969	20.8	10.3	32.4	4.8	11.8	5.7	18.6	1.0
1970-1979	14.7	9.6	33.6	2.5	15.2	2.7	18.8	10.6	1970-1979	14.1	12.8	41.5	−3.6	15.7	2.8	19.3	10.7
1980-1989	4.5	5.7	14.9	−4.9	4.5	2.4	10.0	1.4	1980-1989	5.0	7.0	18.7	−3.9	5.9	1.9	9.9	3.3
1990-1999	−0.5	5.8	11.4	−9.2	1.7	3.2	5.7	−3.9	1990-1999	−2.5	8.7	12.5	−12.9	1.2	3.5	7.1	−4.6
2000-2009	−4.5	3.7	1.0	−10.8	−2.1	2.4	1.4	−5.4	2000-2009	−3.9	7.3	3.6	−20.2	0.0	3.3	5.8	−5.7
2010-2019	4.0	7.1	19.3	−2.4	2.6	5.0	11.2	−5.5	2010-2019	6.3	10.5	31.6	−2.6	3.8	3.4	12.2	0.3
	Public									Civil engineering
Full sample	6.2	10.5	32.6	−11.2	5.1	9.8	24.0	−18.6	Full sample	5.9	10.2	34.7	−9.5	4.9	8.4	23.0	−13.0
1961-1969	18.9	7.5	32.6	11.7	15.1	4.2	24.0	11.5	1961-1969	18.3	7.3	34.7	6.3	15.0	4.0	23.0	11.0
1970-1979	16.7	7.4	25.6	3.0	15.4	3.9	22.8	9.8	1970-1979	16.2	8.2	29.1	6.4	14.2	4.0	21.8	8.3
1980-1989	2.7	3.8	8.7	−2.2	1.4	3.6	10.0	−2.3	1980-1989	3.6	4.7	9.4	−6.3	2.5	3.4	10.1	−1.2
1990-1999	3.0	6.6	12.8	−6.0	3.8	5.2	14.8	−5.8	1990-1999	2.6	6.8	10.1	−7.3	2.3	3.6	7.9	−4.9
2000-2009	−5.5	5.3	7.3	−11.2	−5.4	7.4	14.1	−11.1	2000-2009	−4.9	3.8	2.9	−9.5	−4.7	2.4	−1.1	−8.1
2010-2019	2.5	6.3	14.4	−9.2	1.4	10.9	16.4	−18.6	2010-2019	1.0	5.3	11.5	−7.5	1.4	8.2	10.2	−13.0
	Private
Full sample	6.3	12.8	44.4	−20.4	6.2	6.6	19.9	−8.2
1961-1969	20.5	11.3	33.6	3.6	11.9	5.8	19.1	1.0
1970-1979	13.9	14.4	44.4	−5.2	15.1	2.9	19.9	11.1
1980-1989	5.6	7.4	18.8	−7.0	6.5	1.9	10.0	3.6
1990-1999	−2.5	8.5	14.0	−12.4	0.2	3.7	6.4	−6.0
2000-2009	−3.4	7.8	5.8	−20.4	0.2	4.3	6.8	−8.2
2010-2019	5.1	9.1	26.2	−4.3	3.6	2.3	7.5	−0.1

Table 2. Error measures.

	MAE	RMSE	MdAE	MFE	U		MAE	RMSE	MdAE	MFE	U
	Total						Building
Full sample	5.1	6.8	4.0	0.4	0.6	Full sample	7.2	9.4	5.7	0.1	0.5
1961-1969	7.1	7.9	7.1	6.6	0.6	1961-1969	9.4	10.8	8.9	9.0	0.5
1970-1979	7.0	8.6	5.2	−0.5	0.6	1970-1979	9.6	11.5	8.6	−1.7	0.4
1980-1989	5.0	5.8	4.8	−0.1	1.2	1980-1989	6.3	7.2	6.2	−0.9	1.1
1990-1999	3.9	4.5	4.4	−2.2	0.5	1990-1999	7.0	7.6	6.8	−3.7	0.4
2000-2009	3.2	4.5	1.9	−2.4	0.7	2000-2009	5.2	8.1	3.0	−3.9	0.7
2010-2019	4.8	8.4	2.3	1.3	0.6	2010-2019	5.8	10.4	3.1	2.5	0.4
	Public						Civil engineering
Full sample	4.9	7.0	3.0	1.0	1.1	Full sample	4.7	6.0	3.7	1.0	0.8
1961-1969	4.7	6.6	3.0	3.7	1.1	1961-1969	5.1	5.9	4.9	3.3	0.5
1970-1979	5.8	6.4	5.2	1.4	0.5	1970-1979	6.8	8.0	5.8	2.0	0.8
1980-1989	3.1	4.5	1.8	1.3	1.7	1980-1989	3.5	4.5	2.5	1.1	1.1
1990-1999	4.7	5.7	3.3	−0.9	0.7	1990-1999	4.0	4.6	3.8	0.3	0.3
2000-2009	2.7	3.7	1.6	−0.1	1.2	2000-2009	2.0	2.4	2.4	−0.2	1.0
2010-2019	8.4	11.9	2.8	1.0	1.9	2010-2019	6.7	8.4	4.6	−0.4	1.3
	Private
Full sample	7.1	9.2	5.9	0.1	0.4
1961-1969	9.2	11.0	8.9	8.6	0.4
1970-1979	10.8	13.0	7.7	−1.1	0.5
1980-1989	6.3	7.4	6.8	−1.0	0.9
1990-1999	6.1	6.8	7.2	−2.7	0.4
2000-2009	4.7	6.2	3.9	−3.6	0.4
2010-2019	5.9	9.0	3.6	1.5	0.4

Notes: MAE: Mean absolute error; RMSE: Root mean squared error; MdAE: Median absolute error; MFE: Mean forecast error; U: Theil’s U.

