Not so sweet: Impacts of a soda tax on producers

doi:10.21203/rs.3.rs-2849458/v1

Download PDF

Research Article

Not so sweet: Impacts of a soda tax on producers

https://doi.org/10.21203/rs.3.rs-2849458/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 22 Nov, 2023

Read the published version in International Tax and Public Finance →

You are reading this latest preprint version

Portugal introduced a sugar-sweetened beverages (SSB) tax in 2017. This study uses unique administrative accounting data for all SSB producers/importers in Portugal, and an event study design with bottled water firms as the primary comparison group, to assess the causal impacts of the tax on multiple firm-level outcomes. We find a 6.8% average decrease in domestic SSB sales, vis-à-vis bottled water. The soda tax hindered SSB firms’ financial health, namely net income, ability to convert receivables into cash, and liabilities. SSB producers/importers did not decrease wages, cut jobs, or modify their workforce towards higher R&D capacity. Forgone corporate income tax appears negligible compared to the government revenue generated by the tax itself.

In 2016, the World Health Organization (WHO) urged policy makers to tax sugar-sweetened beverages (SSB, or soda, for short), motivated by the evident link between soda consumption and major diseases such as obesity and type 2 diabetes, and by growing evidence on the effectiveness of soda taxes for curbing sugar intake from soda (WHO, 2016). As of May 2022, more than 60 jurisdictions around the world had already implemented soda taxes (Global Food Research Program UNC, 2022).

This study documents the impacts of the Portuguese soda tax, introduced in 2017, on a set of firm-level outcomes of soda producers and importers, including for example sales, employment, or profits. We rely on a very rich administrative dataset that contains yearly accounting information from the profit and loss (P&L) statement and the balance sheet, as well as workforce-related information, for the universe of SSB producers and importers in Portugal, from 2012 to 2019. To estimate the causal impacts of the tax, we employ event studies and difference-in-differences models, using water bottling firms as primary comparison group.

We find a 6.8% average decrease in domestic SSB sales, vis-à-vis bottled water, and no effects on exports. The soda tax hindered SSB firms’ financial health, namely net income, ability to convert receivables into cash, and liabilities. SSB producers/importers did not decrease wages or cut employment. Forgone corporate income tax revenues for the State appear negligible when compared with the revenue generated by the new tax itself.

We make four main contributions to the literature on SSB taxes. First, by analyzing domestic sales of the universe of SSB producers/importers, we provide evidence on the impacts of soda taxes on total consumption (i.e., in- as well as out-of-home consumption). Almost all previous studies rely on store-level sales data (Castelló & Casasnovas, 2020; Dickson et al., 2021; Gonçalves & Pereira dos Santos, 2020; Seiler et al., 2021; Taylor et al., 2019) or consumer-level supermarket purchases data (Aguilar et al., 2021; Bollinger & Sexton, 2018; Capacci et al., 2019; Cawley et al., 2019a; Cawley et al., 2020; Colchero et al., 2016; Fearne et al., 2019; Fichera et al., 2021; Léger & Powell, 2021; Leider & Powell, 2022; Nakamura et al., 2018; Rojas & Wang, 2021; Silver et al., 2017), from one or more retailers. Other outlets like wholesalers, restaurants and bars, or vending machines —put differently, out-of-home soda consumption—, have largely been ignored (Cornelsen & Smith, 2018), even though in-home and out-of-home soda consumption can potentially respond very differently to soda taxes (Law et al., 2022). The fewer studies that are not limited to retail sales either rely on survey data on all purchases/soda intake (Cawley et al., 2022; Colchero et al., 2017), or use macro level data, relying solely on time variations (Alsukait et al. 2020; Arteaga et al., 2021). Overall, most studies find that soda taxes reduce SSB consumption, including in Portugal (Gonçalves and Pereira dos Santos, 2020). The estimated reductions vary in size, depending on study setting and methodology, as well as consumer income, age, and baseline consumption level (Allcott et al., 2019b; Colchero et al., 2015; Dubois et al., 2020; Sharma et al., 2014).

