The Role of Market Structure and Timing in Determining VAT Pass-Through

Contributor Notes

We examine the role of market characteristics and timing in explaining observed heterogeneity in VAT pass-through. We first extend existing theory to characterize the roles of imperfect competition and product differentiation, then investigate these relationships empirically using a panel of 14 Eurozone countries between 1999 and 2013. We find important roles for product market regulation and product quality, and little impact of advance announcement of reforms. Our findings have important implications for policy-makers considering VAT rate adjustments, by illuminating which of the consumers or the producers would experience the brunt of a reform across different settings.

Abstract

We examine the role of market characteristics and timing in explaining observed heterogeneity in VAT pass-through. We first extend existing theory to characterize the roles of imperfect competition and product differentiation, then investigate these relationships empirically using a panel of 14 Eurozone countries between 1999 and 2013. We find important roles for product market regulation and product quality, and little impact of advance announcement of reforms. Our findings have important implications for policy-makers considering VAT rate adjustments, by illuminating which of the consumers or the producers would experience the brunt of a reform across different settings.

I. Introduction

Value added taxes raise about a fifth of total tax revenues both worldwide and among the members of the OECD (OECD, 2018). Given the relative ease of modifying the rates, they are frequently at the center of policy debates during economic crises – whether for fiscal stimulus (as in the 2009 VAT reform in China) or for domestic revenue mobilization (as in Europe in the 2010s). The current Covid-19 pandemic has led more than 80 countries to undertake VAT reforms, ranging from a temporary 3pp cut in the standard rate of VAT in Germany (to stimulate consumer demand) to a tripling of the VAT rate in Saudi Arabia (to repair state revenues after the oil price crash).2

How the impact of a VAT change will be divided between firms and consumers is critical for policymakers aiming to target their support or to minimize the tax burden for one group relative to the other. Who bears the consequences of a VAT reform is governed by the key parameter of ‘pass-through’ – the elasticity of consumer prices with respect to the VAT rate. Full pass-through to consumer prices of a VAT cut implies that producer prices remain unchanged, with the benefits of lower VAT accruing disproportionally to the consumer, while zero pass-through implies the opposite.3

There is a vast literature estimating the impact of VAT changes on prices. Yet, estimates of VAT pass-through to consumer prices can vary greatly across studies. 4 This paper builds on the recent empirical methodology of Benedek, De Mooij, Keen and Wingender (2015, hereafter BDKW) to explain how differences in VAT pass-through can be related to differences in market characteristics.5 Specifically, we examine the role of market competition, product differentiation and timing in explaining heterogenous VAT pass-through. We find that VAT pass-through is greater for products requiring inputs produced more competitively and for products with greater scope for vertical differentiation, namely quality. We do not find any significant difference in pass-through for reforms announced more in advance.

We start by extending existing theory to identify how supply and demand features can influence the degree of VAT pass-through under different market structures. We develop four simple partial equilibrium models. We first consider equally productive firms competing in prices under monopolistic competition. Building on Dierickx et al. (1988), the next two consider VAT changes in a market with heterogeneous firms where the downstream and upstream sectors in turn produce under Cournot competition. In the three cases, we find that the effect of competition intensity on pass-through depends on whether producers have increasing or decreasing marginal costs. In the intuitive case of increasing marginal costs, pass-through increases with competition because greater competition prevents producers from realizing and passing on savings from scaling down in response to a tax hike.

The fourth model generalizes the ‘quality ladder’ model in Khandelwal (2010) to allow for substitution or complementarity effects between consumer valuation of affordability and quality. We find that variation in pass-through depends on price-quality complementarity. For products with longer ‘quality ladders’ where differences in quality are the starkest, we show that pass-through is larger when there is a high enough degree of price-quality complementarity. In this case, consumers faced with higher prices from higher taxes ask for objects of greater quality, resulting in even higher prices. With less complementarity, consumers prefer lower quality and a lower price increase.

We then investigate empirically the relationships between market characteristics and pass-through using a panel of 14 Eurozone countries between 1999 and 2013. We follow closely the methodology developed in BDKW to systematically quantify the effects of VAT reforms in Europe over time, at the product and country levels. We enrich their specification by interacting VAT reforms with measures of competition and scope for quality. We also examine the role of different varieties of VAT reform (e.g. reforms announced well in advance vs. surprise reforms, or tax hikes vs. tax cuts) in explaining some of the pass-through heterogeneity.

Firstly, we find that changes in regulation in supplier markets play a substantial role, with a one standard deviation rise in the competition-friendliness of regulation (roughly equal to the difference between Austria and relatively uncompetitive Italy in 2013) increasing pass-through by up to 55%. We benchmark this effect against other supply-side characteristics, and find that it is more significant and more important. These results are also significant in a historical context. Liberalizing reforms over the last thirty years have substantially increased the competition-friendliness of regulation in European product markets, so our findings imply that VAT cuts today will be passed on to consumers substantially more than in the past.

Secondly, we investigate the role of product differentiability, and find that the greater the scope for quality differentiation the larger is pass-through. Our empirical results are consistent with our theoretical framework and suggest the existence of complementarity between preferences for quality and price.

Many recent VAT policy changes have been announced significantly before they come into effect.6 This constitutes a form of ‘fiscal forward guidance’ (see e.g. Fujiwara and Waki, 2019), which could have real effects (Leeper et al., 2013; Mertens and Ravn, 2010, 2011, 2012). Even outside times of crisis, fiscal policy uncertainty is large, so it is important to understand the effects of advance communication by policymakers.7 We therefore match data on VAT changes to the Tax Policy Reform Database (Amaglobeli et al., 2018) to create the first cross-sector database of VAT reforms including announcement dates, and use it to provide the first systematic assessment of announcement effects across many product categories.8 We find little overall support for ‘anticipation’ or ‘total effect’ hypotheses.

Together our results imply that market structure should be an important consideration when reforming VAT. For a government seeking to mobilize revenue through raising VAT (e.g. Saudi Arabia in May 2020), a greater share of the burden of higher taxes will fall on consumers relative to firms for products with higher upstream competition or for products characterized by a wider quality range. For a government using a VAT cut to stimulate consumption (e.g. Germany in June 2020), or to support firm profits, the effects are opposite. Firms will retain more of the VAT cut in higher markups, and consumers will experience smaller price reductions, the less competitive the upstream sector or the narrower the range of product quality.

Similarly, our results can inform policymakers of the likely effects of VAT reforms targeted at specific sectors. For instance, pharmaceuticals draw heavily on highly regulated industries, so pass-through is lower. Thus, a decision to waive VAT on coronavirus testing kits (as in the EU in April 2020) is likely to provide relatively greater support to the supply side through higher markups, than to the demand side through lower prices.

The rest of this paper proceeds as follows. Section II outlines the theoretical motivation, then Section III describes the data and outlines the empirical strategy. Section IV presents the results, and Section V addresses their robustness. Section VI concludes.

