Abstract
The empirical findings concerning the validity of Wagner’s Law seem to depend on both economic fundamentals and modelling choices. Using a semi-parametric model and data for a sample of five developed and five developing countries during the period 1960–2007, we find that, for the developing countries, Wagner’s Law holds regardless of whether the relationship between government expenditures and per capita income is modelled in a linear or a nonlinear fashion. For the developed countries, however, model specification matters: If a linear specification is a priori chosen, then we find support of Wagner’s Law, while a nonlinear specification implies on the contrary the opposite result. Hence, model misspecification could have led to incorrectly accept Wagner’s law for the developed countries.
Similar content being viewed by others
Notes
In addition, few other studies tested a priori for nonlinear causality between spending and income (see e.g., Karagianni and Pempetzoglou (2009).
This is important since many datasets are not large enough to ensure accurate nonparametric estimation due to the so called “curse of dimensionality problem”. Moreover, a nonparametric model makes no precise assumptions about the functional form. Instead, the data are allowed to “speak for themselves” (Delgado and Robinson 1992). A nonparametric approach is especially useful at an exploratory level to provide rough indication of (i) which variables are relevant to the analysis of a particular problem, (ii) the functional form of a regression model, or (iii) the distributional form of a disturbance random variable.
I thank a referee for pointing out this alternative.
Throughout the paper when a quotation is used, words or phrases in brackets are our own additions.
In our setting, the SCM suggests that the coefficients of trade openness and population demographics may vary with GDP per capita. Thus, the marginal effects of both trade openness and population demographics depend on the values of GDP per capita. As a result, total government spending may also be a function of GDP per capita. On the contrary, in the PLM it assumed that the slope coefficients \(\beta \) are constant hence per capita GDP can only shift the level of total government spending. In this case, per capita GDP is said to have “neutral” effects on total government spending. In contrast to PLM, the SCM allows GDP per capita to affect total government spending in a non-neutral fashion. See Li and Racine (2007, section 9.3, pp. 301–305) for a detailed account on this.
In our sample, this applies to all countries.
This is the direct counterpart of model selection for parametric approaches.
Other variables that could be used as control variables for government expenditures are the electoral systems and the government type, both available in Persson and Tabelini (1999). However, we do not include these two dummy variables in the empirical analysis as they do not change during the study period for the countries we consider and thus their impact cannot be identified econometrically in our time-series framework.
Notice that the countries included in our sample have not been members of the OECD for the whole period under consideration.
Also Groot et al. (2003) show that the variation in the quality of domestic institutions explains why OECD countries are relatively attractive trade partners.
The idea of a specification test in nonparametric models was first proposed by Ulla (1985) although Bierens (1982) was the first to construct a consistent model specification test, suggesting that one may use the difference between the residual sum of squares from the nonparametric kernel and the residual sum of squares from the parametric model. This may be seen as a natural combination since the most popular parametric model specification test is a F-test based on restricted and unrestricted residual sums of squares and the kernel method is one of the most popular nonparametric estimation methods. However, due to the lack of a satisfactory theoretical justification for the test proposed by Ulla (1985) and any other test based on comparing the nonparametric with the parametric residual sums of squares, these tests are not completely satisfactory. For example, Lee (1994) attempted to establish the asymptotic null distribution of the test statistic proposed in Ulla (1985). To avoid the degeneracy of the distribution of the test statistic under the null hypothesis, Lee (1994) used re-weighting, the consequence of which is the requirement that the errors must be conditionally homoscedastic.
