Commodity taxation and regulatory competition

Figure 1 limits to the period 2004–2014 only, due to constraints to the availability of the Doing Business measure of product market regulation. However, a similar figure can be drawn starting from the 1970s, using the OECD ETCR measure of product market regulation.

A reform on tax rates often requires specification of a single tax figure on which parliament votes, while a regulatory reform involves a long and cumbersome analysis of a nexus of laws and decrees and raises many industry-specific contentions before any vote is held.

There are some studies in environmental economics such as Oates and Schwab’s (1988) or List and Gerking’s (2000) which discuss the impact of environmental regulations on taxes and welfare.

Our main results remain with rent-seeking regulators. Calculations are available upon request.

Terms of trade effects are present in Mintz and Tulkens (1986) and Lockwood (2001). By contrast, Keen and Lahiri (1998), Andersson and Forslid (2003), Haufler and Pfluger (2004), Haufler and Pfluger (2007) assume a numeraire good that dampens terms of trade effects.

Although it is commonly used in international trade theory, Krugman’s (1981) framework is not much discussed in the international tax competition literature where entry issues have not been extensively addressed (Haufler and Pfluger 2004)

In the OECD countries, public procurement ranges between 10% and 30% of GDP and between 20% and 50% of government expenditures (European Commission 2011). Public procurement is an even larger share when interpersonal transfers are excluded from government expenditures.

Regulatory agencies or bodies implement complex market regulatory and supervisory tasks which require economic expertise. To avoid political interference and opportunism, regulatory agencies are generally independent of other branches of government. Some examples of regulatory agencies are the Interstate Commerce Commission and the Food and Drug Administration in the USA, Ofcom in the UK, and AGCOM in Italy.

The parameter \(\theta \) can also be interpreted as a product quality shifter. Accordingly, regulation increases the product quality of national output.

In the OECD countries, public procurement ranges between 20% and 50% of government expenditures (European Commission 2011). Also, in modern economies, public goods and services are usually delivered by small independent public agencies like justice courts, public schools, and universities. One can interpret our setting more narrowly as public procurement or outsourcing.

This is the case for Gorman preferences like quasi-linear or Cobb–Douglas preferences.

See Durevall and Henrekson (2011) for an evaluation of econometric studies on Wagner’s law.

The constant is given by \(K=\left( \frac{f_{0}}{2\beta }\right) ^{2-\frac{1-(\sigma -1)\nu }{\sigma -1} }\cdot \sigma ^{\frac{\sigma +1-(\sigma -1)\nu }{\sigma -1}-2}\cdot (2\sigma -1)^{\frac{\sigma }{\sigma -1}-2} \big [ (2\sigma -1)^{2}-(\sigma -1)\nu \big ] \cdot \)\(\left[ 2\sigma -1-\sigma (\sigma -1)\nu \right] ^{\frac{1-(\sigma -1)\nu }{\sigma -1}-2}\cdot \)\(\left[ 2\sigma ^{2}+\sigma -1-\sigma (\sigma -1)\nu \right] ^{2-\frac{1-(\sigma -1)\nu }{\sigma -1}-\frac{\sigma }{\sigma -1}}\) where all terms under parentheses are positive.

We have \(\frac{\mathrm {d}}{\mathrm {d}\nu }\frac{\mathrm {d}n}{\mathrm {d}n^{*}}>0\) if and only if \(-\left( \sigma +1\right) \left( 2\sigma -1\right) ^{3}+2\nu \left( \sigma ^{2}-1\right) \left( 2\sigma -1\right) +2\nu ^{2}\sigma \left( \sigma ^{2}-\sigma -1\right) \left( \sigma -1\right) ^{2}<0\). This polynomial function of \(\nu \) is negative at \(v=0\) and \(v=1/(\sigma -1)\). For \(\sigma >1.\,618\), we have \(\left( \sigma ^{2}-\sigma -1\right) >0\) so that it is convex and must therefore take negative values for \(\nu \in [0,1/(\sigma -1)]\). For \(\sigma <1.\,618\), it is concave and reaches its maximum at \(\nu =\frac{-\left( 2\sigma -1\right) \left( \sigma +1\right) }{2\sigma \left( \sigma -1\right) \left( \sigma ^{2} -\sigma -1\right) }>\frac{1}{\sigma -1}\). So, the polynomial function is also negative \(\nu \in [0,1/(\sigma -1)]\).

The corresponding analysis of these extensions can be found in Section A of online appendix of the paper.

Equation (18) itself does not have any implication for the specific principle of commodity taxation. Application of the destination principle is guaranteed by the exclusion of origin-based taxes (e.g., excises) from the computation of \(\tau _{it}\) and by the choice of a weighting matrix that minimizes origin-based strategic interactions due to cross-border shopping. See more on this below.