II. Theoretical Motivation

We examine the role of market structure and consumer preferences in determining pass-through by considering four specific cases, building on earlier work by Delipalla and Keen (1992) and Weyl and Fabinger (2013). Consider a good i, with consumer price pi and producer price p˜i subject to ad valorem tax-exclusive rates τi (the VAT), meaning that pi=p˜i(1+τi). As is standard, we define the degree of pass-through to the consumer as the proportionate response of the consumer price to an increase in the tax factor:

γi=lnpiln(1+τi)

We investigate the factors determining γi in the following four scenarios. Full descriptions of the models and derivations of the theoretical results are in the Appendix.

A. Imperfect Competition in the Downstream Sector

We consider the effects of greater competition on VAT pass-through by examining the impact of having more producers under two settings with imperfect competition. In the first setting, we consider equally productive firms competing in prices under monopolistic competition with horizontal differentiation. In the second setting, we follow Dierickx et al. (1988) and consider heterogenous firms competing on quantities. In both cases, we assume that each producer n=1..N, maximizes profits by choosing either the price pn or the produced quantity qn while facing a cost function

Cn=a+cn+b2qn2witha>0;cn>0;

and where b > 0 corresponds to the intuitive case of increasing marginal costs. In the first setting with equally productive firms we have that cn is identical for all n. Producers take the demand function as given and we assume demand to be iso-elastic.9

Proxying ‘competitiveness’ by the number of firms in the market, we show that the impact of competition on pass-through depends on the cost functions, and specifically on the parameter b. Lower demand resulting from higher taxes induces producers to scale back production. With increasing marginal costs, a reduction in scale implies some savings on production marginal costs which, in turn, allows for lower producer prices. Greater competition dampens this cost adjustment. With few large firms with stretched production capacities, a reduction in scale yields large savings. With many smaller firms competing, savings from scaling down are smaller and producers are less able to lower their prices in compensation for higher VAT. In this case, greater competition implies a greater pass-through. Conversely, in the case of decreasing marginal costs when b is negative enough, greater competition implies a lower pass-through. We investigate in the empirical section whether the impact of competition on pass-through is consistent with increasing or decreasing marginal costs.

B. Imperfect Competition in the Upstream Sector

We now consider the case of two sectors, where the downstream sector operates under perfect competition and requires inputs from upstream producers selling under Cournot competition. For the sake of clarity, we assume that upstream producers only sell to the downstream sector and that they face no taxes. Production in the upstream sector is otherwise the same as described in the previous subsection. Both upstream and downstream producers face iso-elastic demand, and that is the case in the upstream sector because the demand for inputs q1 is derived from the downstream producer cost function CF=pIqI=pId(1ρ)qF1/(1ρ) where 0 < ρ < 1 and d> 0.

We obtain the same result as in the previous section. An increase in VAT lowers demand for the final good, and now also reduces demand for upstream inputs. In the case of increasing marginal costs, a reduction in scale for input producers means lower cost, which are then passed through to input prices. Cheaper input costs allow for lower producer prices in the downstream sector. As in the previous case, greater competition dampens the variation in producer costs in response to VAT rate changes. With more firms competing, production capacities are not overly stretched, implying smaller savings from scaling down, and a lower reduction in producer prices. The results are the same as in the single sector case: pass-through increases (decreases) with competition when marginal costs are increasing (decreasing). We investigate in the empirical section whether the impact of competition in the upstream sectors on pass-through is consistent with increasing or decreasing marginal costs.

C. Product Differentiation

We now consider a single-good market in which there are many varieties indexed by n that differ along a horizontal and a vertical dimension as in Khandelwal (2010). Horizontal differentiation is assumed to randomly appeal more to some consumers than others and to be costless, implying that all varieties are consumed in equilibrium.10 By contrast, vertical differentiation, or equivalently ‘quality’, is costly to produce but is regarded by all consumers as superior: holding prices fixed, all consumers would prefer higher quality objects. Each consumer k knows her valuation of horizontal (ξnk) and vertical (λn) characteristics of every variety and chooses the variety n that gives her the highest indirect utility

Vnk=δn+ξnkwithδn(θλnψpnψ)1/ψ,

where δn represents the mean consumer valuation of variety n. δn increases with quality and decreases with price. The parameter ψ controls the degree of substitutability between price and quality, with higher ψ indicating the two characteristics are more easily substituted – i.e. consumers are happy to sacrifice quality for a lower price – while a lower, possibly negative, ψ indicates greater complementarity.11 Greater values of the parameter θ indicate a longer ‘quality ladder’, as defined in Khandelwal (2010), and imply that firms have incentives to produce higher quality.

Each firm produces a variety n subject to a marginal cost function (w + λn/Z) increasing with quality, with wage w, and decreasing with technology Z, and seeks to maximize profits.

We show in the Appendix that the effect of quality on VAT pass-through depends on ψ, the degree of substitutability-complementarity between consumer valuations of price and quality. In the substitution case when ψ >0 (as in Khandelwal, 2010), for a given increase in consumer price resulting from a tax hike, consumers prefer a mitigation in the price increase at the expense of lower quality. Producers respond accordingly and pass-through is lower. The opposite is true in the complementary case when ψ<0 and -ψ is large enough: consumers prefer to tolerate a larger price increase and to be compensated with relatively higher quality. Those effects are magnified by the scope for quality, or ‘quality ladder’ θ. Therefore, pass-through decreases with the quality ladder in the substitution case, while the opposite is true in the complementarity case. We investigate in the empirical section whether the effect of the scope for quality on pass-through is consistent with price-quality complementarity or substitutability.

D. Early Announcement

Early announcement can, in theory, generate anticipation and smoothing effects, i.e. an early and/or prolonged increase in pass-through. On the supply side, the presence of menu costs or Calvo pricing (Calvo, 1983) encourages firms to smooth the price response to an announced VAT change to save on adjustment costs. As discussed in Buettner and Madzharova (2017), for durables there is an extra effect through the demand channel: consumers aware of a future tax fall will defer consumption, reducing demand before the reform and hence lowering prices. Conversely, for an anticipated tax hike, consumers raise pre-reform demand, thereby contributing to higher prices before the rate increase – as observed before the German VAT increase in January 2007 (Danninger et al., 2008).12 Lastly, in a situation of information overload and rational inattention (Sims, 2003), early announcement may increase the salience of a particular reform to consumers and firms, increasing total pass-through. We investigate these ‘anticipation’ and ‘total effects’ in the empirical section.

III. Data and Empirical Specification

We use data on monthly VAT rates across European countries and consumption categories constructed by BDKW from the European Commission publication VAT Rates Applied in the Member States of the European Union and from additional publications by the International Bureau for Fiscal Documentation. The distribution and characteristics of VAT reforms across countries are summarized in Appendix Tables 1 and 2. All the countries studied are in the Eurozone, reducing distortions due to differing exchange rates or monetary policies.13 Data on monthly prices are from Eurostat’s Harmonized Index of Consumer Prices, categorized according to the ‘Classification of Individual Consumption According to Purpose’ (COICOP). We follow BDKW in limiting our sample to those categories for which prices are sufficiently market-driven -excluding, for example, rental accommodation, electricity and healthcare.

