The interest in producing aggregate tax rates may go beyond mechanical comparisons across countries and the assessments of the impact of tax reforms on tax burdens. For instance, the Presidency Conclusions of the Lisbon European Council requested the Council and the Commission to assess “the contribution of public finances to growth and employment”. In the framework of taxation, such a request implies determining the (distortionary) effects of the tax burden and its components on economic efficiency (investment, and employment) and growth. From an economic policy viewpoint, the issue at stake is the effectiveness of tax reform as an instrument to enhance long-run growth and job creation.

The academic literature is rich in papers and monographs with theoretical and empirical evidence on the impacts of taxes on growth and employment. Taxation affects the functioning of the economy in different ways. Taxes interfere with the

incentives to invest, save and work and have sizeable welfare effects49. Capital taxes reduce the after-tax rate of return on physical capital, thus lowering capital accumulation. Capital taxes may also create disincentives to save over the life-cycle. On the other hand, labour taxes, when coupled with rigid labour markets and generous benefit systems, increase the cost of labour, reduce employment and increase unemployment50. Therefore, tax systems may be a key factor to determine the overall efficiency of the economy.

However, such a body of literature is far from generating unanimity about the capacity of taxes to influence long-run growth in a significant way. Where growth effects are concerned, and although economic models predict that changes in tax rates affect growth in the long run51, as far back as in 1964, Harberger52conjectured that in practice tax policy is not effective in influencing growth. Since then, a debate exists in the empirical literature about the size and statistical significance of the effects of taxes on growth. All in all, while taxes and, particularly, capital taxes seem to have a sizeable and unambiguous negative effect on private investment, the effects on long- run growth, measured through the real GDP or the per capita GDP, may be statistically non-significant, or negative but small (Mendoza, Milessi-Ferretti and Asea, 1997). Moreover, it seems that estimates of such growth effects very much depend on the country sample and the period covered. They are also very sensitive to the inclusion or exclusion of other variables in the models, to the instruments used to avoid endogeneity biases, and to the estimation methods applied (see, for example, Agell, Lindh and Ohlsson, 1997, Folster and Henrekson, 1999, and Agell, Lindh and Ohlsson, 1999).

Where the effects of taxation on employment and unemployment are concerned, a body of economic literature argues that the tax burden on labour will be passed on to labour itself, at least in the long-run (see, for instance, Nickell, 1997). Such theoretical arguments have found support in a number of empirical analyses such as, for instance, Layard, Nickell and Jackman (1991) or, Nickell and Layard (1999), which found rather small negative effects of taxes on (un)employment performance. However, other empirical analysis, such as, Elmeskov et al (1999), Blanchard and Wolfers (1999) or Daveri and Tabellini (2000), which deal with the interplay of taxes with benefits and other labour and product market institutions, suggest that such negative effects may be sizeable. Moreover, simulation experiments on the basis of general equilibrium models published in PFR2000 (see European Commission, 2000a) suggest that a reduction in the tax burden of 1% of GDP could increase employment by 0.5 to 1% in the long run (10 years) depending on the characteristics of the tax reform (across-the-board tax cuts, targeted to labour)53.

49See, among many others, de la Fuente (1997), Mendoza et al (1997), Leibfritz, Thornton and Bibbee (1997) and Pissarides (1998).

50 See, Nickell (1997), Daveri and Tabellini (2000), Elmeskov et al (1999) and Blanchard and Wolfers (1999).

51 This particularly applies to endogenous growth models that take account of the whole structure of expenditures and taxes. For instance, Barro (1990) predicts that the impact of fiscal policy on growth depends on the structure of the budget. Kneller, Bleaney and Gemmell (1999) present very interesting empirical evidence in this line. They have shown that, once the government budget constraint is adequately specified, the effects of tax reform on long-run growth depend on how it is financed. 52 See Harberger (1964).

53 Where comparisons with results presented below are concerned, it is worth bearing in mind that such an accumulated impact on employment implies a very low additional annual growth rate of

It is worth clarifying at the outset that the purpose of this section is not to contribute to the debate of the effects of taxes on economic efficiency and long-run growth. This is, for obvious reasons, beyond the scope of this paper. Within a much modest scope, this section only tries to see to what extent the estimated effects of taxes on investment, employment and long-run growth are sensitive to alternative criteria of calculating effective tax rates.

