Chapter 4. Impacts of a Declining and Aging Population on Central and Local Government Expenditures in Japan

Transcription

1 Chapter 4 Impacts of a Declining and Aging Population on Central and Local Government Expenditures in Japan Eiji Sumi, Niigata University Sunchung Oh, Kansai University 1. Introduction Here, we analyze the influence of a declining and aging population on central and local government expenditure. We expect that a population decline in the future will cause a decrease in government expenditure. On the other hand, we expect that population aging will cause an increase in government expenditure. For instance, public welfare plays an important role in a municipality s services. Therefore, we expect that an expansion of expenditure for public welfare caused by population aging will have the biggest influence on a municipality s finances. Moreover, as the degree of population decline and aging differs from region to region, we expect that these two phenomena will have various effects on the expenditures of prefectures and municipalities. Therefore, we consider that it is important to analyze how structural changes in Japanese society, population decline, and population aging will influence central and local government expenditures. However, in general, the Japanese government s medium-term budget forecast is based on only a medium-term economic perspective (e.g. nominal GDP growth rate). Thus, this might make the government choose inappropriate policies. Here, we adopt the following two procedures to forecast central and local government expenditures. Firstly, we regress government expenditures on population and rate of aging to see which factors influence government expenditure. Secondly, if it is verified that population and aging are factors that affect decisions on government expenditure, we put future population and rate of aging into an estimated regression model. We can get forecasting data on government expenditures from FY2006 to FY2030 in this way. The composition of this paper is as follows. In Section 2, we describe concretely the forecast method of central and local government expenditures. In Section 3, we show the estimation results of central and local government expenditures from 71

2 FY2006 to FY2030. Finally, we analyze what factors influence the change in nominal expenditure from FY2006 to FY2030: i.e., price changes, population change and aging change. 2. Methodology for estimating present and future expenditures 2.1 Model for estimating present national expenditure We verify the decision factors of expenditure in central government. First, we explain the model that estimates central government expenditure by population and rate of aging. We define the expenditure of central government as General Expenditure. General Expenditure is total expenditure minus National Debt Service (costs allotted to repaying principal and interest on government debt) and Local Allocation Tax. We define the estimation model as follows. Where t=fy1964,,fy2004, OLDR = the rate of aging We define that u is an error term. We assume AR1 here. We define that is other disturbance elements and i.i.d. The rate of aging (OLDR) is defined as the number of persons aged 65 years or more. We adopt General Expenditure as central government expenditure. This is because National Debt Service depends not on population etc., but on public debt balance and interest rate. Local allocation tax is calculated from a constant ratio of national tax income. This fund is needed by central government to implement policies. We obtain General Expenditure data for the term FY1964-FY2004 from Statistics of Public Finance. We obtain the population in the future and the rate of aging from Population Projections by Prefectures- (2000) - (2030) - March in 2002 of National Institute of Population and Social Security Research (IPSS). Annual expenditure is deflated using the GDP deflator 1. We use data for the 41 years from FY1964 to FY2004 for the estimation. Considering the problem of serial correlation on error term, we estimate this model using the Prais-Winsten method. 1 Benchmark year is

3 We describe the sign condition of each explanation variable here. We expect that the decrease of population will cause a decreasing factor for government expenditure. On the other hand, we expect that aging will cause an increasing factor for government expenditure. Therefore, we forecast that the sign of the coefficient of population is positive, and the one of aging is also positive Methodology for estimating future central government expenditure Next, we obtain the forecast of central government expenditure using the modeled population and rate of aging in the future. We assume FY2006 to be the time when the forecast starts. However, we can obtain financial accounts data only until FY2004, and have budget data only for FY2006. So, we calculate expenditure in FY2006 by the following procedures. Step 1: We obtain the forecast expenditure of the central government by applying population and rate of aging in the future to the regression model. Step 2: There is a difference between the annual expenditure budget in FY 2006 and the forecast expenditure obtained by Step 1. We arrange the constant term of regression from this difference. Step 3: We obtain the forecasted expenditure after FY2007 with the regression model having the corrected constant term exogenously for population and rate of aging in the future. The expenditures calculated according to the above-mentioned procedure are evaluated in FY2000 prices. Therefore, we change them to nominal values using the forecasted GDP deflator Estimation model for local expenditure Now, let us focus on the expenditures of prefectures and municipalities. The expenditure data for municipality totals relate to 47 prefectural administrative divisions. We forecast the influence that population decline and population aging in the future have on local expenditure. We verify the decision factor of local expenditure by regressing local expenditure on population and rate of aging. We define the estimation model as follows. 2 Table 2 (E1) shows that every explanatory variable is positive and significant. 3 Figure 18 shows the forecast GDP deflator for FY