We measure the competition-friendliness of regulation in upstream non-manufacturing industries using the ‘Regimpact’ indicator from the Organization for Economic Co-operation and Development (Conway and Nicoletti, 2006; Égert and Wanner, 2016; Koske et al., 2015). This uses country-specific input-output weights wjk to combine survey-based indicators of the competition-friendliness of regulation in several upstream non-manufacturing industries (REGNMIjt), producing a measure of the degree of regulation affecting final output sectors:1415

Regimpactikt=Σj=1JREGNMIjtwjk

where k denotes the output sectors of interest and j denotes upstream non-manufacturing sectors. The distribution of product market regulation across consumption categories is shown in Figure 5 in the Appendix. The trends in regulation are shown in Figure 6 in the Appendix; in general regulation became much more pro-competitive over the period.

We construct two measures of market competitiveness in the downstream sectors affected by the VAT change, using trade data from UN Comtrade.16 Firstly, we use the sum of imports and exports over total consumption as a measure of openness to trade:

Opennessikt=Importsikt+ExportsiktConsumptionikt

where consumption data are drawn from Eurostat at the 3-digit sector level (rather than the 4-digit level for which VAT rates are available). Secondly, we construct a Herfindahl-Hirschman Index based on import origins to proxy for market concentration:

ImportConcentrationikt=Σc=1Nsickt2

where:

sickt=MicktΣc=1NMickt=ImportsintoifromcTotalimportsintoi

Both of these are imperfect measures of competitiveness, but serve in the absence of relevant firm-level data. Assuming that firms are evenly distributed across producing countries, a high degree of concentration observed among import origins is a necessary consequence of high market concentration among firms, though not sufficient to guarantee it17

We use the scope for product differentiability derived in Khandelwal (2010). The scope for quality, or ‘quality ladder’, is backed out from price and quantity data. High market share conditional on price suggests that a product is high quality and long quality ladders correspond to products with a large dispersion in estimated quality.18 Khandelwal (2010) constructs his product-level measure using trade data on goods, which means ‘quality ladder’ estimates are only available for the subset of good industries and do not vary across countries.19 This prevents us from using the full price and VAT dataset, and some controls, with this measure – so we also perform several robustness checks to verify that our results are not driven by the restrictions related to these data limitations. The distribution of quality scope across consumption categories is shown in Figure 7 in the Appendix.

We standardize all four measures (Regimpact, trade openness, import concentration and quality ladder length) so that their impacts are comparable. The four measures are only weakly correlated, as shown in Table in the Appendix. We also match VAT reforms in the BDKW data to the IMF’s new Tax Policy Reform Database (Amaglobeli et al., 2018), which contains announcement dates. Summary statistics for those VAT changes that we can match to announcement dates are shown in Appendix Table 5. Lastly, we use consumption data from Eurostat to weight observations by their consumption share, and total value added from EU KLEMS in a robustness check. Overall, we use an unbalanced panel of approximately 110k observations spanning January 1998 to December 2013. The variables are summarized in Table 4 in the Appendix.

We build on the specification in BDKW and estimate pass-through from VAT changes to prices by regressing country-product prices on taxes:

Δln(pikt)=β0+Σj=66β1jΔln(1+τikt+j)+Σj=66β2jΔln(1+τikt+j)Xikt+β3Xikt+φit+φkt+φik+εikt

where pikt denotes the price of product k in country i in month t and τikt+j represents the VAT rate in country i for product k in month t. The coefficients of interest β1j capture the average pass-through across products at different horizons j, while β2j measures deviations from the mean pass-through across several covariates. Specifically, the sequences of β1j and β2j capture the magnitude of pass-through adjustments at different times around the reform dates, i.e. at a number of months; before and after the reform date.20 The coefficients φit, φkt and φik are country-time, product-time, and country-product fixed effects, and εikt is the error term.21 Xikt denotes country-product-time covariates of interest, specifically product market regulation, quality range, openness to trade, and import concentration. In our main specification we de-seasonalize and de-trend all price indices, weight observations by their consumption share, and cluster standard errors at the country-product level to account for possible autocorrelation in the error term.

To investigate the effects of early announcement, we run a similar specification with the change in VAT also interacted with a dummy for whether the announcement-to-implementation lag for a particular reform is above or below the median. In this case the interaction of the dummy with the sum of pre-reform coefficients Σj=16β2j tests for an anticipation effect, and the interaction of the dummy with the cumulation of all the β2j terms across the whole window j ∈ {-6, ...,6} tests for a total effect.

IV. Results

This section presents our three main results – on product market regulation, quality scope and early announcement. The subsequent section outlines various robustness checks, while the Appendix includes additional results, for example on the heterogeneity of announcement effects.

A. Product Market Regulation

Table 1 shows results from the main specification in the full dataset; column (1) shows results with no fixed effects, column (2) shows results with individual fixed effects, and column (3) uses interaction fixed effects. The first four estimates correspond to β1 in the main estimating equation above – they estimate the relationship between changes in the VAT rate and changes in prices, i.e. baseline pass-through. ‘Pre-Reform’ refers to the total effect across the six months preceeding the VAT change, and ‘Post-Reform’ refers to that across the six months afterwards; ‘Contemporaneous’ refers to effects in the month of the reform, and ‘Total’ is the sum of effects over the whole window. The remaining estimates correspond to different elements of β2, and in turn reflect the impact of variation in the elements of Xikt – specifically, Opennessikt, ImportConcentrationikt and Regimpactikt – on pass-through.22

Average baseline pass-through of a VAT rise to prices is 31% in column (3).23 As in BDKW’s estimates, this effect is almost entirely driven by the contemporaneous pass-through effect – i.e. by the impact on prices in the month that the reform is introduced. A one standard deviation fall in Regimpact (i.e. a one standard deviation rise in the competition-friendliness of upstream regulation, equivalent to the gap between Italy and relatively competitive Austria in 2013) raises pass-through by a further 18 percentage points, a 56 percent increase in pass-through.

Table 1.

Estimates of Pass-Through Heterogeneity

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption. Regimpact, openness and market concentration are standardized so the coefficients can be interpreted as the impact on pass-through of a one-standard-deviation rise in the regressor. Pre-Reform, Contemporaneous and Post-Reform effects are also estimated for Openness and Concentration, but are not significant so omitted for conciseness.

These effects are more significant and more important than the other supply-side competition measures of openness to trade and import concentration. To the extent that openness and concentration proxy for the competitiveness of the downstream sector, this suggests that the theoretical mechanism outlined in section II.B is stronger than that in II.A. As discussed further in the Appendix, this result aligns with findings elsewhere that upstream reforms affecting inputs can have substantial downstream effects (e.g. Amiti and Konings, 2007; Arnold et al., 2016; Bertrand et al., 2007). A full analysis of the conditions under which such upstream effects can amplify further downstream, rather than decay into insignificance, is beyond the scope of this paper (for details, see e.g. Acemoglu et al., 2012).

