Consumption Tax Incidence: Evidence from

the Natural Experiment in the

Czech Republic

Jan Z�apal �z

j.zapal@lse.ac.uk

�rst draft: October, 2007this draft: October, 2007

� PhD program, London School of Economics and Political Science, Houghton Street,London WC2A 2AE, UK

z Institute of Economic Studies, Faculty of Social Sciences, Charles University,Opletalova 26, Prague 1, 110 00, Czech Republic

I must acknowledge help of Ji�r�� Trexler of Czech Statistical O�ce with the data andhelp of Martin Jare�s of Czech Ministry of Finance with the classi�cation of goods intostandard/reduced VAT rate groups. All remaining errors are mine.

Abstract

This paper tries to estimate the incidence of consumption taxation.

We use data from the natural experiment that took place in 2004 in

the Czech Republic. Not only value added tax (VAT) rates applicable

to range of goods and services changed but also the classi�cation into

standard vs. reduced rate group has been modi�ed. Most importantly,

some of the goods and services experienced no change. This allows us

to use di�erence-in-di�erences estimates to assess the extent to which

taxes are shifted onto consumers.

Our estimates indicate that those goods and services which expe-

rienced decline of the VAT rate from 22% to 19% show now evidence

of decrease in price. We interpret this as the evidence of producers

and vendors taking the full advantage of the tax decline.

On the other hand goods and services belonging to the group which

experienced VAT rate increase from 5% to 19% show lasting increase

of price by up to 6%. This indicates that the higher tax is at least

partially shifted onto the consumers.

JEL classi�cation: H22

Keywords: tax incidence, value added tax, di�erence-in-di�erences

estimation, natural experiment

1 Introduction

Incidence of taxation comprises one of the core topics of public economics.Distinction between those who merely collect the taxes and send the revenueto the government and those who's income changes as a result of the tax hasfascinated generations of economists. Focusing only on indirect taxation, thequestion becomes of how imposition or change in the relevant tax a�ects theprice of the commodity in question.

At the theoretical level the answer is far from clear. Existing models onthe topic deal either with ad valorem tax where the amount of the tax isexpressed as a percentage of the producer's price or with speci�c (excise) taxwhere the amount of the tax is expressed per unit of relevant commodity. Ineither case, some models predict over-shifting, i.e. price of the commodityrises by more than the full amount of the tax, while some models predictunder-shifting, i.e. price of the commodity rises by less than the full amountof the tax (for survey of the models see Fullerton and Metcalf (2002)). Factorsthat in uence those results usually include assumed market structure, degreeof product di�erentiation or elasticity of demand and supply.

At the same time empirical literature estimating degree to which indirecttaxes are shifted on consumers is rather scant. Several studies support theidea of over-shifting. Brownlee and Perry (1967) �nd the evidence of full-shifting following 1965 excise tax reduction in the United States. Usingthe same natural experiment Woodward and Siegelman (1967) analyze thechanges in the prices of automotive replacement parts concluding with lessthan full-shifting. Barzel (1976) and Johnson (1978) �nd evidence of over-shifting using data on the cigarette prices in the US (Sumner and Ward(1981) refute their results). Poterba (1996) �nds over-shifting of the salestaxes (American version of ad valorem tax) using clothing prices followedover the 1925-39 and 1947-77 periods in the series of US cities. Estimates inBesley and Rosen (1999) support over-shifting of sales taxes for at least halfout of the 12 commodities used in the study covering 155 US cities in the1980's.

On the other hand some empirical evidence supports under-shifting. Deli-palla and O'Donnell (2001) analyze European cigarette industry and con-clude that both ad valorem and speci�c taxes tend to be under-shifted. Sim-ilar conclusion reaches Carbonier (2007) using value added tax (Europeanversion of ad valorem tax) reforms in France focusing on housing repair ser-vices and new car market.

1

Given the importance of the question no more than a dozen studies israther surprising. Further discounted by the inde�niteness of their results,economics has little to o�er both to public and interested policy-makers.

We hope to contribute to the topic by analyzing Czech value added tax(VAT) reform of 2004. Not only the standard rate declined from 22% to 19%but also the composition of classes of commodities to which standard andreduced rates apply has changed. Most importantly, certain commoditiesexperienced no change at all and serve us a purpose of control group againstwhich we can measure the e�ect of the reform.