4 We define u i as an element that shows the individual effect. is other disturbance elements and i.i.d. And, we use data on expenditure except public debt servicing expenses. We obtain local expenditure data from Statistical Annual Report of Local Public Finance. We obtain the population in the future and the rate of aging from Population Projections by Prefectures- (2000) - (2030) - March in 2002 of National Institute of Population and Social Security Research. The nominal local expenditure is deflated using the Consumer Price Index 4. The theoretical grounds of the expenditure regression model are the same as the method of calculating Basic Fiscal Needs, which is the basic data used to decide local expenditure. It is clarified that Basic Fiscal Needs are almost explained by population and area as mentioned in the preceding research of Hayashi (1995) and Nagamine (2000). Actually, the population factor is adopted mainly in the process by which Basic Fiscal Needs is calculated. The local expenditure regression model here is based on such research results. We use a panel data set consisting of 47 prefectures observed yearly over 17 years (FY1986 to FY2002) a balanced panel. We adopt a fixed effect model. Table 3 shows a detailed estimate obtained by the fixed effect model. We can confirm the existence of the individual effect by comparing the Least Squares Dummy Variable (LSDV) model with a Without Dummy model. As a result, the F statistic for testing the joint significance of individual effects is strongly in favor of a prefecture specific effect in the data. We describe the sign condition of each explanation variable. We expect that population decrease is a decreasing factor for government expenditure. On the other hand, we expect that aging is an increasing factor for government expenditure. Therefore, we consider that the sign of the coefficient of population is positive, and that of aging is also positive. As shown in Table 3, every explanatory variable shows 4 The data on expenditure are evaluated at 2000 prices. 74

5 a positive sign, and we reject the null hypothesis. Moreover, we can also validate our estimation model with the F-test. 2.4 Methodology for estimating future local expenditure Next, we obtain the forecasted expenditure of the central government by adding population and rate of aging in the future to the model. We assume FY2006 to be the starting time for the forecast. However, only the budget for FY2006 is available. So, we calculate the expenditure in FY2006 by the following procedures. Step 1: We obtain the forecasted expenditure of local government by adding population and rate of aging in the future to the regression model. Step 2: The annual expenditure in FY 2006 is obtained by multiplying the ratio of Local Public Finance Program in FY2006 to Local Public Finance Program in FY2003 by the final accounts in FY2003. There is a difference between the annual expenditure budget in FY2006 and forecasted expenditure obtained by Step 1. We arrange the constant term of regression from this difference. Step 3: We obtain the forecasted expenditures after FY2007 with the regression model having the corrected constant term exogenously for the population of each administrative division in the future and the rate of aging in the future. The expenditure calculated according to the above-mentioned procedure is evaluated at FY 2000 prices. Therefore, we change them to nominal values using the forecast GDP deflator. 3. Prediction of government expenditure 3.1 Prediction of central government expenditure We show the forecasted General Expenditure from FY2006 to FY First, we compare the forecasted General Expenditure based on the forecast model of population and rate of aging with General Expenditure (NGR) extended by the nominal GDP growth rate. As shown in Figure 1, the forecasted General Expenditure based on population and rate of aging stands at 62.1 trillion yen (+33.9% compared 5 The forecast value is nominal. 75