Figure 1 plots the cumulated values of the estimated coefficients β1r for the 12 months surrounding a VAT change for the specification with the most complete set of fixed effects. The dashed line shows pass-through over time for a consumption category with exactly average levels of upstream product market regulation, openness to trade, and market concentration. There is little pass-through prior to the change, then most of the total effect comes within the first month of the reform. The black line illustrates the marginal impact of upstream regulation on these dynamics: it plots the marginal impact on pass-through of having upstream regulation that is one standard deviation more competition-friendly than the average. Again, most of the marginal impact occurs in the month of the VAT reform, with some additional impact in the six months after the reform. This is consistent with the purchaser-supplier relationships described in section II adjusting to the change reasonably quickly. The extent to which forewarning of the reform speeds up such processes is examined in section C below.

Figure 1.
Figure 1.

Cumulative Effect of Upstream Regulation on Pass-Through

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph shows cumulative baseline pass-through and the impact upon this of upstream regulation. The black (blue) lines show cumulative pass-through in a country-product pair with regulation that is exactly one standard deviation more (less) competition-friendly.

Reforms over the last thirty years have substantially increased the competition-friendliness of regulation in European product markets (Égert and Wanner, 2016). The overall median value of the Regimpact measure since 1999 is shown in Figure 2, while the trends in each country and consumption category are shown in Figure 6 in the Appendix. A back-of-the-envelope calculation takes the observed changes in the Regimpact index for each country-product category over the observed period and multiplies them by the coefficient on the VAT-PMR interaction term in Table 1. The smoothed distribution of these estimated changes in VAT pass-though is shown in Figure 3. Because regulations were loosened almost everywhere, our results imply that VAT pass-through increased practically everywhere for all products. The median estimated impact of the large increase in the competition-friendliness of regulation since 1999 is an increase in pass-through of approximately 21 percentage points, while the vast majority of the distribution has an increase in pass-through of more than 10 percentage points. This is a direct extrapolation of our results without proper identification, but this illustrates that changes in upstream regulation are likely to have substantially affected the consequences of most VAT reforms in recent history.

Figure 2.
Figure 2.

Median Index of Regulation Over Time

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph shows the trends over time in the median value, across all countries and products, of the ‘wide’ and ‘narrow’ Regimpact indices of product market regulation. A lower value of the index reflects a more competition-friendly regulatory stance in upstream non-manufacturing industries.
Figure 3.
Figure 3.

Distribution of Estimated Impact of Regulation on Pass-Through

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph shows the smoothed distribution across country-product categories of the estimated increase in pass-through resulting from changes in regulation between 1999 and 2013. It applies the main estimate from Table 1 to the observed change in the Regimpact indicator across the period observed, using only those country-product categories with observations spanning at least ten years.

B. Scope for Quality

Table 2 repeats the analysis for those products for which measures of the scope for quality are available.24 Since the `quality ladder’ data only vary across products, not across countries, we cannot include product-time fixed effects as these would remove all variation. We therefore include only country-product, country-time, product and time fixed effects in the ‘Interaction FEs’ quality specification. This slight loosening has little impact on the Regimpact results, which remain consistent across columns, suggesting that the ‘lighter’ specification still provides informative estimates for the effect of quality range.

The results in Table 2 show that a one standard deviation increase in the length of the ‘quality ladder’ of a product can raise pass-through by more than 40 percentage points. This fits the theory in section II in the case that demand for quality is relatively more important to consumers when prices are higher – i.e. in the ‘complementarity’ case. In this scenario, firms opt to pass on more of a VAT rise rather than reduce quality to dampen the impact on prices; the greater the scope for quality differentiation, the stronger this effect, so the higher is pass-through.

Considering Table 1 and Table 2 together, the regulation and quality effects have comparable magnitudes, while the regulation effect is somewhat more robust across different specifications. Figure 9 in the Appendix below shows the dynamics of the quality scope effect. While there is again a significant effect in the month of the reform, the effect also continues to grow over the six months following the reform.

Table 2.

Estimates of Pass-Through Heterogeneity, Including Quality Range

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01 Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption.

C. Early Announcement

To test whether pass-through differs for reforms announced far in advance, we consider the dynamics of those cases where we can match a VAT change to an announcement date in the TPRD. We create a dummy AnnouncedEarly that equals one if the lag between announcement and implementation is greater than the median implementation lag of 32 days. Interacting this with the pass-through term in the baseline dynamic regression finds that there is no significant anticipation effect in the six months before the reform, as shown in Table 3, though there may be some small cumulative effect over the whole one year window. The full dynamics are illustrated in Figure 4.

To check whether this null result is driven by the above/below median specification, in Table 8 in the Appendix we also present results using a continuous implementation lag variable. We find that an additional month of implementation lag is weakly associated with up to 6% additional total pass-through, but there is again no significant anticipation effect. Anticipation effects through the demand channel may be particularly strong for durables, as noted in section II.D, since they offer greater opportunity to expedite or defer consumption in response to future price changes. We therefore also split the results between durables and non-durables, shown in Table in the Appendix, and again find no evidence for anticipation effects.25 Lastly, a positive result may be obscured by variation in market competitiveness, which we know plays a role as discussed above. Therefore in Table in the Appendix we also include regulation, quality, openness and concentration in the specification, but again find no evidence for announcement effects.

Overall, while there is some weak evidence that reforms announced earlier tend to have slightly larger pass-through, there is no strong support for either the ‘anticipation’ or ‘total effect’ hypotheses.26 We consider this null result a useful and constructive contribution to the literature. To the best of our knowledge this is the first study to systematically match broad country-product VAT reform data to announcement dates – and thus we are able to examine a potential ‘missing variable’ in important works such as BDKW and Benzarti et al. (2017). Our null result reinforces the findings of these papers, by suggesting that, in aggregate, announcement effects are unlikely to be playing a substantial confounding role.

Table 3.

Impact of Early Announcement on Pass-Through

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01 Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption.
Figure 4.
Figure 4.

Marginal Effect on Pass-Through of Early Announcement

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph shows the cumulative marginal impact on baseline pass-through of having an implementation lag (announcement date minus implementation date) above the median.

V. Robustness Checks

To reduce the influence of regulatory outliers, Table in the Appendix replaces Regimpact with RegimpactHML, which takes value 1 if the observation is in the top quartile of the Regimpact distribution, value -1 if in the bottom quartile, and zero otherwise. Results remain similar, with a strong negative relationship between RegimpactHML and pass-through.

Secondly, we check whether pass-through heterogeneity depends on the direction of the VAT change, following recent work on asymmetric pass-through (e.g. Benzarti et al., 2017; Carbonnier, 2007; Politi and Mattos, 2011). Pass-through heterogeneity for increases and decreases are estimated by β2j(inc)andβ2j(dec) in:

Δln(pikt)=β0+Σd{exp,rec}Σj=66β1j(d)Δln(1+τikt+j(d))+Σd{exp,rec}Σj=66β2f(d)Δln(1+τikt+j(d))Xikt+β3Xikt+φit+φkt+φik+εikt

Results comparing pass-through across products impacted differently by regulation are shown in Error! Reference source not found. in the Appendix. The previous literature has found evidence for greater price rigidity with respect to decreases than increases; however, like BDKW, we find little evidence of this in our data – the final column of Table 12. shows few significant differences between the coefficients on increases and decreases. As discussed in BDKW, the mostly insignificant differences are likely due to substantial heterogeneity across product categories in our dataset, without direct association with the reform type (a VAT hike or cut).