The paper proceeds as follows. Next part explains in detail the natureof the Czech VAT reform and describes the data we use. Part 3 describesthe methodology used to estimate the extent of tax shifting that followed thereform. Here we also check whether the data are consistent with the assump-tions we need to make in order to be able to proceed with the estimation.Ensuing part 4 shows the main results of the paper while part 5 concludesthe paper. In the appendix we further check the robustness of the results.

2 Natural experiment design and data

A natural experiment we exploit for the research purposes is the Czech VATreform of 2004 with all measures coming to force on May 1st 2004. There weretwo main reasons for the VAT change. First one was the requirement to alignCzech VAT legislation with the European sixth directive which prescribesrules for the VAT legislation in the EU member states. Second reason for thereform was an attempt to bring down ever increasing public budget de�cit.

The reform had two main component. First, existing standard rate of 22%has been reduced to 19%. We call commodities that experienced this typeof change `treated 1' or T1 for short. Second, many commodities to whichreduced VAT rate of 5% applied previously were relocated to the categoryto which the new standard rate of 19% applies. We use `treated 2' or T2 forthis class. Commodities that were previously in the reduced VAT rate groupand were not relocated subsequently experienced no change at all. This isour `control' group.

To give a avour of commodities in the di�erent groups, the control groupincludes most of the food, medications, personal transportation, press andbooks and items previously exempt. T2 includes veterinary services, vita-mins, contraception, sport and cultural activity entrance fees, food served in

2

restaurants and certain services. Rest comprises the T1 group which includesfor example electronics, housewares, cosmetics, alcohol or tobacco.

The data we use are monthly observations of prices of the commoditiesincluded in the consumption basket used for consumption price index (CPI)calculation by the Czech Statistical O�ce. The data span the whole 2004year and include 790 di�erent commodities.1 Consumption basket is chosensuch as to represent the composition of household consumption. This fact, wehope, increases the relevance of our results being based on the large numberof diverse commodities.

With respect to the VAT reform, we classi�ed 322 commodities into thecontrol group, 408 commodities into the T1 class and remaining 60 into theT2 group. In what follows we use logs of the prices from the original data.This brings additional advantage in that our econometric results then havesimple interpretation of the percentage changes.

Since April 2004 is the last month before the reform, we denote it `month0' with the negative values denoting months before the reform decreasing to`month -3', January 2004. `Month 1' is then �rst month of the new system,May, and positive values denote months after the reform going up to `month8', December 2004.

As a �rst look at the data, we calculated mean log-price for each monthand each of the tree groups. Figure 1 shows the results.

Figure 1: Average log-price in control and treated groups

5.83

5.86

5.89

T1

4.57

4.6

4.63

Con

trol

-5 0 5 10month

Control T1

Average log-price, control vs. first treated

3.82

3.85

3.88

T2

4.57

4.6

4.63

Con

trol

-5 0 5 10month

Control T2

Average log-price, control vs. second treated

-.025

-.02

-.015

-.01

-.005

0

0 2 4 6 8months into treatment

0 -1-2 -3

Base month

Note: full tax shift is -3%

Estimates of tax incidence, T1

.03

.04

.05

.06

0 2 4 6 8months into treatment

0 -1-2 -3

Base month

Note: full tax shift is +14%

Estimates of tax incidence, T2

1 See www.czso.cz for the details on the methodology of the data collection. 790 is morethan 730 actually used for CPI. The discrepancy comes from the fact that as some itemsare being introduced and some phased out the data include more items than is needed.

3

Close inspection of the left panel shows that, at least graphically, thereis little evidence of the tax shifting in the T1 group. Full-shifting wouldrequire sustained decrease of the solid curve by 0.03 since the prices are inlogarithms. On the other hand, right panel of the �gure reveals the increaseof the log-price of T2 commodities by more than 0.03, i.e. more that 3%increase in the prices on average. Although compelling it is far from 14%increase required for the full-shifting.