6 to FY2006), while General Expenditure in FY2006 is 46.4 trillion yen. On the other hand, General Expenditure (NGR) that increases at the nominal economic growth rate stands at 92.9 trillion yen (+100.3%). There is a difference of about 30.8 trillion yen in FY Contribution of factors changing central expenditure We analyze what influence population decline and population aging in the future have on General Expenditure in the future based on the forecast model. Then, we decompose the change in nominal General Expenditure from FY2006 to FY2030 into contribution of population, aging, and factor of price changes, and see what influence those factors have. The results are shown in Figure 2 and Figure 3. The change in nominal General Expenditure from FY2006 to FY2030 is divided into real change and nominal change. Real change is the change in real expenditure, which can be explained by the factors of population and aging. Nominal change is the change in annual expenditure caused by price changes. In short, the following relationship exists between the factors. Change in nominal expenditure = Contribution of prices + Contribution of real expenditure As shown in Figure 2, the change in nominal General Expenditure from FY2006 to FY2030 is decomposed into the change in expenditure caused by price changes, and in real terms by both factors of population and aging. (In Figure 2, 46.4 trillion yen at the starting point of FY2006 is replaced with 0.) As shown in Figure 2, nominal General Expenditure increases by 15.7 trillion yen (+33.9%) in FY2030 compared to FY2006. (In short, this nominal general expenditure stands at 62.1 trillion yen.) The contribution of price rises is trillion yen (+36.8%), and the contribution of real expenditure is 1.3 trillion yen ( 2.8%), as shown in Figure 2. Real General Expenditure declines by 1.3 trillion yen in FY2030 compared to FY2006, as shown in Figure 3. (In short, this real General Expenditure stands at 45.1 trillion yen.) This shows that the contribution of the factor of aging is +9 trillion yen (+19.4%), and the contribution of population decline is 10.3 trillion yen ( 22.3%). 76

7 Note: Exp (NGR) is the expenditure extended by nominal economic growth rate. Figure 1: Forecast of the central General Expenditure Note: The change from FY2006 to FY2030. Figures in parentheses are percentages. Figure 2: Contribution of factors of expenditure change 77

8 Note: The change from FY2006 to FY2030. Figures in parentheses are percentages. Figure 3: Contribution of factors of real expenditure change 3.3 Forecast of local expenditure We show the forecasted expenditure for prefectures (47 administrative divisions) and municipalities from FY2006 to FY First, we compare the forecasted local expenditure based on the forecast model of population and rate of aging with local Expenditure (NGR) extended by nominal GDP growth rate. As shown in Figure 4, the forecasted prefecture expenditure based on population and rate of aging stands at 71.9 trillion yen (+55.0%), while expenditure in FY2006 is 46.4 trillion yen. On the other hand, prefecture expenditure (NGR) that expands at the nominal economic growth rate stands at 93.3 trillion yen (+100.3%). There is a difference of about 21.1 trillion yen in FY2030. As shown in Figure 4, the forecasted municipality expenditure based on population and rate of aging stands at 75.7 trillion yen (+57.5%), while expenditure in FY2006 is 48.1 trillion yen. On the other hand, municipality expenditure (NGR) that expands at the nominal economic growth rate stands at 96.3 trillion yen (+100.3%). There is a difference of about 20.6 trillion yen in FY Each value is nominal. 78

9 This shows that the entire local expenditure increases from 94.5 trillion yen in FY2006 to trillion yen (+56.3%) in FY2030. On the other hand, expenditure that expands in proportion to nominal GDP growth rate stands at trillion yen (+100.3%) in FY2030. Table 1 shows nominal prefecture and municipality (47 administrative divisions) expenditure in FY2006 and FY Contribution of factors changing local expenditure We have analyzed what influence population decline and population aging in the future has on local expenditure in the future based on the forecast model. Then, we decomposed the change in nominal local expenditure from FY2006 to FY2030 into the contributions of population, aging, and the factor of price changes, and see what influence those factors have. The results are shown in Figures 6~9. First, we look at prefecture expenditure. As shown in Figure 6, the change in nominal prefecture expenditure from FY2006 to FY2030 is decomposed into the change in expenditure caused by price changes and in real terms by changes of both factors of population and aging. (In Figure 6, 46.4 trillion yen at the starting point of FY2006 is replaced by 0.) As shown in Figure 6, nominal prefecture expenditure increases by 25.5 trillion yen (+55.0%) in FY2030 compared to FY2006. (In short, this nominal prefecture expenditure stands at 71.9 trillion yen.) As shown in Figure 6, the contribution of price rises is trillion yen (+42.1%). Real prefecture expenditure increases by 6.4 trillion yen (+13.7%) in FY2030 compared to FY2006, as shown in Figure 7. (In short, real prefecture expenditure stands at 52.8 trillion yen.) This shows that the contribution of the factor of aging is +9.5 trillion yen (+20.5%), and the contribution of population decline is 3.1 trillion yen. ( 6.8%) Next, we look at municipality expenditure. As shown in Figure 8, the change in nominal municipality expenditure from FY2006 to FY2030 is decomposed into the change in expenditure caused by price changes and in real terms by both factors of population and aging. (In Figure 8, 48.1 trillion yen at the starting point of FY2006 is replaced by 0.) As shown in Figure 8, nominal municipality expenditure increases by 27.7 trillion yen (+57.5%) in FY2030 compared to FY2006. (In short, this nominal prefecture expenditure stands at 75.7 trillion yen.) 79