Table in the Appendix repeats this exercise for those observations with quality data. In this case, greater pass-through for products with a longer `quality ladder’ as estimated in section IV.B appears to be essentially driven by reforms with VAT increases. According to our theoretical framework, the result would suggest that producers respond to a VAT hike by increasing quality, while they choose to leave quality unchanged in the case of VAT cuts.

Thirdly, we use a similar method to investigate whether pass-through varies with the business cycle. We use recession indicators from the OECD (Federal Reserve Bank of St. Louis, 2020; OECD, 2020), constructed by using statistical methods to identify turning points in the time series of industrial output and GDP. We run:

Δln(pikt)=β0+Σd{exp,rec}Σj=66β1j(d)Δln(1+τikt+j(d))+Σd{exp,rec}Σj=66β2f(d)Δln(1+τikt+j(d))Xikt+β3Xikt+φit+φkt+φik+εikt

where β1j(rec)andβ1j(exp) reflect baseline pass-through in recessionary and expansionary periods respectively, and β2j(rec)andβ2j(exp) reflect heterogeneity likewise. The results are shown in Table 14 in the Appendix. We find some evidence that pass-through effects are stronger in expansions, possibly because prices are more flexible when inflation is higher, but ultimately cannot reject equality of pass-through coefficients across expansionary/contractionary periods.

Lastly, in additional specifications (available on request) we allow for differential effects of regulation and quality across types of VAT change – specifically standard rate changes, reduced rate changes and reclassifications, as discussed in detail in BDKW. However, with current data we cannot make clear inferences about the triple interaction between reform, regulation/quality and reform-type, as our results may simply be driven by the composition of reforms in our dataset For instance, the majority of reforms in our dataset are standard rate changes, affecting relative standard errors in estimates across the varieties. The average size and spread of the reforms also vary substantially across type, as shown in Table 2, which could affect the estimated coefficients if the relationship between reform size and pass-through is non-linear. We therefore focus on the pooled effects, but also note that Figure 2 of BDKW shows similar effects across reform types – particularly once the reform is introduced, i.e. in the period for which we find regulation and quality to be important.

VI. Conclusion

This paper investigates the role of market structure and timing in pass-through heterogeneity. We extend existing theory by modelling three different settings in which market competitiveness can influence pass-through. We test these relationships empirically using a consumption panel across 14 Eurozone countries, and find that upstream product market regulation and quality have a substantial impact – both in absolute terms and relative to other market characteristics. Our results indicate that pass-through to consumer prices is greater the more competitive the upstream sector or the wider the quality range of the taxed product.

Extending such analysis beyond pricing behavior – e.g. to direct observation of firm markups and marginal costs – is likely to be a fruitful area for future research. We model imperfect competition in upstream and downstream sectors independently and in a partial equilibrium framework, so future work could also extend the theory to a GE setting – allowing for broader linkages between sectors.

Our results suggest that the substantial loosening in regulations to encourage greater competition in upstream sectors at the beginning of the century may have put downward pressure on pass-through in recent history. We also provide the first systematic evidence on ‘fiscal forward guidance’ with respect to VAT reforms, finding that early announcement is unlikely to have large anticipation or total effects.

Together our results are relevant for governments considering VAT reforms with the view of stimulating demand, supporting supply or protecting either side of the market. Because pass-through affects whether supply or demand is more affected by a VAT reform (Weyl and Fabinger 2013), policy-makers should factor in market characteristics. A greater or smaller VAT rate change may be needed to achieve a certain price variation objective depending on market characteristics. In the cases where pass-through is such that producer or consumer prices are unresponsive to VAT change, policy makers willing to achieve some targeted support could look for more cost-effective instruments than VAT changes.

VII. References

  • Acemoglu, D., V.M. Carvalho, A. Ozdaglar, A. Tahbaz-Salehi, 2012. “The Network Origins of Aggregate Fluctuations,” Econometrica 80, 19772016. https://0-doi-org.library.svsu.edu/10.3982/ECTA9623.

    • Search Google Scholar
    • Export Citation
  • Amable, B., L. Demmou, I. Lezdema, 2007. “Competition, Innovation and Distance to Frontier (No. 0706),” CEPREMAP Working Papers (Docweb). CEPREMAP, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Amaglobeli, D., V. Crispolti, E. Dabla-Norris, P. Karnane, F. Misch, 2018. “Tax Policy Measures in Advanced and Emerging Economies: A Novel Database,” IMF Working Paper 18/1, International Monetary Fund, Washington, D.C.

    • Search Google Scholar
    • Export Citation
  • Amiti, M., J. Konings, 2007. “Trade Liberalization, Intermediate Inputs, and Productivity: Evidence from Indonesia,” American Economic Review 97, 16111638. https://0-doi-org.library.svsu.edu/10.1257/aer.97.5.1611.

    • Search Google Scholar
    • Export Citation
  • Andrade, P., M. Carré, A. Bénassy-Quéré, 2015. “Competition and pass-through on international markets: Firm-level evidence from VAT shocksCEPII, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Arnold, J., G. Nicoletti, S. Scarpetta, 2008. “Regulation, Allocative Efficiency and Productivity in OECD Countries: Industry and Firm-Level Evidence,” OECD Econ. Dep. Working Paper, OECD, Paris, France. https://0-doi-org.library.svsu.edu/10.1787/241447806226.

    • Search Google Scholar
    • Export Citation
  • Arnold, J. M., B. Javorcik, M. Lipscomb, A. Mattoo, 2016. “Services reform and manufacturing performance: Evidence from India,” Economic Journal, 126, 139.

    • Search Google Scholar
    • Export Citation
  • Baker, S. R., N. Bloom, S.J. Davis, 2016. “Measuring Economic Policy Uncertainty,” Quarterly Journal of Economics. 131, 15931636. https://0-doi-org.library.svsu.edu/10.1093/qje/qjw024.

    • Search Google Scholar
    • Export Citation
  • Benedek, D., R. de Mooij, M. Keen, P. Wingender, 2015. “Estimating VAT Pass Through,” CESifo Working Papers.

  • Benzarti, Y., D. Carloni, 2017. “Who Really Benefits from Consumption Tax Cuts? Evidence from a Large VAT Reform in France,” Working Paper No. 23848, National Bureau of Economic Research, Cambridge, MA. https://0-doi-org.library.svsu.edu/10.3386/w23848.