Figure 1 makes another important point. As will become clear shortly,the validity of our estimates heavily rests on the assumption that the devel-opment in the control and treated group prior to policy change is the same.In other words, for the estimates to be valid, we need to assume that themean log-price in the control and treated group had the same trend priorto the reform. This allows us to conjecture that, absent the reform, the dif-ference between the mean log-price in the control and treated group wouldremain the same into the future.

Whereas we can never test the conjecture regarding the developmentswithout the reform, we can test the hypothesis that the di�erence in themean log-price between the groups is the same in the four months before thereform. Inspection of the �gure 1 then shows that the hypothesis is not likelyto be rejected.

3 Methodology

This section explain the rationale and logic of the di�erence-in-di�erences(D-i-D) estimation we are about to use to estimate the extent to which VAThas been shifted following the 2004 reform.2

Suppose a researcher is asked to assess the e�ect of certain either naturalor controlled experiment on the variable of interest. She is presented withthe data on this variable. Furthermore, each observation indicates whetherit has been made either before or after the experiment and whether it comesfrom the control or treated group.

In general, D-i-D estimation acknowledges the di�erence in the variableof interest between the treated and control groups and uncovers the e�ectof the experiment as a di�erence in those di�erences before and after theexperiment, hence its name.

2 See Angrist and Krueger (1999) for more in-depth discussion of di�erence-in-di�erences estimation and Meyer (1995) for the discussion of its possible pitfalls.

4

Figure 2 shows stylized example. Development of the variable of interestin both groups is captured by the solid lines. Straight line for the controlgroup indicates steady trend due to the absence of an experiment relatedchange. On the other hand change in the slope of the treated group linecaptures the e�ect of the experiment on the variable of interest.

Figure 2: Di�erence-in-di�erences estimation

-

6

time-1 0 1policy change

����

����

����

����

����

����

������

�����������

��������������

control

treated

6

?

6

?

6

?

6

?

�

�

�

r

r

r

r

r

r

r

Uncovering the e�ect of the experiment means estimating � from theavailable data. There are numerous ways to do so. One of them is to computethe mean of the variable of interest. For control group before the experimentthis gives �, for the control group after the experiment �+ , for the treatedgroup before the experiment �+� and �nally for the treated group after theexperiment � + � + + �. � is then simply calculated from the estimatedmeans.

Rather more convenient way of the estimation which also readily providesthe variance of the estimates is running the following regression

ln(pi) = � + � Ti + Ai + � (Ti � Ai) + �i (1)

where we already use notation relevant to our data. For the dependentvariable, ln(pi) denotes the log-price of the commodity i, dummy variable Ti

5

indicates whether the observation comes from the control or treated group(unity for treated), dummy variable Ai indicates whether the observationcomes from before or after the experiment (unity for after) and �i is the errorterm.

Notice that the use of the same �, �, and � in the �gure 2 and equation(1) is not coincidental. For the observations from the control group beforethe experiment both dummy variables will always be zero and the estimate of� from the regression will be simply mean of ln(pi) in this group. Similarly,estimate of �+ � from the regression is mean of ln(pi) for the treated groupbefore the experiment as only Ti dummy will be unity. Exactly the same logicapplies to both groups after the experiment. The advantage of regressionbased estimation is that it provides the variance of estimated � which allowsfor standard hypothesis testing.

We must stress that the validity of the D-i-D heavily rests on the assump-tion that if it were not for the reform, the di�erence between the control andtreated group would remain the same. With reference to �gure 2 this assump-tion is equivalent to assuming that the solid control group line and dashedtreated group line after the policy change are parallel.

While there is no way how to test this equal trend assumption, we caninfer how likely is it to hold from the development before the policy change.To do so, we estimate �'s for the four months before the VAT reform an testwhether they are equal. Table 1 shows the results of the test and convey themessage that the assumption we need to proceed with the D-i-D estimationis likely to hold in the data for both treated groups.

Our empirical strategy warrants few further comments. In general, D-i-Destimation does not require panel data. In other words, observations on thevariable of interest before and after the experiment can come from di�erentindividuals as long as they can be unambiguously classi�ed into control ortreated group.

If the data indeed have panel structure and include the observation be-fore and after the experiment for each individual as our data do, standarderrors estimated by conventional method can be invalid due to possible cor-relation of unobservable error for each individual. To overcome this problemwhen computing standard errors we cluster on individual commodities of theconsumption basket.