10 The contribution of price rises is trillion yen (+41.8%), as shown in Figure 8. Real municipality expenditure increases by 7.6 trillion yen (+15.7%) compared to FY2006 in FY2030, as shown in Figure 9. (In short, real prefecture expenditure stands at 55.7 trillion yen.) This shows that the contribution of the factor of aging is trillion yen (+33.7%), and the contribution of population decline is 8.6 trillion yen ( 18.0%). Compared to the influences population decline and population aging have on real expenditure in prefectures and municipalities in Figure 7 and Figure 9, you can understand that both population decline and population aging have big influences on municipalities. While in municipalities the level of expenditure reduction caused by population decline is relatively high, the level of expenditure increase caused by population aging is far higher than that of population decline. Actually, in FY2030 the influence of the population factor is a 3.1 trillion yen decrease in prefectures and a 8.6 trillion yen decrease in municipalities. On the other hand, the influence of the factor of aging is a 9.5 trillion yen increase in prefectures and a 16.6 trillion yen increase in municipalities. The reason why the influence of aging is large in municipalities is that a lot of welfare services for the aged are provided by municipalities. Therefore, it can be forecast that financial demand caused by aging is generated in municipalities. In the same way as can be seen in Figures 6~9, we can see that the change in nominal prefecture or municipality expenditure from FY2006 to FY2030 is decomposed into the change of expenditure caused by price changes and in real terms by both factors of population and aging in each administrative division in Figures 10~13 7 As shown in Figure 10 and Figure 12, the contribution of real expenditure is negative only in Hokkaido and Osaka Prefecture. It can be seen that population decline and aging contribute to the change in real expenditure in each administrative division in Figure 11 and Figure 13. In FY2030, population decline causes reductions of 0.26 trillion yen and 0.34 trillion yen, respectively, in Hokkaido and Osaka Prefecture. Likewise, it causes reductions of 0.73 trillion yen and 0.94 trillion yen, respectively, in Hokkaido and Osaka s municipalities. The level of expenditure increase caused by population 7 The data for municipalities is combined with that for prefectures. 80

11 decline is far higher than that of population aging in only Hokkaido and Osaka Pref. The reason is that Hokkaido and Osaka Prefecture have relatively large-scale population declines. It is forecasted that there will be a decline of 847,000 persons in Hokkaido and 1,097,000 persons in Osaka Prefecture from FY2006 to FY2030. This means a substantial reduction in the financial load compared to the present situation in Hokkaido and Osaka Prefecture. Conversely, there will be population increases in Shiga Prefecture and Okinawa Prefecture, although the total population of Japan will decrease according to the data of IPSS by FY2030. In Shiga Prefecture and Okinawa Prefecture, population increases respectively by 67,000 persons and 125,000 persons. Therefore, an additional increase in financial demand caused by the population increase is generated in Shiga Prefecture and Okinawa Prefecture. 81

12 Note: Exp(NGR) is expenditure extended by nominal GDP growth rate. Figure 4: Forecast of prefectural expenditure Note: Exp(NGR) is expenditure extended by nominal GDP growth rate. Figure5: Forecast of municipal expenditure 82

13 Prefecture Table.1: Forecast of local expenditure Municipality Note: Unit: Trillion Yen %(percentage change in FY2030 compared to FY2006 ) 83

14 Note: The change from FY2006 to FY2030. Figures in parentheses are percentages. Figure 6: Contribution of factors of nominal prefectural expenditure change Note: The change from FY2006 tofy2030. Figures in parentheses are percentages. Figure 7: Contribution of factors of real prefectural expenditure change 84