The idea that the implementation of product market reforms takes at least 1 year is consistent with descriptive evidence for the OECD countries (see Conway and Nicoletti 2006 and the World Bank’s Doing Business report World Bank 2007).

The reflection problem arises whenever strategic interactions occur among countries in a fixed reference group. Our weighting matrix specifies a different reference group for each country (importer), as shown in Table 5. For example, our matrix accounts for the fact that trade relationships are stronger between countries with common legal origins (e.g., Belgium, Spain, France, Greece, Italy, the Netherlands, and Portugal, which have common French legal origins) or common language (e.g., Austria and Germany which are German speaking). It accounts for the fact that a EU country may have stronger trade links with other EU countries, relative to non-EU ones (since EU countries are geographically closer and are more likely to share common legal and colonial origins).

For example, our weights are not affected by strategic interactions between tax policies that might affect the size of trade flows. They also are not affected by the product market regulations in specific sectors (e.g., energy, transport, postal services) that influence international transportation costs.

The countries we consider are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, the UK, and the USA. Section B.1 in online appendix describes the data sources and the construction of variables. It also presents our control variables and/or the variables used for the robustness checks.

We exclude excise taxes, customs and import duties, profits from public monopolies, and taxes on specific services whose revenues may partly reflect application of the origin principle to consumers’ transactions.

The long time span of this series makes it well suited to a panel study. In the robustness analysis, we also proxy regulation by the number of days required to start up a new business (see Djankov et al. 2002). This measure applies to the whole economy but is available for the shorter time period 1999–2008. The two series have a correlation coefficient equal to 0.50 and significant at the \(1\%\) level (see Section B.1 of online appendix).

WVS and EVS data consist of fully comparable survey waves. They describe social attitudes, which are persistent in each country over the years covered. Thus, it can be argued that social preferences change between two consecutive waves, while remaining constant in the years covered by each single wave. (Details are given in Section B.1 of online appendix).

At the end of the 1980s, social unrest increased in Finland and Sweden due to the rise of social equality movements and the contrast between Swedish majority and minority groups. Also, the collapse of the Soviet Union and the ensuing great economic depression in the first half of the 1990s increased the demand for social, political, and economic stabilization. Similarly, in Belgium demand for order increased as a consequence of serial crime episodes and the dioxin food crisis during the mid-1990s.

At the beginning of the 1990s, the levels of trust in Australia were very low driven mostly by political inertia and economic depression. The rise in trust level may be attributed to the election of a liberal government that launched successful waves of liberalization and structural reforms. The path was somewhat similar in New Zealand where, up to the early 1990s, national governments carried out reforms that may not have reflected the mood of the electorate. The rising level of trust in the 2000s may reflect the success and application of a referendum in favor of a mixed-proportional political representation (Castles and Mitchell 1993).

It is generally acknowledged that the shock that triggered the resurgence of distrust in these countries during the 1990s was the fall of the Communist regime in Russia and Eastern Europe. Also, the fall of the Communist system and the Yugoslav wars fostered fear and opposition to rising immigration. Immigrants were often perceived as ‘dangerous’ to national communities, and this has led to the election of governments supported by extreme right and nationalist parties (see European Commission Canoy et al. 2006). In Italy, the increase in distrust and demand for freedom and autonomy was also triggered by the ‘Mani Pulite’ political scandal.

The larger effect reported in the 2SLS estimation suggests that the OLS estimate in Column [2] is upward biased as should be the case if governments introduce policy packages that jointly increase regulation levels and commodity taxes. Notice that in Table 8 of Appendix B we show the negative impact of product market regulation holds on the commodity tax revenue. This isolates factors affecting only the numerator of the commodity tax rate.

In the model government regulatory agencies set firms’ entry requirements, and then governments set their commodity tax rates, as regulatory agencies’ processes and standards are more difficult to (re-)structure than commodity tax rates.

For example, reverse causality may go from consumption taxes toward imports’ shares if the level of taxes in country i influences the decision of country j to export to country i or in some other country \(-i\). Along similar lines, product market regulation in country j determines the relative prices of its goods, thus influencing the decision of country i over whether to import from j or from some other country \(-j\). Omitted variable bias may arise if unobserved structural characteristics exist in a country which affect both its over time variation in taxation, regulation, and imports.

in Eq. (B-4) the coefficient of contig is very weakly significant at the \(10\%\) probably due to the fact that in our sample of 21 OECD countries the variation in the geographical position is mostly captured by the distwces variable. Nevertheless, we decided to include contig in ( B-4) due to the strong theoretical a priori in favor of the importance of shared borders to imports.