References
Akitoby B, Clements B, Gupta S, Inchauste G (2006) Public spending, voracity, and Wagner’s law in developing countries. Eur J Polit Econ 22(4):908–924
Alesina A, Wacziarg R (1998) Openness, country size and government. J Public Econ 69:305–321
Bairam EI (1992) Variable elasticity and Wagner’s law. Public Finance 3:491–495
Bairan EI (1995) Level aggregation, variable elasticity and Wagner’s law. Econ Lett 48:341–344
Biehl D (1998) Wagner’s Law: an introduction to and a translation of the last version of Adolph Wagner’s text of 1911. Public Finance 53(1):102–111
Bierens HJ (1982) Consistent model specification tests. J Econom 20:105–134
Brown CV, Jackson PM (1982) Public sector economics. Martin Robertson, Oxford
Cameron D (1978) The expansion of the public economy: a comparative analysis. Am Polit Sci Rev 72(4):1243–1261
Chang T (2002) An econometric test of Wagner’s law for six countries based on cointegration and error-correction modeling techniques. Appl Econ 34:1157–1169
Delgado MA, Robinson PM (1992) Nonparametric and semi-parametric methods for economic research. J Econ Surv 6(3):201–249
Dickey DA, Fuller WA (1979) Distribution of the estimators for autoregressive time series with a unit root. J Am Stat Assoc 74:427–431
Durevall D, Henrekson M (2011) The futile quest for grand explanation of long-run government expenditure. J Public Econ 95:708–722
Fan Y, Li Q (2000) Consistent model specification tests: Kernel based tests versus Bierens’ ICM tests. Econom Theory 16:1016–1041
Ganti S, Kolluri BR (1979) Wagner’s law of public expenditures: some efficient results for the United States. Public Finance 34:225–233
Gonzalez-Eiras M, Niepelt D (2012) Aging, government budgets, retirement, and growth. Eur Econ Rev 56(1):97–115
de Groot HLF, Linders GM, Rietveld P (2003) Why do OECD-countries trade more? Tinbergen Institute Discussion Paper 092/3. Tinbergen Institute, Amsterdam
Gupta SP (1967) Public expenditure and economic growth: a time series analysis. Public Finance 22:423–471
Gupta SP (1968) Public expenditure and economic development: a cross-section analysis. Finanzarchiv 28:26–41
Härdle W, Muller M, Sperlich S, Werwatz A (2004) Nonparametric and semi-parametric models. Springer, Berlin
Henrekson M (1993) Wagner’s law: a spurious relationship? Public Finance 48:406–415
Johansen S (1988) Statistical analysis of cointegration vectors. J Econ Dyn Control 12(3):231–254
Johansen S, Juselius K (1990) Maximum likelihood estimation and inference on cointegration with applications to the demand for money. Oxf B Econ Stat 52:169–210
Karagianni S, Pempetzoglou M (2009) Evidence for non-linear causality between public spending and income in the European Union countries. J Appl Bus Res 25(1):69–82
Kolluri BR, Panik MJ, Wahab MS (2000) Government expenditure and economic growth: evidence from G7 countries. Appl Econ 32:1059–1068
Koop G, Tobias JL (2006) Semiparametric Bayesian inference in smooth coefficient models. J Econom 134:283–315
Lee BJ (1994) Asymptotic distribution of the Ullah-type specification test against the nonparametric alternatives. J Quant Econ 10:73–92
Li Q, Huang CJ, Li D, Fu TT (2002) Semiparametric smooth coefficient models. J Bus Econ Stat 20(3):412–422
Li Q, Racine JS (2007) Nonparametric econometrics: theory and practice. Princeton University Press, Princeton
Li Q, Wang S (1998) A simple consistent bootstrap test for a parametric regression function. J Econom 87:145–165
Meltzer A, Richard S (1981) A rational theory of the size of government. J Polit Econ 89(5):914–927
Oxley L (1994) Cointegration, causality and Wagner’s Law: a test for Britain 1870–1913. Scot J of Polit Econ 41:286–297
Payne JE, Ewing BT (1996) International evidence on Wagner’s hypothesis: a cointegration analysis. Public Finance 51:258–274
Peacock A, Scott A (2000) The curious attraction of Wagner’s Law. Public Choice 102:1–17
Persson T, Tabelini G (1999) The size and scope of government: comparative government with rational politicians. Eur Econ Rev 43:699–735
Phillips PC, Perron P (1988) Testing for a unit root in time series regression. Biometrika 75:335–346
Ram R (1987) Wagner’s hypothesis in time series and cross section perspectives: evidence from “real” data for 115 countries. Rev Econ Stat 69:194–204
Ram R (1992) Use of Box–Cox models for testing Wagner’s hypothesis: a critical note. Public Finance 47:496–504
Robinson P (1988) Root-N-consistent semi-parametric regression. Econometrica 56:931–954
Rodrik D (1998) Why do more open economies have bigger governments? J Polit Econ 106(5):997–1032
Shelton CA (2007) The size and composition of government expenditure. J Public Econ 91:2230–2260
Singh S, Sahni BS (1984) Causality between public expenditure and national income. Rev Econ Stat 66:630–643
Stein E, Talvi E, Grisanti A (1998) Institutional arrangements and fiscal performance: the Latin American experience. Working paper 367, Office of the Chief Economist, Inter-American Development Bank
Stock JH (1989) Nonparametric policy analysis. J Am Stat Assoc 84:567–577
