Tax buoyancy in OECD countries: New empirical evidence

https://doi.org/10.1016/j.jmacro.2020.103189Get rights and content

Abstract

A renewed interest in the link between business cycle and tax revenues has recently emerged, especially during economic crises. In this paper, we provide an empirical analysis on 35 OECD countries over the period 1995–2016 to estimate both short-run and long-run tax buoyancies, taking into account the macroeconomic framework, changes in governments’ tax policies, budgetary and political variables possibly affecting how taxes react to GDP fluctuations. By adopting the dynamic common correlated effects estimator, we find that both short- and long-run tax responses are lower than those reported in previous cross-country studies. We suggest that this slightly lower than expected reaction of tax revenue can be interpreted as a reduced power of both automatic stabilization in the short-run and fiscal sustainability in the long-run. Results are robust to possible endogeneity issues between tax revenues and business cycles.

Introduction

In recent years, international organizations, national governments, and academics have shown a renewed interest in the way in which tax revenues react to the business cycle, especially during economic crises (Arnold et al., 2011; Acosta-Ormaechea and Yoo, 2012; Gemmell et al., 2014; Baiardi et al., 2018; Aizenman et al., 2019). Measuring this reaction is crucial both for monitoring and forecasting governments’ public finances, as it helps to predict how to implement stabilization policies in response to expansionary or contractionary stages of the economic cycle (Blanchard and Perotti, 2002; Arachi et al., 2015; Lagravinese et al., 2018), and for the general sustainability of public finances. This issue has even greater importance in the presence of financial stress – as in the last decade – where the tax revenue management has been affected by a certain degree of uncertainty and instability. In this context, how taxpayers respond either to tax incentives or to discretionary tax changes might be unpredictable (Mourre and Princen, 2019), especially when the monetary policy has limited room for country-specific macroeconomic policies as for most European countries in the Euro area.

Since the first contribution on the elasticity of tax revenues (Groves and Kahn, 1952), the literature has gradually enriched over the years. For example, many studies have estimated the elasticity of single taxes (e.g. Huton and Lambert, 1980; Fries et al., 1982; Clausing, 2007), while other contributions have provided either case-study analyses (Sobel and Holcombe, 1996; Dye 2004; Bruce et al., 2006; Wolswijk 2009; Poghosyan, 2011; Koester and Prismeier, 2012; Havranek et al., 2016; Lagravinese et al., 2018) or comparative analyses of tax buoyancy and tax elasticity using different samples of countries (Sancak et al., 2010; Brückner, 2012; Fricke and Sussmuth, 2014; Belinga et al., 2014; Dudine and Jalles, 2018; Boschi and d'Addona, 2019; Mourre and Princen, 2019).

Over time, the econometric methodologies have also been refined, moving from the Ordinary Least Squares (OLS) and Dynamic OLS (DOLS) estimators to more sophisticated econometric techniques that take into account the presence of cross-sectional dependence among panel units (Pesaran, 2006; Chudik and Pesaran, 2015). This is a crucial point for fiscal policies, which are often influenced by supranational constraints (e.g., the European rules on public budget). For this reason, our analysis is carried out using a new panel data method that takes into account unobserved heterogeneity, temporal persistence, and cross-sectional dependence. In particular, in the presence of cross-sectional dependence in the data, the output obtained with the fixed effect estimator cannot be relied upon.

Most importantly, when addressing the tax revenues reaction to the economic cycle, an important distinction between the concepts of tax buoyancy and tax elasticity should be made. Tax buoyancy is a measure of how taxes react to economic growth, without disentangling the impact of discretionary and automatic tax changes. Tax elasticity, instead, is a measure of the reaction of taxes due to the built-in flexibility of the tax system, which disregards the impact of discretionary tax changes.

