Economic Growth and Budgetary Components: a Panel Assessment for the EU

Transcription

1 Economic Growth and Budgetary Components: a Panel Assessment for the EU December, 2008 António Afonso (ECB), Juan González Alegre (UPO)

2 Outline 1. Motivation 2. Theoretical underpinnings 3. Empirical specifications 4. Empirical analysis 5. Conclusions

3 1. Motivation (1) Fiscal variables are still quite heterogeneous among member states; The Lisbon process and the need to enhance productivity growth in Europe; Shortcomings associated to past methodological approaches; The need to better understand the mechanisms that link fiscal variables and growth.

4 1. Motivation (2) Stylized facts 30 GDP A (Ireland, Greece, Luxembourg, Spain and Finland) B (Belgium, Germany, France and Italy) difference (B-A) Miles Euro per capita. Constant prices basis=2000.

5 1. Motivation (3) Stylized facts. Public Expenditure 6 Public Consumpt ion 0.10 Public Invest ment A (Ireland, Greece, Luxembourg, Spain and Finland) B (Belgium, Germany, France and Italy) difference (B-A) All variables are expressed in miles Euro per capita.

6 1. Motivation (4) Stylized facts. Public Revenues Direct taxation Social Cont ribut ions A (Ireland, Greece, Luxembourg, Spain and Finland) B (Belgium, Germany, France and Italy) difference (B-A) All variables are expressed in miles Euro per capita.

7 2. Theoretical underpinnings (1) Public Expenditure Traditionally modelled as a determinant of multifactor productivity or a separate production factor: Aschauer (1989), Barro and Sala-i-Martin (1992) Y = AK L G α γ δ t t t 1, t According to the cost-function approach, public expenditure may incentive private investment: Devarajan et al. (1996), Moreno et al. (2003) G2, = (1 s ) K t t t

8 2. Theoretical underpinnings (2) Public Expenditure Some public spending items may have a direct impact on labour markets: Agenor (2007), van der Ploeg (2006); = % µ ν η t t 3, t t L w G N It can also represent a consumption good, affecting savings rates: Turnovsky (1996); U t (1 ) j C θ = β t G θ 4, t t= j

9 2. Theoretical underpinnings (3) Public Revenues Consumption tax has an impact on growth only in the presence of externalities: Liu and Turnovski (2005); Perverse effect of capital tax, Chamley (1989); Taxes on labour income affect the labour market; Taxation should have no effect on TFP? (1 + τ ) C + s K = (1 τ ) wl + (1 τ ) π + rk τ c t t+ 1 t+ 1 l t t π t t t τ Where c is the consumption tax-rate, l is the labour-income tax-rate, and represents the corporate profits tax-rate. τ π

10 2. Theoretical underpinnings (4) Growth and productivity Public Expenditure Taxation Productivityenhancing (G 1 ) Capital enhancing (G 2 ) Labourenhancing (G 3 ) Consumption good (G 4 ) Consumption tax Labour Income tax Corporate profits tax GDP Only short term eff. Only short term eff. - - Lab. Prod Only short term eff. Only short term eff. + - TFP + No effect No effect Only short term eff. Only short term eff. No effect No effect Assuming α > 0, 0>s>1 and ν > 0

11 3. Empirical specifications (1) First models Use of (five-year) moving average data to account for long-term, Easterly and Rebelo (1993); Static linear models, estimated usually by fixed-effects; Include a large number of countries and small period range, Barro and Sala-i-Martin (1992) include 98 countries; Considerable variety of explanatory variables.

12 3. Empirical specifications (2) First critics Developed and developing countries in the same panel? (Devarajaan, Swaroop and Zou, 1996); The consequences of ignoring the budget constraint, Kneller, Bleaney and Gemmell (1999); Dependent variables may be endogenous; The moving-average process may affect the results, Levine and Renelt (1992), Bleaney et al. (2001). Recent proposals Common long-term effect and unit specific short-term effect estimated by PMGE, Bassanini and Scarpetta (2001); Common long-term and short term effects from lagged values of the explanatory variables, Romero de Avila and Strauch (2003); Autoregressive term for GDP growth, Gupta et al. (2005).

13 3. Empirical specifications (3) A dynamic framework ARDL model, lags of dependent and fiscal variables: p q y = µ + ν + Σ λ y + Σ δ fiscal + τ other + ε i, t i t j i, t j s i, t s i i, t i, t j = 1 s = 0 Estimated by GMM Arellano and Bond (1991); Compute long-term relations by assuming an economy on its steadystate: long Σ δ s s= 0 run = p 1 Σ λ q j= 1 j

14 3. Empirical specifications (4) A dynamic framework Budget constraint. The estimated coefficients describe the impact of: An increase in public expenditure financed by adjusting the omitted public expenditure variables. An increase in taxed compensated by an equivalent decrease in the omitted public revenue variables. For example, the coefficient X attached to public consumption means that an increase in public consumption of one percent over GDP, financed by an equivalent decrease in the remaining public expenditure (keeping taxation and deficit unaltered), increases the GDP growth rate by X percentage points.

