Abstract
We study the price effects of a temporary VAT reduction in Germany using a web-scraped dataset of daily prices of more than 60000 supermarket products. For causal identification, we compare the development of German prices to those in Austria. We find that the reduction of VAT rates led to a price decrease of 1.3%, implying that 70% of the tax cut were passed on to consumers. Moreover, the pass-through is higher for vertically integrated products (private label) than for independent brands. This is consistent with menu cost theories and theories predicting that price markups act as a buffer for cost shocks.
This is a preview of subscription content, log in via an institution to check access.
Data source: Our World in Data, COVID-19 Dataset, Series: new_cases_smoothed_per_million
Data source: Our World in Data, COVID-19 Dataset, Series: new_deaths_smoothed_per_million
Data source: Google COVID-19 Community Mobility Reports (Color figure online)
Data source: Eurostat
Similar content being viewed by others
Notes
One could object that, even without pass-through, a VAT rate cut may boost economic activity because it allows firms to secure jobs or maintain investment. But a general VAT rate cut helps firms in proportion to their sales. The economic consequences of the COVID-19 pandemic for firms were very heterogeneous. A general VAT cut would channel most of the help to the winners of the crisis, like digital companies or supermarkets, which were exempt from lockdowns. This suggests that, to help firms affected by the crisis, other instruments targeting the most affected sectors and companies would be more effective.
Austria as well as Germany introduced a sector specific VAT cut for hotels and restaurants. In Germany this was on top of the general VAT cut. This will be explained in greater detail in Sect. 2.
Among other things, we investigate whether the pass-through varies across products taxed at the regular VAt rate vs. products taxed at the reduced VAT rate and private label vs. brand products. This cannot be done using CPI data.
Benzarti et al. (2020) show that the pass-through of a VAT rate hike for hairdressing services in Finland was inversely related to hairdressers’ profit margins, suggesting that high cost markups may buffer tax shocks. There is also evidence that the pass-through of excise taxes on alcohol (Hindriks & Serse, 2019) cigarettes (Harding et al., 2012), and fuel (Doyle Jr. & Samphantharak, 2008) is related to the intensity of spatial competition between retailers.
We obtain this number by dividing the revenue in German supermarket retail by private consumption spending. The data are taken from Gesellschaft für Konsumforschung and the German Federal Statistical Office and refer to the year 2018.
Detailed accounts of the stimulus measures are provided by Dorn et al. (2020) for Germany and Baumgartner et al. (2020) for Austria. See also the IMF database on fiscal policy responses to COVID 19: https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19#A.
These programs allow employers to reduce their employees’ working hours without laying them off. Employees in Germany receive a share of the net loss in income incurred of at least 60 and 67%, respectively, in the case of an employee without children and with at least one child. These allowance rates increase over time to 80 and 87%, respectively, after a period of six months. Austria has a very similar program that grants allowances of at least 80% of the previous net salary and up to 90% for smaller salaries under the current legislation until the end of March 2021.
Before the reduction, the VAT rate for restaurants was 19% in Germany and 20% (for food) and 10% (for drinks), respectively, in Austria.
Source: The Nielsen Company.
We started collecting price data from REWE for a different purpose, which is why our sample period begins long before the announcement of the temporary VAT reduction.
For example, in the Billa online shop, ‘tea, coffee, and cocoa’ is a sub-category of ’beverages.’ In the REWE online shop, ‘tea, coffee, and cocoa’ is a main category. The opposite is true for ‘bread and baked goods.’ In other cases, the collection of goods belonging to the same sub-category or sub-subcategory differs between the online shops.
COICOP is published by the UN Department of Economic and Social Affairs and serves as the international reference classification of household expenditure. It is an integral part of the System of National Accounts and used for many statistical purposes such as, for instance, the establishment of weights for the computation of consumer price indexes and the calculation of purchasing power parities.
The fact that the number of REWE products still exceeds the number of Billa products is due to the larger number of brands and product varieties offered by REWE. For example, the REWE online shop features 185 types of flour, 416 types of bread, and 462 types of cereals. At Billa, there are 22 types of flour, 106 types of bread, and 99 types of cereals.
For the USA, Agrawal and Shybalkina (2023) show that this shift to online shopping led to a redistribution of sales tax revenue from urban centers to rural areas.
Unfortunately, data on the online share of grocery spending for Austria are not available for earlier years.
The REWE (Billa) price index represents a weighted average of the price indexes of the single REWE (Billa) products. In both cases we use the CPI weights for Germany.