Compared to the previous studies, our paper concentrates exclusively on the tax buoyancy of 35 OECD countries during the period 1995–2016. By focusing on the tax buoyancy, instead of tax elasticity, we are able to capture the reaction of both structural characteristics and discretionary tax policies to GDP changes, not least because of the debatable methods by which discretionary changes are isolated in the common practice. Indeed, estimating tax elasticities through a regression of tax changes on tax bases may fail to adequately disentangle automatic and discretionary changes, thus being not able to assess their distinctive impacts (Sen, 2006).

The time interval used in this paper is also particularly important as it crosses relevant institutional changes and economic cycles for the countries in our sample. In particular, we extend previous works that are normally based on a time span that ends at years 2013 or 2014 (Sancak et al., 2010; Belinga et al., 2014; Boschi and d'Addona, 2019; Dudine and Jalles, 2018), by adding further empirical evidence capturing the impact of the recent Great Recession on both short- and long-run responses of taxes to GDP changes.

On the institutional side, all OECD countries belonging to the European Union (EU), since 1992, have experienced tighter budget constraints and fiscal discipline, which might have conditioned the response of taxes compared to those OECD countries that do not belong to the EU. As for the economic cycle, our period includes both a severe recession and a recovery. This element is particularly important for measuring both short- and long-run tax reactions. In doing so, we also exploit both the heterogeneity of tax responses across countries and the different intensity of recessions in different countries.

From the econometric side, by limiting the analysis to OECD countries that share similar institutional systems and economic development, one can expect either weak or strong forms of cross-sectional correlation that, if ignored, may lead to biased estimates (Pesaran 2006; Pesaran and Tosetti, 2011; Everaert and De Groote, 2016). To address this issue, we adopt the Dynamic Common Correlated Effects estimator (DCCE) as developed by Ditzen (2018) to correct for small sample time-series bias (Chudik and Pesaran, 2015). The DCCE controls for dependence by adding cross-sectional means and lags (Everaert and De Groote, 2016; Pesaran and Tosetti, 2011) and by testing, at the same time, for cross-sectional dependence in the error terms.

The main results of the paper show a significant mitigation of the levels of tax buoyancy over the period investigated both in the short- and the long-run, with respect to estimates provided by previous studies. Such divergence can be mostly attributed to the different time span included in the analysis and to the novel use of the dynamic estimator in calculating short- and long-run tax responses. In detail, the short-run coefficients for total revenue display a slightly lower reaction to the economic cycle than expected for a good automatic stabilizer. Nevertheless, this average response is the outcome of a significant variability of the short-run buoyancies of specific tax items. On the other hand, when considering long-run tax buoyancies, all coefficients are significantly lower than 1. Since long-run responses would give information on the sustainability of fiscal policies, they reveal that both the aggregate of total taxes and single tax items could not guarantee fiscal sustainability in the long-run, as they do not seem to converge to a long-run equilibrium in the period analysed.

When considering the macroeconomic conditions, we note that inflation and unemployment contribute to foster tax buoyancy at least in the short-run, while they do not significantly affect estimates of long-run tax buoyancy. This would suggest that while the short-run tax reaction is impaired by automatic sources, the long-run response could be mostly affected by the combination of both automatic and discretionary changes in governments’ policy. Most interestingly, when taking into account the role of country- specific economic downturns occurred during the observed period, short- and long-run tax buoyancies are lower than those obtained with the baseline results.

Robustness checks are performed by considering changes in governments’ tax policy, and the presence of European institutions that could differently impact on the tax system of the selected countries and, eventually, on the tax reactions to the economic cycle. Finally, other robustness checks concern the use of instrumental variables to take into account for possible endogeneity issues and the use of fixed effects only to test the stability and the accuracy of our dynamic estimator, also compared to previous studies mostly based on the former.

The rest of the paper is organized as follows. Section 2 describes the importance of tax buoyancy approach used in this paper. Section 3 presents the methodological framework. Section 4 shows the main results for the long and short-run tax buoyancies, also reporting country-specific coefficients, and some robustness checks. Section 5 provides additional evidence taking into account the role of budgetary parameters, shadow economy and political variables possibly affecting how taxes react to GDP changes. Section 6 briefly concludes.