    • Search Google Scholar
    • Export Citation
  • Benzarti, Y., D. Carloni, J. Harju, T. Kosonen, 2017. “What Goes Up May Not Come Down: Asymmetric Incidence of Value-Added Taxes,” Working Paper No. 23849, National Bureau of Economic Research, Cambridge, MA. https://0-doi-org.library.svsu.edu/10.3386/w23849.

    • Search Google Scholar
    • Export Citation
  • Bertrand, M., A. Schoar, D. Thesmar, 2007. “Banking Deregulation and Industry Structure: Evidence from the French Banking Reforms of 1985,” J. Finance 62, 597628. https://0-doi-org.library.svsu.edu/10.1111/j.1540-6261.2007.01218.x.

    • Search Google Scholar
    • Export Citation
  • Bi, H., E. Leeper, C. Leith, 2013. “Uncertain Fiscal Consolidations,” Economic Journal F31F63.

  • Bourlès, R., G. Cette, J. Lopez, J. Mairesse, G. Nicoletti, 2013. “Do Product Market Regulations in Upstream Sectors Curb Productivity Growth? Panel Data Evidence For OECD Countries,” Review of Economics and Statistics 95, 17501768.

    • Search Google Scholar
    • Export Citation
  • Braila, C., G. Rayp, S. Sanyal, 2010. “Competition and regulation, Belgium, 1997 to 2004,” Working Papers 1003, Federal Planning Bureau, Belgium.

    • Search Google Scholar
    • Export Citation
  • Buettner, T., B. Madzharova, 2017. “The Effects of Pre-Announced Consumption Tax Reforms on the Sales and Prices of Consumer Durables,” SSRN Electron. J. https://0-doi-org.library.svsu.edu/10.2139/ssrn.3016386

    • Search Google Scholar
    • Export Citation
  • Calvo, G. A., 1983. “Staggered prices in a utility-maximizing framework,” J. Monet. Econ. 12, 383398.

  • Carbonnier, C., 2007. “Who pays sales taxes? Evidence from French VAT reforms, 1987–1999,” J. Public Econ. 91, 12191229.

  • Cette, G., J. Lopez, J. Mairesse, 2014. “Product and Labor Market Regulations, Production Prices, Wages and Productivity,” Working Paper No. 20563, National Bureau of Economic Research, Cambridge, MA, https://0-doi-org.library.svsu.edu/10.3386/w20563.

    • Search Google Scholar
    • Export Citation
  • Cette, G., J. Lopez, J. Mairesse, 2013. “Upstream product market regulations, ICT, R&D and productivity,” Working Paper, Banque de France, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Conway, P., G. Nicoletti, 2006. “Product Market Regulation in the Non-Manufacturing Sectors of OECD Countries: Measurement and Highlights,” https://0-doi-org.library.svsu.edu/10.1787/362886816127.

    • Search Google Scholar
    • Export Citation
  • Copenhagen Economics, 2013. “Regulation and productivity in the private service sectors,” Background report prepared for the Danish Productivity Commission.

    • Search Google Scholar
    • Export Citation
  • Crossley, T., H. Low, C. Sleeman, 2014. “Using a temporary indirect tax cut as a fiscal stimulus: evidence from the UK,” IFS Working Papers 14/16, Institute for Fiscal Studies, London, U.K.

    • Search Google Scholar
    • Export Citation
  • Danninger, S., A. Carare, 2008. “Inflation Smoothing and the Modest Effect of VAT in Germany,” IMF Working Paper 08/1, International Monetary Fund, Washington, D.C.

    • Search Google Scholar
    • Export Citation
  • De Loecker, J., P.K. Goldberg, A.K. Khandelwal, N. Pavcnik, 2016. “Prices, Markups, and Trade Reform,” Econometrica 84, 445510. https://0-doi-org.library.svsu.edu/10.3982/ECTA11042.

    • Search Google Scholar
    • Export Citation
  • Delipalla, S., M. Keen, 1992. “The comparison between ad valorem and specific taxation under imperfect competition,” J. Public Econ. 49, 351367.

    • Search Google Scholar
    • Export Citation
  • Di Ubaldo, M., I. Siedschlag, 2018. “Determinants of firms’ inputs sourcing choices: The role of institutional and regulatory factors,” ESRI Working Paper No. 599, Economic and Social Research Institute, Dublin, Ireland.

    • Search Google Scholar
    • Export Citation
  • Dierickx, I., C. Matutes, D. Neven, 1988. “Indirect taxation and Cournot equilibrium,” International Journal of Industrial Organization, Volume 6, Issue 3, Pages 385399. https://0-doi-org.library.svsu.edu/10.1016/S0167-7187(88)80019-5.

    • Search Google Scholar
    • Export Citation
  • Égert, B., I. Wanner, 2016. “Regulations in services sectors and their impact on downstream industries,” https://0-doi-org.library.svsu.edu/10.1787/5jlwz7kz39q8-en.

    • Search Google Scholar
    • Export Citation
  • European Commission, 2007. Quarterly report on the Euro area (No. 6 (4)).

  • Fujiwara, I., Y. Waki, 2019. “Fiscal forward guidance: A case for selective transparency,” J. Monet. Econ. https://0-doi-org.library.svsu.edu/10.1016/j.jmoneco.2019.10.007.

    • Search Google Scholar
    • Export Citation
  • Gaarder, I., 2018. “Incidence and Distributional Effects of Value Added Taxes,” Econ. J. 0. https://0-doi-org.library.svsu.edu/10.1111/ecoj.12576.

    • Search Google Scholar
    • Export Citation
  • Gaulier, G., S. Zignago, 2010. “BACI: International Trade Database at the Product-Level. The 1994–2007 Version (No. 2010–23),” Working Papers, CEPII, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Goldberg, P. K., A.K. Khandelwal, N. Pavcnik, P. Topalova, 2010. “Imported Intermediate Inputs and Domestic Product Growth: Evidence from India,” Q. J. Econ. 125, 17271767. https://0-doi-org.library.svsu.edu/10.1162/qjec.2010.125.4.1727.

    • Search Google Scholar
    • Export Citation
  • Havik, K., K.M. Morrow, W. Röger, A. Turrini, 2008. “The EU-US total factor productivity gap: An industry perspective (No. 339),” European Economy – Economic Papers 2008 – 2015. Directorate General Economic and Financial Affairs (DG ECFIN), European Commission.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund (Ed.), 2015. “Uneven growth: short- and long-term factors,” World Economic Outlook, International Monetary Fund, Washington, D.C.

    • Search Google Scholar
    • Export Citation
  • Khandelwal, A., 2010. “The Long and Short (of) Quality Ladders,” Rev. Econ. Stud. 77, 14501476.

  • Koske, I., I. Wanner, R. Bitetti, O. Barbiero, 2015. “The 2013 update of the OECD’s database on product market regulation,” https://0-doi-org.library.svsu.edu/10.1787/5js3f5d3n2vl-en.

    • Search Google Scholar
    • Export Citation
  • Leeper, E. M., T.B. Walker, S.-C.S. Yang, 2013. “Fiscal foresight and information flows, Econometrica 81, 11151145.