Austria only reduced the VAT rate for hotels, restaurants, and certain cultural institutions like theaters, museums or natural parks. Germany also provided a targeted VAT cut for food served in restaurants and hotels, where the reduced rather than the standard rate was applied as from July 1, 2020. This was in addition to the general VAT cut which is the focus of our analysis.
We use the COICOP weights used to calculate the Consumer Price Index for Germany to compute weighted averages. The weights can be found here: https://www.destatis.de/DE/Themen/Wirtschaft/Preise/Verbraucherpreisindex/FAQ/anteil-gueter-warenkorb.html.
We compute average weekly price indexes because prices hardly vary within a week.
For example, a price reduction of 1.26% in response to the tax cut for goods subject to the standard VAT rate would imply a pass-through of 50%.
The fixed effects refer to specific products, not product categories or classes. Our baseline specification thus includes roughly 68,000 product-fixed effects.
Clustering standard errors at the product class level would be justified in case similar products are exposed to similar shocks. This assumption is plausible if price developments are mainly driven by production conditions or demand-side factors.
Table 6 of Appendix B shows the estimates in table format.
This result is the opposite of what Benzarti et al. (2020) find. The authors exploit a VAT rate reduction for hairdressing services in Finland that was communicated to be permanent, but eventually repealed five years later. Their findings suggest that the price change in response to the VAT rate hike was two times larger than the price change following the VAT rate cut.
Note that we compute White-robust standard errors instead of clustering standard errors at the 4-digit product class level since the number of clusters would be very small. I.e., the product group ‘food and non-alcoholic beverages’ comprises 22 4-digit product classes, the product group ‘alcoholic beverages and tobacco’ seven 4-digit product classes, and the product group ‘cosmetic and hygiene products’ five 4-digit product classes.
From a theoretical perspective, the implications of market structures for tax incidence are generally ambiguous and often depend on the curvature of demand and cost functions as well as the particular market setting. For instance, more market power in oligopolistic markets may lead to more or less shifting of a consumption tax (Weyl & Fabinger, 2013). In an earlier version of this paper (Fuest et al., 2021), we analyze consumption tax incidence in a love of variety model and show that tax shifting to consumers may be higher or lower in product groups with more varieties, depending on the properties of marginal cost and demand functions.
Based on data from Spanish gas stations, Bajo-Buenestado and Borrella-Mas (2022) find that the pass-through of a tax on prices is significantly higher for vertically integrated gas stations.
There is disagreement in the literature about whether retail stores should be viewed as two-sided markets. In Armstrong (2006)’s opinion, the answer is yes. His opinion is based on the notion that retail stores share features of platforms as they allow an exchange between producers and their customers. Rysman (2009), however, views retail stores as one-sided markets, arguing that “the retailer takes possession of the product and sells the product as it wishes, and the wholesaler has no concern for how many units the retailer is able to sell” (p. 126). In our view, Armstrong’s characterization provides an accurate description of the relationship between independent suppliers featuring their own brands, supermarkets, and customers, whereas Rysman’s characterization best describes the situation in which supermarkets are responsible for the pricing and branding of a product.
A theoretical model which generates this prediction for non-tax cost shocks is developed by Hong and Li (2017).
While REWE and EDEKA are ‘conventional’ supermarket chains, LIDL and ALDI are discounters. Together, these four chains represent about 70% of the supermarket retail market in Germany.
The data are provided by Gesellschaft für Konsumforschung the largest market research institute in Germany, and based on a representative sample of roughly 30,000 German households.
References
Agrawal, D. R., & Shybalkina, I. (2023). Online shopping can redistribute local tax revenue from urban to rural America. Journal of Public Economics, 219, 104818.
Armstrong, M. (2006). Competition in two-sided markets. RAND Journal of Economics, 37, 668–691.
Bajo-Buenestado, R., & Borrella-Mas, M. Á. (2022). The heterogeneous tax pass-through under different vertical relationships. The Economic Journal, 132(645), 1684–1708.
Baumgartner, J., Fink, M., Moreau, C., Rocha-Akis, S., Lappöhn, S., Plank, K., Schnabl, A., & Weyerstrass, K. (2020). Wirkung der wirtschaftspolitischen maßnahmen zur abfederung der covid-19-krise. mikro-und makroökonomische analysen zur konjunkturellen, fiskalischen und verteilungspolitischen wirkung. WIFO Studies.