15 Note: The change from FY2006 to FY2030. Figures in parentheses are percentages. Figure 8 Contribution of factors of nominal municipal expenditure change Note: The change from FY2006 to FY2030. Figures in parentheses are percentages. Figure 9 Contribution of factors of real municipal expenditure change 85

16 Note: The change from FY2006 to FY2030. Figure 10 Contribution of factors of nominal prefectural expenditure change Note: The change from FY2006 to FY2030. Figure 11. Contribution of factors of real prefectural expenditure change 86

17 Note: The change from FY2006 tofy2030. Figure 12 Contribution of factors of real municipal expenditure change (division of prefecture) Note: The change from FY2006 to FY2030. Figure 13 Contribution of factors of real municipal expenditure change (division of prefecture) 87

18 4. Conclusion General Expenditure increases by 33.9% (+15.7 trillion yen) in FY2030 from FY2006. This consists of contributions of +36.8%(+17.1 trillion yen) from price changes and -2.8% (-1.3 trillion yen) change of real expenditure. The impacts of population factor and aging factor are -22.3% (-10.3 trillion yen) and +19.4% (+9.0 trillion yen), respectively. Prefecture expenditure increases by 55.0% (+25.5 trillion yen) in FY2030 from FY2006. This consists of contributions of %( trillion yen) from price changes and %(+6.4 trillion yen) change of real expenditure. The impacts of the population factor and aging factor are -6.8%(-3.1 trillion yen) and %(+9.5 trillion yen), respectively. Municipality expenditure increases by 57.5% (+27.7 trillion yen) in FY2030 from FY2006. This consists of contributions of +41.8% (+20.1 trillion yen) from price changes and +15.7% (+7.6 trillion yen) change of real expenditure. The impacts of the population factor and aging factor are -18.0% (-8.6 trillion yen) and +33.7% (+16.2 trillion yen), respectively. Compared to the influences that population decline and population aging have on real expenditures of prefecture and municipalities, both population decline and population aging have major influences on municipalities. The level of the expenditure increase caused by population aging is far higher than that of population decline, although in municipalities the level of expenditure reduction caused by population decline is relatively high. The reason why the influence of aging is large in municipalities is that a lot of welfare services for the aged are provided by municipalities. Therefore, it can be forecast that financial demand caused by aging is generated in municipalities. For both central and local expenditures, expenditures that only expand at the nominal economic growth rate trace a higher growth route than expenditures based on the factors of population and aging. Thus, this might make the government choose inappropriate policies. In reality, expenditure that became unnecessary because of population decline, as expenditure has increased with population increase, is not necessarily reduced. For instance, most education expenses are personnel expenses for teachers. The level of education expenses is strongly sustained by employment security and aging of teachers, regardless of any decrease in the number of students. 88

19 A case in which expenditure does not decrease with population decline can occur due to the short lag that occurs when taking action for public policy. However, this rigidity of expenditure does not necessarily continue into the future. If consolidation and reorganization of schools are promoted and the personnel system becomes flexible, education expenses can be expected to decrease in the future with a declining population of children. Therefore, we consider that the forecast here reflects reality more precisely when public-sector reform progresses and the expenditures are reviewed effectively as required. 89

20 References Hiroahi, Hayashi (1995) Sozei seisaku no keiryo bunseki {The positive analysis of tax policy}, Nihonhyoron sha. Jun-Ichi, Nagamine(2000) Chihou koufuzei no santei kouzou/haibun kouzou ni kansuru bunseki {The analysis of calculation and allocation structure in Local Allocation Tax}, Koukyou sentaku no kenkyu {Public Choice Studies}, Vol.35,Keiso-Shobo. National Institute of Social Security Research (2000) Population Projections by Prefectures - (2000) - (2030) - March in 2002 Association of Local Public Statistics of Local Public Finance annual report The Ministry of Finance Statistics of Public Finance 90