  • Mariscal, R., A. Werner, 2018. “The Price and Welfare Effects of The Value-Added Tax: Evidence from Mexico,” IMF Staff Paper 18, International Monetary Fund, Washington, D.C.

    • Search Google Scholar
    • Export Citation
  • Mertens, K., M. Ravn, 2010. “Measuring the Impact of Fiscal Policy in the Face of Anticipation: A Structural VAR Approach,” Econ. J. 120, 393413.

    • Search Google Scholar
    • Export Citation
  • Mertens, K., M.O. Ravn, 2012. “Empirical Evidence on the Aggregate Effects of Anticipated and Unanticipated US Tax Policy Shocks,” Am. Econ. J. Econ. Policy 4, 145181. https://0-doi-org.library.svsu.edu/10.1257/pol.4.2.145.

    • Search Google Scholar
    • Export Citation
  • Mertens, K., M.O. Ravn, 2011. “Understanding the Aggregate Effects of Anticipated and Unanticipated Tax Policy Shocks,” Rev. Econ. Dyn. 14, 2754.

    • Search Google Scholar
    • Export Citation
  • OECD, 2018. “Consumption Tax Trends 2018: VAT/GST and Excise Rates, Trends and Policy Issues, Consumption Tax Trends,” OECD Publishing, Paris, France.

    • Search Google Scholar
    • Export Citation
  • Pallan, H., 2019. “Do Investors Care About Consumption Taxes? Evidence from Advanced and Emerging Markets,” Graduate Institute of International and Development Studies, Geneva, Switzerland.

    • Search Google Scholar
    • Export Citation
  • Politi, R. B., E. Mattos, 2011. “Ad-valorem tax incidence and after-tax price adjustments: evidence from Brazilian basic basket food,” Can. J. Econ. Can. Déconomique 44, 14381470. https://0-doi-org.library.svsu.edu/10.1111/j.1540-5982.2011.01680.x.

    • Search Google Scholar
    • Export Citation
  • Ramey, V. A., 2011. “Identifying Government Spending Shocks: It’s all in the Timing,” Q. J. Econ. 126, 150.

  • Sims, C. A., 2003. “Implications of rational inattention,” J. Monet. Econ. 50, 665690. https://0-doi-org.library.svsu.edu/10.1016/S0304-3932(03)00029-1.

    • Search Google Scholar
    • Export Citation
  • Topalova, P., A. Khandelwal, 2010. “Trade Liberalization and Firm Productivity: The Case of India,” Rev. Econ. Stat. 93, 9951009. https://0-doi-org.library.svsu.edu/10.1162/REST_a_00095.

    • Search Google Scholar
    • Export Citation
  • Weyl, E. G., M. Fabinger, 2013. “Pass-Through as an Economic Tool: Principles of Incidence under Imperfect Competition,” J. Polit. Econ. 121, 528583. https://0-doi-org.library.svsu.edu/10.1086/670401.

    • Search Google Scholar
    • Export Citation
  • Yahmed, S. B., S. Dougherty, 2012. “Import Competition, Domestic Regulation and Firm-Level Productivity Growth in the OECD (No. 980),” OECD Economics Department Working Papers. OECD Publishing, Paris, France.

    • Search Google Scholar
    • Export Citation

VIII. Appendix: Literature Review and Additional Findings

A. Related Literature

A substantial literature exists estimating the effects of specific tax changes. Carbonnier (2007) considers the impact of decreasing VAT on cars and housing repairs in France; Benzarti and Carloni (2017) consider a VAT cut for French restaurants, Mariscal and Werner (2018) consider the impact of differences in VAT for Mexican border cities, and Gaarder (2018) considers a cut in the VAT on food in Norway. A few studies consider effects across multiple countries: while Benzarti et al. (2017) focus on changes in the VAT on hairdressing in Finland, they also consider all VAT changes across EU member states, and Andrade et al. (2015) consider the impact on French export prices of VAT changes in several destination markets. This paper builds primarily on the work of Benedek, De Mooij, Keen and Wingender (2015), who constructed the core dataset of European VAT rates used in this paper. Like BDKW, in estimating VAT pass-through across a broad range of countries and consumption categories we aim to provide more general results than can be reached in studies of a small number of countries, sectors or reforms. We also use the same identification strategy and a product-country panel which, by comparing products across countries and countries across products, provides better controls than product-specific studies or economy-wide cross-country studies.

To measure the impact of upstream regulation on a sector, we use the ‘Regimpact’ indicator developed by the OECD (Conway and Nicoletti, 2006; Égert and Wanner, 2016; Koske et al., 2015). This has been widely used to study the impacts of regulation on productivity (Amable et al., 2007; Arnold et al., 2008; Bourlès et al., 2013; Cette et al., 2014, 2013; Copenhagen Economics, 2013; European Commission, 2007; Havik et al., 2008; International Monetary Fund, 2015; Yahmed and Dougherty, 2012), and to a lesser extent to study the impacts on competitiveness (Braila et al., 2010) and firms’ input sourcing decisions (Di Ubaldo and Siedschlag, 2018). These studies generally find a positive effect of deregulation on productivity, competitiveness, and the propensity of firms to purchase inputs rather than source them intra-firm through FDI. To the best of our knowledge the ‘Regimpact’ indicator has not previously been used to investigate VAT pass-through.

Other studies of upstream service sector reform have found substantial downstream effects on firms. Arnold et al. (2016) construct a measure of services liberalization in India, and find a strong positive effect on the productivity of manufacturing firms intensive in the liberalizing services. Bertrand et al. (2007) find similar effects on French manufacturing firms of banking deregulation in the 1980s. Our finding that features of the upstream market have substantial downstream effects also parallels an established result from the trade literature that input tariffs can have major effects in output markets (e.g. Amiti and Konings, 2007; De Loecker et al., 2016; Goldberg et al., 2010; Topalova and Khandelwal, 2010).

Several previous studies consider the impact of anticipated fiscal shocks on aggregate economic variables, both in theory and empirically (Bi et al., 2013; Fujiwara and Waki, 2019; Mertens and Ravn, 2012, 2011; Ramey, 2011). To the best of our knowledge this is the first study to consider product-level announcement effects across many different VAT reforms. In using the Tax Policy Reform Database (Amaglobeli et al., 2018) to identify announcement effects of VAT reforms on consumer prices, our paper also parallels the work of Pallan (2019), who considers the impact on stock prices.

B. Descriptive Statistics

Table 1.

Summary of VAT Reforms by Country

article image
Table 2.

Summary of Observed VAT Rates and Prices

article image
Table 3.

Pairwise Correlation Between Competitiveness Variables

article image
* shows significance at the 0.05 level
Table 4.

Summary Statistics for Main Variables

article image
* indicates standardized variables
Table 5.

VAT Changes for Which Announcement Dates are Observed

article image
Table 6.

Correlation Among Other Market Structure Variables

article image
* shows significance at the 0.05 level
Figure 5.
Figure 5.