Sun Y, Li Q (2011) Data-driven bandwidth selection for nonstationary semiparametric models. J Bus Econ Stat 29(4):541–551
Thornton J (1999) Cointegration, causality and Wagner’s law in 19th century Europe. Appl Econ Lett 6:413–416
Uhlenberg P (2009) International handbook of population aging. Springer, New York
Ulla A (1985) Specification analysis of econometric models. J Quant Econ 1:187–209
Wang Q, Phillips PCB (2009) Structural non-parametric cointegrating regression. Econometrica 77(6):1901–1948
Wagner A (1911) Der Staat in nationalökonomischer Hinsicht. In: Conrad J et al (eds) Handwörterbuch der Staatswissenschaften, \(3^{\rm {rd}}\) completely rev. ed., vol 7, Jena, pp 734–736
Wei S-J (2000) Natural openness and good government, NBER Working paper 7765. Cambridge, MA
Wu MA, Lin M (2012) Determinants of government size: evidence from China. Public Choice 151(1–2):255–270
Yatchew A (2003) Semi-parametric estimation for the applied econometrician. Cambridge University Press, Cambridge
Acknowledgments
I would like to thank Qiying Wang for a constructive discussion on an estimation issue and two anonymous referees for valuable comments and suggestions on earlier versions of the paper.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Unit root and cointegration tests
We use the Augmented Dickey–Fuller (ADF) test (1979) and the Phillips–Perron (PP) test (1988) to examine the time-series properties of the data. The ADF test is referred to the t-statistic of the \(\theta _2 \) coefficient on the following regression:
where \(\chi =\left\{ {y,X,Z} \right\} \). The ADF regression tests for the existence of unit root of \(\chi _t \), namely in the logarithm of all model variables at time t. The variable \(\Delta \chi _{t-1} \) expresses the first differences with k lags, and \(\varepsilon _t \) is the variable that adjusts the errors of autocorrelation. The coefficients \(\theta _0,\theta _1, \theta _2 \) and \(\lambda _i \) are estimated. The null and the alternative hypotheses for the existence of unit root in variables \(\chi _t \) are:
The PP test is an extension of the Dickey–Fuller test, which corrects for autocorrelation and is more robust in the case of weakly autocorrelated and heteroscedastic regression residuals. Also it is more powerful than the ADF test for aggregate data.
The results are shown in Table 4. We find that the variables under examination are integrated of order 1 in all ten countries. Hence, we also perform a cointegration test. The null hypothesis of non-cointegration is tested against the alternative of cointegration. We follow the maximum likelihood procedure of Johansen (1988) and Johansen and Juselius (1990) where a p-dimensional \((p\times 1)\) vector autoregressive model with Gaussian errors can be expressed by its first differenced error correction form:
where \(Y_t =[y\;\;\;X\;\;Z]\) is a \(p\times 1\) vector containing all model variables; \(\mu \) is the \(p\times 1\) vector of constant terms; \(\Gamma { }_i=-I+A_1 +A_2 +\cdots +A_i (i=1,2,\ldots ,p-1)\) is the\(p\times p\)matrix of coefficients; \(\Pi =I-A_1 -A_2 -\cdots -A_p \) is the\(p\times p\) matrix of coefficients; and \(u_t \) is the \(p\times 1\) vector of the disturbance term coefficients.
The \(\Pi \) matrix conveys information about the long-run relationship between \(Y_t \) variables, and the rank of \(\Pi \) is the number of linearly independent and stationary linear combinations of the examined variables. Thus, testing for cointegration involves testing for the rank of \(\Pi \) matrix by examining whether the eigenvalues of \(\Pi \) are significantly different from zero. Johansen (1988) and Johansen and Juselius (1990) propose two test statistics for testing the number of cointegrating vectors (or the rank of \(\Pi )\) in the VAR model. These are the trace test and the maximum eigenvalue test. The likelihood ratio statistic for the trace test is:
where \(\hat{{\lambda }}_{r+1} ,\ldots ,\hat{{\lambda }}_p \) are the \(p-r\) smallest eigenvalues. The null hypothesis to be tested is that there are at most \(r\) cointegrating vectors. That is, the number of cointegrating vectors is less than or equal to \(r\), where \(r\) is 0, 1, 2,..., and so forth. In each case, the null hypothesis is tested against the general alternative. Alternatively, the maximum eigenvalue statistic is:
In this test, the null hypothesis of \(r\) cointegrating vectors is tested against the alternative hypothesis of \(r+1\) cointegrating vectors. Thus, the null hypothesis \(r = 0\) is tested against the alternative \(r = 1, r = 1\) against the alternative \(r = 2\), and so forth. It is well known that the cointegration tests are very sensitive to the choice of lag length. We use the Schwartz criterion and the likelihood ratio test to select the number of lags required in the cointegration test (Table 5).
Rights and permissions
About this article
Cite this article
Paleologou, SM. The long-run tendency of government expenditure: a semi-parametric modelling approach. Empir Econ 50, 753–776 (2016). https://doi.org/10.1007/s00181-015-0959-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-015-0959-2