Benedek, D., De Mooij, R. A., Keen, M., & Wingender, P. (2020). Varieties of VAT pass through. International Tax and Public Finance, 27, 890–930.
Benzarti, Y., Carloni, D., Harju, J., & Kosonen, T. (2020). What goes up may not come down: Asymmetric incidence of value-added taxes. Journal of Political Economy, 128(12).
Blundell, R. (2009). Assessing the temporary VAT cut policy in the UK. Fiscal Studies, 30(1), 31–38.
Carbonnier, C. (2007). Who pays sales taxes? Evidence from French VAT reforms, 1987–1999. Journal of Public Economics, 91(5–6), 1219–1229.
Cole, A. J. (2009). Christmas in August: Prices and quantities during a sales tax holiday. PhD Dissertation, University of Michigan.
Crossley, T. F., Low, H. W., & Sleeman, C. (2014). Using a temporary indirect tax cut as a fiscal stimulus: Evidence from the UK. IFS Working Papers.
Crossley, T. F., Low, H., & Wakefield, M. (2009). The economics of a temporary VAT cut. Fiscal Studies, 30(1), 3–16.
D‘Acunto, F., Hoang, D., & Weber, M. (2018). Unconventional fiscal policy. In AEA Papers and proceedings (Vol. 108, pp. 519–23).
Dorn, F., Fuest, C., & Neumeier, F. (2020). After the great economic collapse: Germany’s stimulus package to recover the economy in times of covid-19. In CESifo Forum (Vol. 21, pp. 38–46). ifo Institute Munich.
Doyle, J. J., Jr., & Samphantharak, K. (2008). \$2.00 Gas! Studying the effects of a gas tax moratorium. Journal of Public Economics, 92(3–4), 869–884.
Feldstein, M. (2002). The role for discretionary fiscal policy in a low interest rate environment. NBER Working Paper 9203.
Fuest, C., Neumeier, F., & Stöhlker, D. (2021). The pass-through of temporary VAT rate cuts: Evidence from German supermarket retail. CESifo Working Paper 9149.
Gopinath, G., & Itskhoki, O. (2010). Frequency of price adjustment and pass-through. Quarterly Journal of Economics, 125, 675–727.
Harding, M., Leibtag, E., & Lovenheim, M. F. (2012). The heterogeneous geographic and socioeconomic incidence of cigarette taxes: Evidence from Nielsen Homescan data. American Economic Journal: Economic Policy, 4(4), 169–198.
Harper, R. K., Hawkins, R. R., Martin, G. S., & Sjolander, R. (2003). Price effects around a sales tax holiday: An exploratory study. Public Budgeting & Finance, 23(4), 108–113.
Hellerstein, R., & Villas-Boas, S. B. (2010). Outsourcing and pass-through. Journal of International Economics, 81, 170–183.
Hindriks, J., & Serse, V. (2019). Heterogeneity in the tax pass-through to spirit retail prices: Evidence from Belgium. Journal of Public Economics, 176, 142–160.
Hindriks, J., & Serse, V. (2020). The incidence of VAT reforms in electricity markets: Evidence from Belgium. CORE Discussion Paper 2020/12.
Hong, G. H., & Li, N. (2017). Market structure and cost pass-through in retail. Review of Economics and Statistics, 99, 151–166.
Kosonen, T. (2015). More and cheaper haircuts after VAT cut? On the efficiency and incidence of service sector consumption taxes. Journal of Public Economics, 131, 87–100.
Levy, D., Bergen, M., Dutta, S., & Venable, R. (1997). The magnitude of menu costs: Direct evidence from large U.S. supermarket chains. Quarterly Journal of Economics, 112, 791–825.
Montag, F., Sagimuldina, A., & Schnitzer, M. (2020). Are temporary value-added tax reductions passed on to consumers? Evidence from Germany’s stimulus. arXiv preprint arXiv:2008.08511.
Rysman, M. (2009). The economics of two-sided markets. Journal of Economic Perspectives, 23, 125–143.
Schmidheiny, K., & Siegloch, S. (2020). On event studies and distributed-lags in two-way fixed effects models: Identification, equivalence, and generalization. ZEW Discussion Paper 20-017.
Weyl, E. G., & Fabinger, M. (2013). Pass-through as an economic tool: Principles of incidence under imperfect competition. Journal of Political Economy, 121(3), 528–583.
Author information
Authors and Affiliations
Contributions
All authors contributed equally to the paper.