Distribution of Regulation Across Consumption Categories

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: These plots summarize the distribution of the Regimpact measure across consumption categories. A lower value of the indicator reflects a more competition-friendly regulatory stance among input industries. Each box depicts the 25th, 50th and 75th percentiles, with extending lines to the minimum and maximum values, excluding outliers (defined as 1.5IQR below/above the lower/upper quartile).
Figure 6.
Figure 6.

Upstream Regulation Over Time, Country and Consumption Category Panel A. Regimpact by Country Over Time – 25th, 50th and 75th percentiles

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Figure 7.
Figure 7.

Distribution of Quality Scope Across Consumption Categories

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph depicts the estimated quality range across different consumption categories. A higher value of the indicator reflects a longer average ‘quality ladder’ (Khandelwal, 2010).
Figure 8.
Figure 8.

Upstream Industries Included in Regimpact Indicator, and Categories Upon Which They are Scored

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Source: Égert and Wanner, 2016, p. 7

C. Additional Figures and Results

Decomposing the PMR Effect by Upstream Sector

We also use an alternative measure of upstream regulation to decompose the impact across upstream sectors. Table 7 below repeats the analysis using the ‘narrow’ Regimpact indicator, which includes only the first three of the eight industries considered in the ‘wide’ Regimpact indicator (Energy, Transport, Communications, Retail, Accounting, Legal, Engineering, Architecture). The size and significance of the impact of PMR dramatically decreases, suggesting that the impact is driven by the omitted industries, i.e. Retail, Accounting, Legal, Engineering, and Architecture.

To further narrow down which sectoral regulations matter most, we then compare the role of the regulated sectors in supply chains. Compared to the output of professional services, the output of retail is less often used as an input to other sectors. Figure 10 in the Appendix shows the share of intermediate demand in gross output of non-manufacturing sectors across countries. Around 80% of the output of ‘Business Activities’ (comprising Accounting, Legal, Engineering and Architecture) is used as an intermediate input to other sectors of the economy, compared to just 40% in Retail. This is reflected in a substantially higher average weight being attached to the former in the construction of the ‘wide’ Regimpact indicator (Égert and Wanner, 2016), leading us to conclude that the impact of product market regulation on pass-through heterogeneity is most likely to be driven by the impact of regulation in professional services.

Heterogeneity in the Effects of Early Announcement

The top-left panel of Figure 11 below shows contemporaneous pass-through by implementation lag for all country-product pairs for which we have data on announcement dates. The specific cases highlighted illustrate the heterogeneity: the bottom-left panel shows a relatively large possible announcement effect for a rise in VAT on package holidays in Luxembourg, while the bottom-right panel shows no announcement effect for a rise in VAT on restaurants and cafés in Portugal. Future research to gather more complete data on announcement dates will allow systematic evaluation of the factors determining whether advance announcement impacts pass-through.

Figure 9.
Figure 9.

Cumulative Effect of Quality Scope on Pass-Through

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Notes: This graph shows cumulative baseline pass-through and the impact upon this of quality scope. The blue (black) lines show cumulative pass-through in a country-product pair with a quality ladder that is exactly one standard deviation longer (shorter) than the mean.
Table 7.

Estimates Using ECTR Only

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption. Regimpact, openness and market concentration are standardized so the coefficients can be interpreted as the impact on pass-through of a one-standard-deviation rise in the regressor. Pre-Reform, Contemporaneous and Post-Reform effects are also estimated for Openness and Concentration, but are not significant so omitted for conciseness.
Figure 10.
Figure 10.

Share of Intermediate Demand in Gross Output of Non-Manufacturing Sectors

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Source: Égert and Wanner, 2016, p. 9Notes: These graphs show the share of intermediate demand in gross output of non-manufacturing sectors across countries in the mid-2000s. The ‘wide’ Regimpact measure includes the first five sectors, while the ‘narrow’ measure includes only ‘Electricity, gas and water supply’, ‘Transport and storage’, and ‘Post and telecommunications’.
Figure 11.
Figure 11.

Heterogeneity in Announcement Effects

Citation: IMF Working Papers 2021, 061; 10.5089/9781513571546.001.A001

Table 8.

Impact of Early Announcement on Pass-Through, for Continuous Implementation Lag

article image
p-values are shown below coefficients. * p < 0.10, ** p < 0.05, *** p < 0.01. ‘X_ikt’ refers to the inclusion of Regimpact, trade openness and import concentration. Specifications (4) and (5) also control for value added, consumption and whether the reform was part of a package. ‘Implementation Lag’ is measured in months, so e.g. a coefficient of 0.01 implies that announcing a VAT reform one additional month in advance is associated with a 1% increase in pass-through.
Table 9.

Impact of Early Announcement on Pass-Through, for Continuous Implementation Lag, by Durability

article image
p-values are shown below coefficients. * p < 0.10, ** p < 0.05, *** p < 0.01. The ‘Individual FEs’ and ‘Interaction FEs’ specifications correspond to models (2) and (3) in Table 8, but with coefficients estimated independently for Non-Durables and Durables. ‘Implementation Lag’ is measured in months, so a coefficient of 0.01, for example, implies that announcing a VAT reform one additional month in advance is associated with a 1% increase in pass-through.
Table 10.

Regulation, Quality and Announcement Effects

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01

Table above considers the interaction between early announcement and openness, concentration, regulation and quality. Once again there is no evidence of an anticipation or total effect. The impact of regulation is driven by reforms which were announced fewer than 32 days in advance, but this is likely driven by the composition of that group – it contains a substantially higher share of changes to the reduced rate, which have the strongest effects as discussed above. The quality range effect is similar across implementation lag groups.

Additional tables (available on request) repeat the main specifications using country-level clustering and product-level clustering in turn. Results are similar with product-level clustering, while with country-level clustering the contemporaneous effect of Regimpact remains significant while the total effect is marginally insignificant.

D. Additional Tables for Robustness Checks

Table 11.

Estimates Using Discrete PMR Variable

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption. RegimpactHML is a discrete variable taking value 1 if the observation is in the top quartile of the Regimpact distribution, value -1 if in the bottom quartile, and zero otherwise. Openness and market concentration are standardized so the coefficients can be interpreted as the impact on pass-through of a one-standard-deviation rise in the regressor. Pre-Reform, Contemporaneous and Post-Reform effects are also estimated for Openness and Concentration, but are not significant so omitted for conciseness.
Table 12.

Estimates by Direction of VAT Change

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption. Regimpact, openness and market concentration are standardized so the coefficients can be interpreted as the impact on pass-through of a one-standard-deviation rise in the regressor. The final column presents p-values from a Wald test of equality between the Increase and Decrease coefficients.
Table 13.

Estimates by Direction of VAT Change, Including Quality Range

article image
p-values in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01Estimates are the sum of the price elasticity coefficients with respect to tax changes over each period. Prices are de-trended and de-seasonalized, and observations are weighted by their share of national consumption. Regimpact, openness, market concentration and quality range are standardized so the coefficients can be interpreted as the impact on pass-through of a one-standard-deviation rise in the regressor. The final column presents p-values from a Wald test of equality between the Increase and Decrease coefficients.