Table 17 shows the results of simple regressions between ECFIN tax rates, on the one hand, and the private investment rate, the employment rate, the unemployment rate and the real GDP growth rate for a panel data involving the 15 Member States, the US and Japan over the 1970-1997 period54.

Table 17. Investment, growth and (un)employment effects. ECFIN tax rates, 1970-1997. Investment rate Growth rate Employment rate Unemployment rate Coef. (t-ratio) Coef. (t-ratio) Coef. (t-ratio) Coef. (t-ratio) LETR -0.13 (-4.95)* 0.20 -0.07 (-4.89)* 0.19 0.05 (0.52) 0.00 0.05 (1.17) 0.01 LITR -0.12 (-4.05)* 0.14 -0.08 (-5.35)* 0.22 -0.09 (-0.98) 0.01 0.06 (0.06) 0.02 CETR -0.06 (-1.57) 0.02 -0.02 (-1.03) 0.01 CITR -0.09 (-1.56) 0.02 -0.03 (-0.99) 0.01 KETG -0.24 (-5.82)* 0.25 -0.00 (-0.22) 0.00 KETN -0.05 (-3.23)* 0.09 -0.02 (-2.45)* 0.06 KITG -0.27 (-7.51)* 0.36 -0.05 (-1.83) 0.03 KITN -0.12 (-6.39)* 0.29 -0.03 (-2.81)* 0.07 * Significant at 5%

In order to capture long-run or trend effects, we have eliminated cyclical fluctuations by taking five-year averages over the period 1970-1994 (1970-1974, 1975-1979, 1980-1984, 1985-1989 and 1990-1994) plus the three-year averages for the period 1995-1997. The table only shows the estimate (and the t-ratio) of the coefficient of the tax rate in row in the regression of the indicator in column. The R² of the regression is displayed in bold. Econometric results in the table are very much in line with the literature mentioned above.

First, labour and capital taxes negatively affect the private investment rate in the long run. However, consumption taxes seem to be rather neutral with respect to investment55. Leaving aside KETN, the explicative power of capital taxes is larger than that of labour taxes. All in all, an increase of 1 percentage point in capital tax rates may reduce the investment rate by a quarter of point. The same increase in labour taxes leads to half such a reduction in the investment rate (slightly more than a tenth of a percentage point).

Second, taxes are overall negatively correlated with growth. However, their effects on growth are lower than those on investment. In the case of labour taxes, an increase of a percentage point in the effective tax rate on labour reduces growth in the long run by less than 0.1%. In the case of capital taxes, the effects on growth are negligible when compared with those on investment. In some cases, such effects are statistically non-significant56.

Finally, and unsurprisingly, since our “models” do not take account of the interplay of labour taxes with benefits and other market institutions, there are no significant effects of taxes on the employment and the unemployment rates57. Overall, when excluding any other potential explicative variable, labour tax rates seem to be neutral with respect to employment performance.

There is a question now to ask whether such conclusions are supported by the tax rates calculated by the OECD or by EC2000. Table 18 compares the investment, growth and employment effects of changes in the tax rates in ECFIN and in EC2000.

Table 18. Comparing investment, growth and (un)employment effects of ECFIN and EC2000 tax rates

Investment rate Growth rate Employment rate Unemployment rate

Coef. Coef. Coef. Coef.

LITR -0.09* 0.08 -0.08* 0.16 0.16 0.03 -0.03 0.00 LT1T -0.09* 0.11 -0.05* 0.09 0.18 0.04 0.02 0.00 LT2T -0.09* 0.12 -0.05* 0.07 0.22* 0.06 0.01 0.00 CITR -0.08 0.02 0.00 0.00 CTRT -0.09 0.02 0.06 0.03 KITN -0.09* 0.23 -0.01 0.01 KT1T -0.06* 0.13 -0.02 0.04 KT2T -0.08* 0.11 -0.02 0.01 * Significant at 5%

55 Mendoza et al (1997) also found zero correlation between consumption taxes and investment. 56

See Kneller, Bleaney and Gemmell (1999) for an interpretation of this result, which would be due to absence of the government budget constraint in the model.