Corresponding author
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work benefited from comments by participants at various seminars and conferences including the EconPol Workshop on Public Policy Evaluation, the ZEW Public Finance Conference and the Workshop on ‘Temporary VAT Cuts and other (non) Conventional Fiscal Policies’ at the German Federal Ministry of Finance. Pascal Zamorski and Ludwig Oetker provided excellent research assistance. The usual disclaimer applies.
Appendices
Appendix A: Background information and description of data
The characteristics of supermarket retail are very similar in Germany and Austria. In both countries, the market is highly concentrated. In Germany, the largest retailers are EDEKA (market share in 2019: 24.5%), REWE (market share in 2019: 17.8%), LIDL (market share in 2019: 16.5%), and ALDI (market share in 2019: 11.7%).Footnote 32 They all operate nationwide. In Austria, Billa’s (market share in 2019: 34.1%) biggest competitor is SPAR, with a market share in 2019 of 32.8%. The product assortments of the largest retailers in both countries comprise brands of independent suppliers as well as products marketed under brands owned by the retailers themselves (private labels). REWE’s three major private labels, for example, are Ja! (Yes!), a discount label, REWE Beste Wahl (REWE Best Choice), which is part of the medium price segment, and REWE Bio (REWE Organic), which is REWE’s label for organic products. Billa also has a discount label named Clever, a medium price private label named Billa (same name as the chain), and an organic label named Billa Bio (Billa Organic).
Since our analysis covers only one supermarket chain, it is important to get an idea about its customer base. As a starting point, Fig. 20 of Appendix B shows the geographical distribution of REWE and Billa stores in Germany and Austria, respectively. Stores of both chains can be found across the whole of both countries, although—unsurprisingly—the concentration of stores is higher in the proximity of large cities. Altogether, there are 3693 REWE stores in Germany and 1248 Billa stores in Austria (as of September 2021), which makes 4.4 REWE stores per 100,000 inhabitants in Germany and 14.1 Billa stores per 100,000 inhabitants in Austria.
Reflecting its huge geographical outreach, Fig. 21 shows that the majority of German households regularly buy at REWE. The figure compares the share of regular REWE customers to the share of regular customers of REWE’s biggest competitors in Germany—EDEKA, LIDL, and ALDI. The numbers in Fig. 21 indicate what share of German households has regularly made purchases at the stores of the four supermarket chains in 2020, the year of the temporary VAT cut,Footnote 33 According to these numbers, three quarters of German households regularly bought goods at REWE. Due to that, it seems fair to say that price responses of the temporary VAT rate cut observable at REWE have a noticeable impact on German households.
Table 3 shows the number of products as well as average product prices included in our dataset separately for ‘Classification of Individual Consumption According to Purpose’ (COICOP) product classes and the two supermarket chains. In addition, the table also shows the weights the single COICOP classes have in the German CPI.
COICOP is published by the UN Department of Economic and Social Affairs. Products are grouped into divisions indicated by 2-digit codes, groups indicated by 3-digit codes, and classes indicated by 4-digit codes. COICOP is an integral part of the system of National Accounts and used, inter alia, for the establishment of weights for the computation of consumer price indexes and the calculation of purchasing power parities.
Source REWE Marktfinder and Billa Marktfinder
Geographical distribution of REWE and Billa stores. Notes The figure shows the geographic distribution of all 3693 physical Rewe stores in Germany and 1248 physical Billa stores in Austria.
Data source: Gesellschaft für Konsumforschung
Share of German household regularly buying at REWE & Co. Notes The figure shows the share of German households indicating that they regularly buy at REWE, EDEKA, ALDI, and LIDL.
Figures 22 and 23 show the composition of regular REWE, EDEKA, LIDL, and ALDI customers in terms of household income (differentiating between three income brackets: less than 1000 EUR, 1000–2000 EUR, and more than 2000 EUR) as well as the population number of the town where the customers live (four brackets: less than 20,000 inhabitants, 20,000–100,000 inhabitants, 100,000–500,000 inhabitants, more than 500,000 inhabitants). The figures show that with regard to household income, the composition of the customers of the four supermarket chains is very similar. When it comes to town size, though, REWE has a larger share of customers living in large cities and a smaller share of customers residing in small towns.
Data source: Gesellschaft für Konsumforschung
Share of German household regularly buying at REWE & Co.—by Income Group. Notes The figure shows the share of German households indicating that they regularly buy at REWE, EDEKA, ALDI, and LIDL depending on household income.