Table 3. Rationality tests.

	Standard	Fluctuation
Total	0.38	8.80
Public	1.73	25.9**
Private	2.09	14.8**
Building	0.92	5.67
Civil engineering	0.73	10.2

Notes: Standard indicates the standard rationality test. Fluctuation indicates the Fluctuation Rationality test. Newey–West standard errors with a lag of 1 are used.

** indicates the 1% significance levels.

Figure 3. Test statistic of Fluctuation Rationality test.

Forecast rationality is primarily violated in recent decades along with the largest postwar external and domestic shocks, the global financial crisis of 2008-09 and the Great East Japan Earthquake in 2011.

Table 4. Information advantage tests.

	GDP		Private investment		Residential investment		Public investment		Public works related budget
	Standard	Fluctuation	Standard	Fluctuation	Standard	Fluctuation	Standard	Fluctuation	Standard	Fluctuation
Total	0.12	31.8**	0.03	55.0**	1.13	41.8**	0.07	8.57	0.05	6.85
Public	6.82*	62.8**	2.34	43.5**	9.19**	68.9**	2.61	5.82	3.03	3.50
Private	2.03	21.5**	1.93	20.1**	0.00	47.9**	0.00	19.0**	0.10	11.1
Building	0.95	15.1*	0.25	19.5**	0.23	38.3**	0.17	13.5*	1.87	8.15
Civil engineering	3.42	52.1**	1.58	58.8**	2.75	38.9**	1.01	6.40	4.81*	6.60

Notes: Standard indicates the standard rationality test. Fluctuation indicates the Fluctuation Rationality test. Newey–West standard errors with a lag of 1 are used.

**, * indicates the 1%, and 5% significance levels, respectively.

Figure 4. Test statistic of information advantage test regarding real GDP growth.

The government’s GDP forecast provides additional information.

Figure 5. Test statistic of information advantage test regarding private investment.

The government’s private investment forecast provides additional information.

Figure 6. Test statistic of information advantage test regarding residential investment.

The government’s residential investment forecast provides additional information.

Figure 7. Test statistic of information advantage test regarding public investment: Private.

The government’s public investment forecast provides additional information for private construction forecast.

Table 5. Rationality and information advantage tests: Robustness.

	Rationality test	Information advantage test
		GDP	Private investment	Housing	Public investment	Public works related budget
Window size at 20
Total	27.4**	73.3**	78.8**	86.3**	46.6**	12.6
Public	21.3**	100.4**	39.6**	89.6**	77.4**	6.08
Private	20.1**	78.8**	47.9**	132.7**	30.1**	27.6**
Building	13.3*	58.5**	32.8**	129.2**	29.4**	26.8**
Civil engineering	13.6*	43.7**	79.0**	107.5**	51.7**	8.54
Window size at 30
Total	5.57	24.6**	40.2**	13.2	3.17	5.78
Public	12.9*	21.1**	36.3**	14.8*	4.18	3.50
Private	15.4**	20.3**	22.2**	11.8	4.83	13.4*
Building	5.36	13.9*	23.0**	7.52	3.50	9.23
Civil engineering	4.04	18.4**	38.7**	18.8**	7.24	8.48

Notes: Newey–West standard errors with a lag of 1 are used. **, * indicates the 1%, and 5% significance levels, respectively.

Figure A2. Test statistic of information advantage test regarding real GDP growth.

The government’s GDP forecast provides additional information for all construction forecasts.

Figure A3. Test statistic of information advantage test regarding private investment.

The government’s private investment forecast provides additional information for all construction forecasts.

Figure A4. Test statistic of information advantage test regarding residential investment.

The government’s residential investment forecast provides additional information for all construction forecasts.

Figure A5. Test statistic of information advantage test regarding public investment: Building.

The government’s public investment forecast provides additional information for building construction forecast.

Figure A6. Test statistic of information advantage test regarding real GDP growth: Robustness.

There is no information advantage of the GDP forecast in the early years of the sample period.

Figure A7. Test statistic of information advantage test regarding private investment: Robustness.

There is no information advantage of the private investment forecast in the early years of the sample period.

Figure A8. Test statistic of information advantage test regarding residential investment: Robustness.

There is also no information advantage of the residential investment forecast for the civil engineering construction forecast until the late 2000s. While the residential investment forecast gained an information advantage for the public construction forecast in 2008 and 2009, the residential investment forecast lost an information advantage to the public construction forecast in the early years of the sample period.

Figure A9. Test statistic of information advantage test of private investment regarding public works related budget.

The public works related budget has an information advantage for the private investment forecast in 1993.

Data availability statement

The data that support the findings of this study are available from the author upon reasonable request.