Section snippets

Why tax buoyancy?

The focus of the paper is on tax buoyancy, which is a measure of how taxes react to economic growth that does not disentangle the impact of discretionary and automatic tax changes. Controlling for discretionary tax changes, instead, would isolate tax elasticity, which is the reaction of taxes that is only due to the built-in flexibility of the tax system for a given tax structure (i.e. remaining constant when discretionary tax changes occur).1

Specification and estimation technique

The empirical specification is based on a balanced panel of 35 OECD countries over the period 1995–2016.8 To estimate tax buoyancy, we adopt an

Main results

Tables 1–8 show the results from the DCCE estimator for total revenue and disaggregated taxes as dependent variables, without and with some control variables.10

Additional empirical evidence

In this section, we provide some additional results on tax buoyancies taking into account the role of budgetary parameters, shadow economy and political variables possibly affecting how taxes react to changes in GDP. The related results for total revenue and total taxes are reported in Table 9, respectively in the upper panel and in the bottom panel. Details on each column are described in the next sub-sections.

Conclusions

The most striking result of this paper is that the long-run tax buoyancy in OECD countries is consistently below 1 for various specifications of the base model. By interpreting the long-run responses in terms of fiscal sustainability, a first conclusion emerges that both the institutional changes and the deep recession occurred in the period analysed have reduced both the power of taxes in performing automatic stabilization and their ability to grow in line with GDP growth. This would imply

References (55)

  • J.M. Arnold et al.

    Tax policy for economic recovery and growth

    Econ. J.

    (2011)
  • M. Ashraf et al.

    Institutional determinants of tax buoyancy in developing nations

    J. Emerg. Econ. Islamic Res.

    (2018)
  • D. Baiardi et al.

    Tax policy and economic growth: does it really matter?

    Int. Tax Publ. Finance

    (2018)
  • S. Barrios et al.

    Discretionary Measures and Tax Revenues in the Run-up to the Financial Crisis

    (2010)
  • R.J. Barro et al.

    Macroeconomic effects of the 2017 tax reform

    Brookings Pap Econ. Act

    (2018)
  • V. Belinga et al.

    Tax buoyancy in OECD countries

    (2014)
  • L. Bettendorf et al.

    The Stabilities of Tax Elasticities in the Netherlands

    (2013)
  • O. Blanchard et al.

    An empirical characterization of the dynamic effects of changes in government spending and taxes on output

    Q. J. Econ.

    (2002)
  • M. Boschi et al.

    The stability of tax elasticities over the business cycle in European countries

    Fiscal Stud.

    (2019)
  • D. Bruce et al.

    Tax base elasticities: a multi-state analysis of long-run and short-run dynamics

    South Econ. J.

    (2006)
  • Coppedge, M., Gerring, J., Lindberg, S.I., Skaaning, S.-E., Teorell, J., Altman, D.,Bernhard, M., Fish, M.S., Glynn,...
  • K.A. Clausing

    Corporate tax revenues in oecd countries

    Int. Tax Publ. Finance

    (2007)
  • J. Ditzen

    Estimating dynamic common-correlated effects in stata

    Stata J.

    (2018)
  • P. Dudine et al.

    How buoyant is the tax system? new evidence from a large heterogeneous panel

    J. Int. Dev.

    (2018)
  • R.F. Dye

    State revenue cyclicality

    Natl. Tax J.

    (2004)
  • R.F. Engle et al.

    Co-Integration and error correction: representation, estimation, and testing

    Econometrica

    (1987)
  • G. Everaert et al.

    Common correlated effects estimation of dynamic panels with cross-sectional dependence

    Econ. Rev.

    (2016)
  • Cited by (17)

    View all citing articles on Scopus
    View full text