Lastly, up to now we have distinguished only before and after the exper-iment periods. Although su�cient for D-i-D estimation, our data have theadvantage that for each commodity they include four monthly observations

6

Table 1: Test of hypothesis of equal trends before the treatment

T1 January February MarchFebruary 0.00 (1).

0.990March 0.00 (1) 0.00 (1).

0.985 0.995April 0.00 (1) 0.00 (1) 0.00 (1)

0.980 0.990 0.995

T2 January February MarchFebruary 0.00 (1).

0.979March 0.00 (1) 0.00 (1).

0.978 0.998April 0.00 (1) 0.00 (1) 0.00 (1)

0.978 0.998 1.000

Note: Test of the null hypothesis of equal trends in the treated and control group before

the treatment. Comparing the di�erence between mean log-price in the treated and control

group in column vs. row months. �2 and (degrees of freedom) of the test in the upper

part of each cell. p-value of the test in the lower part (probability that the null hypothesis

is the correct one).

before the reform and eight monthly observations after the reform.Question then becomes which observations to choose for the actual esti-

mation. For the benchmark results we present in the next section, we useApril as base month for the period before the reform. Individual columnsthen correspond to di�erent months used for the period after the reform giv-ing us eight estimates of the extent of tax shifting. Additional advantage ofthis approach is that we are able to see the development of tax shifting overthe time. Appendix then includes similar tables for both treated groups withdi�erent base months for the period before the reform.

7

4 Results

We are now in the position to present our main results. Table 2 depicts theresults for the �rst treated group and table 3 for the second treated group.

Each column in both tables estimates the speci�cation in (1) where Ai

becomes Ati for t 2 f1 : : : 8g and denotes the di�erent months after thereform used in the estimation. For example �fth column of table 2 estimatesthe degree of tax shifting for the commodities from the �rst treated groupT1. In doing so the regression includes observations on log-prices from April2003, our base month for the whole table, representing the period before thereform. For the period after the reform �fth column uses data from month5, i.e. September 2004.

The estimates have straightforward interpretation explained in detail inthe previous section. Since all the prices are in logarithms the estimatedcoe�cients have the interpretation of percentage changes. Estimate of taxe�ect of 0.03 say, then means that the price of relevant commodities increasesby 3% as a result of the reform.

Inspection of table 2 reveals that for those commodities where the VATrate decrease from 22% to 19% resulting change in the prices is rathermarginal. Largest decrease in the table 2 can be found in the �fth col-umn. Yet it still reaches only -1.9% and is not signi�cant. Inspection ofother columns reveals similar picture. All the estimated tax e�ect are notsigni�cant. We interpret this result as the evidence of producers and vendorstaking the advantage of the VAT rate decrease.

On the other hand table 3 reveals completely di�erent picture for thecommodities that has been reclassi�ed from the 5% VAT rate group into the19% VAT rate group. The estimates of the e�ect of the reform on the pricerange from 3.3% in the �rst month after the reform to 5.5% in the �fth monthafter the reform.

Furthermore, all the tax e�ect estimates are statistically signi�cant indi-cating that the e�ect lasts well after the reform. Taking the 5.5% increaseat a face value means that roughly 40% of the tax has been shifted on theconsumers (5.5 percentage points out of 14).

Hence our results point to the asymmetry in the tax shifting since increasein the VAT rate has been re ected in prices while decrease in the VAT rateleft the prices unchanged. While certainly in uenced by the extend of thechange we suspect it to be a manifestation of the more general pattern.

8

Table2:

Estimates

oftaxincidence

withAprilas

abasemonth,�rsttreatedgroupT1

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

-0.003

-0.002

0.001

-0.014

-0.019

0.002

-0.003

-0.011

(0.003)

(0.005)

(0.008)

(0.020)

(0.021)

(0.010)

(0.008)

(0.008)

T1

1.263���

1.263���

1.263���

1.263���

1.263���

1.263���

1.263���

1.263���

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

A1

0.002

(0.002)

A2

0.003

(0.004)

A3

-0.002

(0.007)

A4

-0.007

(0.009)

A5

-0.008

(0.010)

A6

-0.004

(0.009)

A7

0.000

(0.007)

A8

0.006

(0.007)

constant

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N1460

1460

1460

1460

1460

1460

1460

1460

R2

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

Note:T1isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT1andAt.