Data source: Gesellschaft für Konsumforschung
Share of German household regularly buying at REWE & Co.—by town size. Notes The figure shows the share of German households indicating that they regularly buy at REWE, EDEKA, ALDI, and LIDL, depending on the population size of the town where the household lives.
Appendix B: Additional tables and figures
See Tables 4, 5, 6 and Figs. 24, 25, 26, 27, 28, 29, 30, 31.
Comparison of prices levels—reduced set of products. Notes The density plot is based on price data collected from the REWE and Billa online shops in the first week of June which serves at the base period in our analysis
Data source: Eurostat
Differences between prices in Germany and Austria since 2017. Notes The figure shows the absolute difference between the average monthly price indexes for 168 product groups across Germany and Austria (cf. Fig. 8). The whiskers indicate 95% confidence intervals.
Data source: Eurostat
Differences between prices for selected product groups in Germany and Austria since 2017. Notes The figure shows the absolute difference between the average monthly price indexes for food and non-alcoholic beverages, alcoholic beverages and tobacco, and hygiene and cosmetic products across Germany and Austria (cf. Fig. 7). The whiskers indicate 95% confidence intervals.
Results of event study analysis (unweighted regression): impact of the VAT Rate Cut on Prices. Notes The figure shows the coefficient estimates for the event dummies along with the 95% confidence intervals. The y-axis indicates the price change in percent. The first week of June serves as a reference and the corresponding coefficient estimate is zero. Results are based on Eq. (1). The first solid vertical line indicates the day of the VAT rate cut (July 1, 2020), and the second solid vertical line indicates the day of the expiration of the VAT rate cut (January 1, 2021). Standard errors are clustered at the product class level (COICOP 4-digit level). The number of products is 63,841, and the number of clusters is 47
Results of event study analysis (unweighted regression): pass-through of the VAT rate cut to prices. Notes The figure shows the coefficient estimates when the event study indicators measure the change in the VAT burden relative to a product’s after-tax price along with the 95% confidence intervals. The y-axis indicates the pass-through to prices (\(-0.5\) means that the pass-through is 50%, and 1 means that the pass-through is 100%). The first week of June serves as a reference and the corresponding coefficient estimate is zero. Results are based on Eq. (1). The first solid vertical line indicates the day of the VAT rate cut (July 1, 2020), the second solid vertical line indicates the day of the expiration of the VAT rate cut (January 1, 2021). Standard errors are clustered at the product class level (COICOP 4-digit level). The number of products is 63,841, and the number of clusters is 47
Results of event study analysis (event dummies): reduced product set. Notes The figure shows the coefficient estimates for the event dummies along with the 95% confidence intervals. The y-axis indicates the price change in percent. The first week of June serves as a reference and the corresponding coefficient estimate is zero. Results are based on Eq. (1). The first solid vertical line indicates the day of the VAT rate cut (July 1, 2020), and the second solid vertical line indicates the day of the reversal of the VAT rate cut (January 1, 2021). Standard errors are clustered at the product class level (COICOP 4-digit level). The number of products is 45,388, and the number of clusters is 26
Results of event study analysis (pass-through rate): reduced product set. Notes The figure shows the coefficient estimates when the event study indicators measure the change in the VAT burden relative to a product’s after-tax price along with the 95% confidence intervals. The y-axis indicates the pass-through to prices (\(-0.5\) means that the pass-through is 50%, and 1 means that the pass-through is 100%). The first week of June serves as a reference and the corresponding coefficient estimate is zero. Results are based on Eq. (1). The first solid vertical line indicates the day of the VAT rate cut (July 1, 2020), and the second solid vertical line indicates the day of the expiration of the VAT rate cut (January 1, 2021). The number of products is 45,388, and the number of clusters is 25
Results of event study analysis (pass-through rate): placebo tests. Notes The figure shows the results of placebo tests. For the test, the treatment date was moved 6 (upper panel), 10 (middle panel) and 14 (lower panel) weeks back. Dark blue dots indicate coefficient estimates, and light blue areas indicate 95% confidence intervals. Results are based on Eq. (1). The solid vertical line indicates the day of the placebo treatment. Standard errors are clustered at the product class level (COICOP 4-digit level). The number of clusters is 47
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Fuest, C., Neumeier, F. & Stöhlker, D. The pass-through of temporary VAT rate cuts: evidence from German supermarket retail. Int Tax Public Finance (2024). https://doi.org/10.1007/s10797-023-09824-7
Accepted:
Published:
DOI: https://doi.org/10.1007/s10797-023-09824-7