21 Appendix 1. Relations between expenditure and nominal rate of economic growth In general, the Japanese government medium-term budget forecast is based only on a medium-term economic perspective (e.g. nominal GDP growth rate). Thus, this might make the government choose inappropriate policies. As is shown in Figure 14, the year-to-year percentage change at annual rates of actual expenditures of central and local governments, except debt servicing, has no clear relation with that of the nominal GDP growth rate from FY1985 to FY2003. (The correlation coefficient between nominal GDP growth rate and year-to-year percentage change at annual rates of actual expenditures are (central), (prefectures), and 0.515(municipalities) respectively. It seems that expenditure of local governments has a relation to rate of nominal GDP growth, but not with national general expenditure.) Figures 15~17 show that the actual expenditure of central and local governments (represented by solid line), and expenditure (NGR) that is increased by the rate of economic growth (represented by dotted line). It does not strictly correlate to actual expenditure and expenditure that is increased by the rate of nominal GDP growth. The gap between actual expenditure and expenditure that is increased by the rate of nominal GDP growth grows mainly in the long-term. Moreover, there is the tendency for actual expenditure to grow faster than expenditure that is increased by the rate of nominal GDP growth. The analysis of the decision factors of expenditure in this paper has no significant relation between General Expenditure of prefectures and municipalities, and GDP growth rate 8. 8 Therefore, they are excluded from explanation variables. The Table 2 shows the results of an estimation of central government expenditure. 91

22 Figure 14 Correlation between nominal GDP growth rate and actual expenditure change rate Note: Base 85 means General Expenditure (NGR) that takes FY1985 as the starting point, and is extended by nominal GDP growth rate. Figure 15 Comparison between actual central expenditure and NGR 92

23 Note: Same as Figure 15. Figure 16 Comparison between actual prefectural expenditure and NGR Note: Same as Figure 15. Figure 17 Comparison between actual municipal expenditure and NGR 93

24 2. Results of estimation by expenditure model We explain the estimation results in detail. First, we analyze the factors deciding expenditure of central government using the following regression model. In addition, to confirm whether or not economic growth rate influences expenditure, we also adopt a model (E2) by which real economic growth rate (PEGR) is added to the explanatory variables. (E1) (E2) Where t=fy1964,,fy2004,regr= Real Economic Growth Rate It is defined that u is the error term. We assume AR1. We adopt the Prais-Winsten method to solve the problem of serial correlation. The estimate results are shown in Table 2. In the E1 model, the null hypothesis of =0 is accepted with DW Statistics=1.714 according to the result of the Durbin-Watson Test, and the auto-correlation has been removed. The signs of the coefficient of population and rate of aging are both positive, and these variables affect expenditure significantly. In the E2 model, which adds real economic growth rate (PGR) to the explanatory variables, the null hypothesis of =0 is accepted with DW Statistics=1.712 as the result of the Durbin-Watson Test, and the auto-correlation has been removed. The real economic growth rate does not affect expenditure significantly. On the other hand the signs of the coefficient of population and rate of aging are both positive, and these variables affect expenditure significantly. Next, we analyze the factors to decide expenditure of local governments using the following regression model. (E3) Where t=fy 1986,, FY 2002, i = 1,,47(Prefectures) We define u i as an element that shows individual effects. is other disturbance elements and i.i.d. And, we use data on expenditure except public debt servicing expenses. We use the panel data set consisting of 47 Prefectures observed every year for 17 94

25 years (FY1986 to FY2002) a balanced panel. We adopt the fixed effect model. Table 3 shows a detailed estimate obtained by the fixed effect model. We can confirm the existence of individual effects by comparing the Least Squares Dummy Variable (LSDV) model with the Without Dummy model. As a result, the F-statistic (F(n-1, nt-n-k+1)) for testing the joint significance of individual effects is strongly in favor of the prefecture specific effect in the data. Here, n is the number of individuals, T is period, and k is the number of explanation variables. As shown in Table 3, the signs of the coefficient of population and rate of aging are both positive, and these variables affect expenditure significantly. Moreover, we can also validate our estimation model from the F-test. Finally, the expenditure estimated is evaluated in FY 2000 prices. Therefore, we must deflate them to the change nominal value using the forecast GDP deflator. The forecasting GDP deflator is shown in Figure 18. Table 2. Estimation Results (E1) Dependent Variable: Real General Expenditure (E2) Dependent Variable: Real General Expenditure Note: D.W. = Durbin-Watson Statistics. N= Number of observations. 95

26 Table 3 Estimation Results:Local Government Prefecture Municipality Note: Fixed effect model. Id-test is a test for fixed effect. F statistic for testing the joint significance of individual effects is F(n-1,nT-n-k+1). 96

27 Fig 18 GDP Deflator 97