Fulltaxshiftingwould

requiretaxe�ectestimatesof-3%

or-0.03.Robust

clustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

9

Table3:

Estimates

oftaxincidence

withAprilas

abasemonth,secondtreatedgroupT2

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

0.033���

0.035���

0.043���

0.050���

0.055���

0.052���

0.051���

0.046���

(0.004)

(0.006)

(0.009)

(0.010)

(0.011)

(0.010)

(0.009)

(0.009)

T2

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

A1

0.002

(0.002)

A2

0.003

(0.004)

A3

-0.002

(0.007)

A4

-0.007

(0.009)

A5

-0.008

(0.010)

A6

-0.004

(0.009)

A7

0.000

(0.007)

A8

0.006

(0.007)

constant

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

4.598���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N764

764

764

764

764

764

764

764

R2

0.03

0.03

0.03

0.03

0.03

0.03

0.03

0.03

Note:T2isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT2andAt.Fulltaxshiftingwould

requiretaxe�ectestimatesof+14%

or+0.14.Robustclustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

10

We further note that the extent of the tax shifting indicated by the resultseven for the T2 group belongs to the lower range compared to the empiricalresults brie y surveyed in the introduction. Hence, our results are morein line with those studies that �nd under-shifting, Delipalla and O'Donnell(2001) and Carbonier (2007). Coincidentally, the very same studies dealwith the VAT rather than the sales or speci�c taxes which are in focus of thestudies supporting over-shifting.

As already hinted, we re-run all the estimations using di�erent basemonths to check the robustness of our �ndings. Although detailed tables areincluded in the appendix we summarize the results using �gure 3 which showsthe estimated tax e�ects. Di�erent lines represent di�erent base months usedand the horizontal axis denotes the month after the reform used in the esti-mation. With reference to �gure 3 we note the robustness of our �ndings.

Figure 3: Estimated tax incidence with di�erent base months

5.83

5.86

5.89

T1

4.57

4.6

4.63

Con

trol

-5 0 5 10month

Control T1

Average log-price, control vs. first treated

3.82

3.85

3.88

T2

4.57

4.6

4.63

Con

trol

-5 0 5 10month

Control T2

Average log-price, control vs. second treated

-.025

-.02

-.015

-.01

-.005

0

0 2 4 6 8months into treatment

0 -1-2 -3

Base month

Note: full tax shift is -3%

Estimates of tax incidence, T1.0

3.0

4.0

5.0

6

0 2 4 6 8months into treatment

0 -1-2 -3

Base month

Note: full tax shift is +14%

Estimates of tax incidence, T2

11

5 Conclusion

This paper tries to assess the extent of tax shifting of the VAT. For thispurpose we use natural experiment - Czech 2004 VAT reform. Using monthlydata on prices of almost eight hundred commodities included in the CPIbasket we use di�erence-in-di�erences estimation. Two main conclusionsemerge based on our �ndings.

Firstly, for those commodities that experienced decrease in the applicableVAT rate from 22% to 19% there is no evidence of the e�ect of the reformon the prices.

Secondly, for those commodities for which VAT rate increased from 5%to 19% due to reclassi�cation there is evidence of less than full tax shifting.Increase of the VAT rate by 14 percentage points translates into at most 6%increase in the prices of the a�ected goods. Nevertheless, the estimates arestatistically and we believe also economically signi�cant.

Furthermore, the asymmetry in our results hints on the asymmetry in thetax incidence in general. We leave this conjecture for the future research aswell as the observation that the tax incidence studies dealing with VAT asopposed to sales or speci�c taxes in general conclude with the lower extentof the tax shifting.

References

[1] Angrist, J. D. and A. B. Krueger, (1999) `Empirical Strategies in LaborEconomics', Handbook of Labor Economics, Elsevier, Volume 3, Part 1:1277-1366.

[2] Barzel, Y., (1976) `An Alternative Approach to the Analysis of Taxation',Journal of Political Economy, 84(6): 1177-1197.