57 Therefore, the results in this section are not comparable with simulations in, for instance, European Commission (2000a). Models in table 17 represent, by definition, partial analyses, while simulation experiments in “Public Finances in EMU –2000” give general equilibrium effects, which take account

In particular, we compare effects of the two labour implicit tax rates published in European Commission (2000c) with their best counterpart in the ECFIN databank – LITR-. In the case of consumption, we compare the EC2000 CTRT with CITR in (28). Finally, we compare the EC2000 capital rates KT1T and KT2T with the ECFIN rate KITN as defined in (39). It is worth keeping in mind that regression results in table 18 may not be comparable with those in table 17; the reason being that the sample of countries and the periods are different in both sets of regressions. Since EC2000 only includes the 15 Member States, samples in table 18 exclude the US and Japan. Moreover, unlikely in table 17, the period for some countries is 1980-1997 instead of 1970-1997. The main conclusions can be summarised as follows:

First, where investment effects of taxation are concerned, both ECFIN and EC2000 tax rates lead to almost exactly the same results. Overall, an increase of 1 percentage point in labour or capital tax rates would reduce the private investment rate by a (statistically significant) figure slightly lower than a tenth of a percentage point. However, ECFIN and EC2000 consumption tax rates are neutral with respect to investment.

Second, capital and consumption taxes seem to be neutral with respect to long-run growth. As in table 17, the effects of capital taxes on growth are negligible and, in any case, much lower than those on investment. In accordance with table 17, the effects of labour taxes on growth seem to be significant, although lower than those on investment. The explicative power of ECFIN’s LITR is slightly higher than that of its counterparts in EC2000, LT1T and LT2T.

Finally, leaving aside LT2T in the regression for the employment rate, which exhibits a (wrong) positive sign, likely due to specification errors, the effects of labour taxes on employment and unemployment are statistically insignificant. Consequently, such results seem to suggest that when it comes to analyse growth and employment effects of taxation, ECFIN and EC2000 databanks lead to comparable conclusions.

Table 19 is the equivalent to table 18 for the comparisons between ECFIN and OECD tax rates.

Table 19. Comparing investment, growth and (un)employment effects of ECFIN and OECD tax rates

Investment rate Growth Rate Employment rate Unemployment rate

Coef. Coef. Coef. Coef.

LETR -0.05 0.04 -0.05* 0.11 0.07 0.00 -0.02 0.00 LTRO -0.04 0.03 -0.04* 0.07 0.17 0.03 -0.04 0.01 CETR -0.04 0.01 -0.01 0.00 CT1O -0.06 0.02 0.01 0.00 CT2O -0.05 0.03 0.00 0.00 KETG -0.11 0.05 0.05 0.03 KTGO -0.07 0.05 -0.02 0.01 KETN 0.02 0.01 -0.02 0.03 KTNO 0.00 0.00 -0.01 0.02 * Significant at 5%

Again, the results in table 19 are not comparable with those in tables 17 and 18. On the one hand, the country sample is different since Luxembourg is not included in it, but the US and Japan are in the sample. On the other hand, the time span only starts in 1980 and there are no data for some countries during the eighties.

Overall, results in the table indicate that ECFIN and OECD tax rates lead to the same type of efficiency and growth effects.

• No tax rate has significant effects on the investment rate. Yet, tax rates of capital, measured on the basis of the gross operating surplus, seem to have larger (and negative) effects than those calculated on the basis of the net operating surplus. • Albeit rather small, labour taxes have negative effects on long-run growth. An

increase of 1 percentage point in the effective tax rate reduce growth in the long run by only 0.05%. In the sample of table 19, consumption and capital taxes do not affect long-run growth.

• Finally, the employment and the unemployment rates are invariant with respect to the labour effective tax rates produced by both ECFIN and OECD.

In document ECFIN's effective tax rates. Properties and comparisons with other tax indicators. Economic Papers No. 146, October 2000 (Page 51-56)