15 3. Empirical analysis (1) Data Annual data for the period Source: AMECO EU15 countries, full period; EU New member states, ; Functional classification of spending,

16 4. Empirical analysis (2) (1) (2) (3) (4) (5) (6) PE consumption *** PE compensation of employees PE social transfers ** PE subsidies PE Investment PE total PR direct taxation PR social contributions PR Indirect taxation PR total Public deficit Private Investment Terms of trade Labour force growth *** *** ** *** *** *** ** *** *** *** *** *** ** *** *** ** ** ** *** *** *** ** *** *** *** *** ** An increase in public consumption of one percent over GDP, financed by an equivalent decrease in the remaining public expenditure (keeping taxation and deficit unaltered), pushes down the GDP growth rate around 0.7 pp. Population growth *** *** *** *** *** *** Observations

17 4. Empirical analysis (2) Table 4 Table 5 Table 6 Table 7 Table 9 Countries EU 15 EU 15 EU 15 NMS EU 15 Sample period Model ARDL ARDL ARDL ARDL 5-year m.a. PE consumption *** (.225) (.469) *** (.359) *** (.713) *** (.055) PE comp. of employees (.297) ** (.471) *** (.399) *** (1.35) *** (.063) PE social transfers ** (.154) ** (.287) *** (.197) *** (.664) (.043) PE subsidies.3222 (.240) ** (.378) (.530) *** (1.59) *** (.062) PE Investment ** (.327) *** (.500) (.454) (1.69) (.074) PR direct taxation (.219) (.348) (.286) ** (.929) (.057) PR social contributions ** (.254) (.488) *** (.319) *** (1.42) * (.065) PR Indirect taxation (.225) (.349) ** (.360) ** (1.22) ** (.053) Observations

18 4. Empirical analysis (3) Full sample (Table 4): Public consumption and Social transfers seem to be an obstacle for economic growth Public Investment, on the contrary, is growth-enhancing Social contributions is the source of revenues more harmful for growth Sub-samples (Tables 5 and 6): As suggested by previous results (Caselli et al., 2000, Afonso and St. Aubyn, 2007 ), there seems to exist a break around 1970 Public expenditure is becoming less productive

19 4. Empirical analysis (4) New member States (Table 7): Poor data availability Fiscal variables seem less efficient than in the EU15 sample Comparison with traditional methodology (Table 4 vs. Table 9): Some results are not extremely different from the estimations using the standard 5-year averaging method estimated by fixedeffects However most of them present dramatic changes (compensation of employees, subsidies, public investment, social contributions, indirect taxation) The estimates for the control variables also differ significantly

20 4. Empirical analysis (5) Table 4 Table 10 Table 11 Dependent variable GDP gr. Lab. Prod. gr. TFP gr. PE consumption PE compensation of employees PE social transfers PE subsidies PE Investment PR direct taxation *** (.225) (.297) ** (.154).3222 (.240) ** (.327) (.219) *** (.201) (.261) *** (.126) (.196) *** (.296) (.158) *** (.122) *** (.132) (.077) (.148) *** (.147) (.103) - The negative sign on table 11 suggests that compensation of employees is productivityretarding (G 1 ). - But the results on tables 4 and 10 would not match, unless the variable behaves simultaneously as the capital-enhancing type of public expenditure (G 2 ). PR social contributions ** (.254) *** (.184) (.122) PR Indirect taxation (.225) (.199) (.094) Observations

21 4. Empirical analysis (6) Public Expenditure Taxation Labourenhancing Productivityenhancing Capital - enhancing Consumption good Consumption tax Labour Income tax Corporate profits tax PE cons (-) PE emp (-) PE inv (-) PE cons (-) PE emp (+) PE inv (+) PE soc trans.(-) PEinv (-) PE emp PE soc transf. PR social contr PR Ind. tax PR social contr. (-) The consumption good and consumption tax behaviours are not observable in the long-term coefficients.

22 4. Empirical analysis (7) (1) (2) (3) (4) (5) PE Econ. Affairs (.326) ** (.234) PE Health *** (.344) *** (.317) PE Education (.551) *** (.503) PE Social Protection (.170) *** (.189) Total Public revenues *** (.137) *** (.128) *** (.134) *** (.129) (.117) Public Deficit (.140) * (.112) (.131) (.125) (.120) Omitted variables: Private Investment * (.001) *** (.001) * (.001) ** (.001) ** (.001) General Public Services, Defence Labor force growth (.205) (.179) (.199) (.171) (.156) Public Order and Safety Environment Protection Terms of Trade * (.0007) (.0006) * (.0007) (.0006) ** (.0006) Housing and Community amenities Population growth (.913) * (.745) (.872) (.752) (.630) Recreation, Culture and Religion Observations

23 5. Conclusions (1) The omission of relevant dynamics in growth regressions may be an important source of bias. The identification of the source of the enhancing/retarding effect of some fiscal variables could be possible with the analysis of alternative dependent variables. There seems to exist a change on the impact of public expenditure. Our results confirms previous findings. In the 12 New Member States of the EU, fiscal policy seems to be less helpful in enhancing growth.

24 5. Conclusions (2) Public consumption retards economic growth, and have a negative impact on productivity. Public wages also affects negatively productivity, but incentives capital accumulation. Social transfers, on the contrary, is capital retarding. Public Investment have variable impact on the alternative variables, with an important evidence of a crowding-in effect on private investment. Social contributions have a behaviour similar to social transfers, while the other categories of public revenues seem not to have a relevant impact on growth.