[3] Besley, T. J. and H. S. Rosen, (1998) `Sales Taxes and Prices: An Em-pirical Analysis', NBER working paper, no. 6667.

[4] Brownlee, O. and G. L. Perry, (1967) `The E�ects of the 1965 FederalExcise Tax Reduction on Prices', National Tax Journal, 20(3): 235-249.

[5] Carbonnier, C., (2007) `Who Pays Sales Taxes? Evidence from FrenchVAT Reforms, 1987-1999', Journal of Public Economics, 91(5-6): 1219-1229.

12

[6] Delipalla, S. and O. O'Donnell, (2001) `Estimating Tax Incidence, MarketPower and Market Conduct: The European Cigarette Industry', Interna-tional Journal of Industrial Organization, 19(6): 885-908.

[7] Fullerton, D. and G. E. Metcalf, (2002) `Tax Incidence', NBER workingpaper, no. 8829.

[8] Johnson, T. R., (1978) `Additional Evidence on the E�ects of AlternativeTaxes on Cigarette Prices', Journal of Political Economy, 86(2): 325-328.

[9] Meyer, B. D., (1995) `Natural and Quasi-Experiments in Economics',Journal of Business & Economic Statistics, 13(2): 151-161.

[10] Poterba, J. M., (1996) `Retail Price Reactions to Changes in State andLocal Sales Taxes', National Tax Journal, 49(2): 165-176.

[11] Sumner, M. T. and R. Ward, (1981) `Tax Changes and Cigarette Prices',Journal of Political Economy, 89(6): 1261-1265.

[12] Woodward, F. O. and H. Siegelman, (1967) `E�ects of the 1965 FederalExcise Tax Reduction upon the Prices of Automotive Replacement Parts,A Case Study in Tax Shifting', National Tax Journal, 20(3): 250-257.

6 Appendix

This appendix checks the robustness of the results from the main part of thepaper. Tables 4, 5 and 6 di�er from the table 2 only in using di�erent basemonth for the estimation. Similar di�erence links tables 7, 8, 9 and table 3.The tax e�ect estimates are summarized in �gure 3.

13

Table4:

Estimates

oftaxincidence

withMarch

asabasemonth,�rsttreatedgroupT1

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

-0.005

-0.004

0.000

-0.016

-0.020

0.000

-0.005

-0.012

(0.004)

(0.005)

(0.009)

(0.021)

(0.021)

(0.010)

(0.008)

(0.008)

T1

1.264���

1.264���

1.264���

1.264���

1.264���

1.264���

1.264���

1.264���

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

A1

0.003

(0.002)

A2

0.003

(0.004)

A3

-0.001

(0.008)

A4

-0.006

(0.009)

A5

-0.007

(0.010)

A6

-0.003

(0.009)

A7

0.001

(0.007)

A8

0.007

(0.006)

constant

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N1460

1460

1460

1460

1460

1460

1460

1460

R2

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

Note:T1isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT1andAt.

Fulltaxshiftingwould

requiretaxe�ectestimatesof-3%

or-0.03.Robust

clustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

14

Table5:

Estimates

oftaxincidence

withFebruaryas

abasemonth,�rsttreatedgroupT1

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

-0.006

-0.005

-0.001

-0.017

-0.022

-0.001

-0.006

-0.014*

(0.004)

(0.006)

(0.009)

(0.021)

(0.021)

(0.010)

(0.009)

(0.008)

T1

1.265���

1.265���

1.265���

1.265���

1.265���

1.265���

1.265���

1.265���

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

A1

0.004

(0.003)

A2

0.005

(0.004)

A3

0.000

(0.008)

A4

-0.005

(0.009)

A5

-0.006

(0.010)

A6

-0.002

(0.009)

A7

0.002

(0.008)

A8

0.008

(0.007)

constant

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N1460

1460

1460

1460

1460

1460

1460

1460

R2

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

Note:T1isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT1andAt.

Fulltaxshiftingwould

requiretaxe�ectestimatesof-3%

or-0.03.Robust

clustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

15

Table6:

Estimates

oftaxincidence

withJanuaryas

abasemonth,�rsttreatedgroupT1

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

-0.009**

-0.008

-0.004

-0.020

-0.024

-0.004

-0.009

-0.017**

(0.005)

(0.006)

(0.009)

(0.023)

(0.023)

(0.010)

(0.009)

(0.008)

T1

1.268���

1.268���

1.268���

1.268���

1.268���

1.268���

1.268���

1.268���

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

(0.162)

A1

0.004

(0.004)

A2

0.005

(0.005)

A3

0.000

(0.009)

A4

-0.005

(0.010)

A5

-0.006

(0.011)

A6

-0.002

(0.010)

A7

0.002

(0.008)

A8

0.008

(0.007)

constant

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N1460

1460

1460

1460

1460

1460

1460

1460

R2

0.07

0.07

0.07

0.07

0.07

0.07

0.07

0.07

Note:T1isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT1andAt.

Fulltaxshiftingwould

requiretaxe�ectestimatesof-3%

or-0.03.Robust

clustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

16

Table7:

Estimates

oftaxincidence

withMarch

asabasemonth,secondtreatedgroupT2

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

0.033���

0.035���

0.043���

0.050���

0.055���

0.052���

0.051���

0.046���

(0.005)

(0.006)

(0.009)

(0.011)

(0.011)

(0.010)

(0.009)

(0.009)

T2

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

-0.759���

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

A1

0.003

(0.002)

A2

0.003

(0.004)

A3

-0.001

(0.008)

A4

-0.006

(0.009)

A5

-0.007

(0.010)

A6

-0.003

(0.009)

A7

0.001

(0.007)

A8

0.007

(0.006)

constant

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

4.597���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N764

764

764

764

764

764

764

764

R2

0.03

0.03

0.03

0.03

0.03

0.03

0.03

0.03

Note:T2isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT2andAt.Fulltaxshiftingwould

requiretaxe�ectestimatesof+14%

or+0.14.Robustclustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

17

Table8:

Estimates

oftaxincidence

withFebruaryas

abasemonth,secondtreatedgroupT2

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

0.033���

0.036���

0.043���

0.050���

0.055���

0.053���

0.052���

0.046���

(0.005)

(0.007)

(0.009)

(0.011)

(0.012)

(0.011)

(0.010)

(0.009)

T2

-0.760���

-0.760���

-0.760���

-0.760���

-0.760���

-0.760���

-0.760���

-0.760���

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

A1

0.004

(0.003)

A2

0.005

(0.004)

A3

0.000

(0.008)

A4

-0.005

(0.009)

A5

-0.006

(0.010)

A6

-0.002

(0.009)

A7

0.002

(0.008)

A8

0.008

(0.007)

constant

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N764

764

764

764

764

764

764

764

R2

0.03

0.03

0.03

0.03

0.03

0.03

0.03

0.03

Note:T2isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT2andAt.Fulltaxshiftingwould

requiretaxe�ectestimatesof+14%

or+0.14.Robustclustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

18

Table9:

Estimates

oftaxincidence

withJanuaryas

abasemonth,secondtreatedgroupT2

Dependentvariable:logofprice

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

taxe�ect

0.039���

0.042���

0.049���

0.056���

0.061���

0.059���

0.058���

0.052���

(0.006)

(0.007)

(0.010)

(0.012)

(0.012)

(0.011)

(0.010)

(0.009)

T2

-0.766���

-0.766���

-0.766���

-0.766���

-0.766���

-0.766���

-0.766���

-0.766���

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

(0.160)

A1

0.004

(0.004)

A2

0.005

(0.005)

A3

0.000

(0.009)

A4

-0.005

(0.010)

A5

-0.006

(0.011)

A6

-0.002

(0.010)

A7

0.002

(0.008)

A8

0.008

(0.007)

constant

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

4.596���

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

(0.094)

N764

764

764

764

764

764

764

764

R2

0.03

0.03

0.03

0.03

0.03

0.03

0.03

0.03

Note:T2isdummyvariablefortreatedgroup.Atisdummyfort-th

month

into

thetreatm

ent.

Taxe�ectin

t-th

columnis

interactionterm

betweenT1andAt.Fulltaxshiftingwould

requiretaxe�ectestimatesof+14%

or+0.14.Robustclustered

standard

errors

(onindividualcommodities)

inparentheses.���,��,�denotessigni�cance

on1%,5%

and10%

respectively.

19