The second main contribution is that we indirectly explore manufacturers’ reformulation activity, by looking at changes in the firms’ workforce, namely the number of employees working in research and development (R&D). This contribution relates to the specific design of the Portuguese soda tax, the first multi-tier soda tax in the world. The tax is levied on producers/importers and is structured in several brackets, based on drinks’ sugar content. This multi-tier design appears to have incentivized soda producers to reformulate recipes towards lower sugar content.1 Recipe reformulation is a main channel through which multi-rate soda taxes can reduce sugar intake from soda, besides reducing soda consumption. Studies show the superiority of multi-rate soda taxes in terms of welfare (O’Connell & Smith, 2021), and economic and public health gains (Grummon et al., 2019). Since Portugal implemented its soda tax in 2017, and reports of reformulation and reductions in soda consumption started to emerge, other countries were motivated to (re)design their soda taxes in a similar manner, e.g., France, Ireland, the UK. If substantial reformulation activity was going on around the time the tax was implemented, then we expect to find a positive impact of the Portuguese soda tax on the number of employees working in R&D, and potentially higher average wages. The only study, to date, that directly explores the effects of a soda tax on reformulation is Dickson et al. (2021). The authors estimate that the UK soda tax reduced calorie intake from soda by around 6,500 calories per annum per resident, with more than 80% of that reduction attributable to manufacturers’ reformulation activities.

The third main contribution is that we consider the impacts of a soda tax for economic agents that have been largely overlooked in this literature, namely producers/importers and, indirectly, workers. Similarly to other countries, in Portugal the soda tax is levied on producers/importers. Producers/importers can adjust to the new tax along two main margins. The first is specific to the case of multi-tier taxes, like the Portuguese one: producers may reduce the sugar content of their drinks to pay a lower tax. This option is limited by consumer preferences, because if consumers dislike the new recipe, they will stop buying. Reformulation also entails costs, with R&D, relabeling, rebranding, as well as renegotiations with retailers and other clients. As stated above, there is evidence of reformulation activity following Portugal’s and UK’s soda taxes. The second main margin of adjustment is changing prices: producers/importers must decide how much of the tax to absorb, and how much to pass on to retailers, wholesalers, restaurants, and other clients. This will depend on multiple factors, such as the relative market power of each agent, price elasticity of demand, and firms’ drinks portfolio. All available evidence on soda tax pass-through pertains to overall shifts to final consumer prices (Aguilar et al., 2021; Alsukait et al., 2020; Berardi et al., 2016; Bollinger & Sexton, 2018; Capacci et al., 2019; Cawley & Frisvold, 2017; Cawley et al., 2018a; Cawley et al., 2018b; Dickson et al., 2021; Etilé et al., 2018; Gonçalves & Pereira dos Santos, 2020; Grogger, 2017; Léger & Powell, 2021; Leider & Powell, 2022; Rojas & Wang, 2021; Seiler et al., 2021; Silver et al., 2017; Stacey et al., 2019). Pass-through to consumer prices is usually large; however, it is unclear how much of it is attributable to producers/importers, and how much to retailers.2 Overall, few studies have considered the impacts of soda taxes for SSB producers/importers and workers. Using time series data, Law et al. (2020a, b) find short-lived negative impacts of the UK soda tax on stock returns and domestic turnover of UK soda manufacturers. Guerrero-López et al. (2017) and Lawman et al. (2019) find no aggregate unemployment effects of soda taxes in Mexico and Philadelphia. In this paper, we consider firms’ “financial health”, which we assess based on net income (an indicator of profitability: the difference between total income revenues and total expenses), and cash, receivables, and liabilities (which together provide insights on liquidity and solvency), as well as employment and wages.

The fourth and last main contribution is an estimation of the impact of the soda tax on corporate income tax payments. This is a relevant outcome from a government revenue perspective, as forgone corporate income taxes may partly offset the additional revenue from the soda tax.

To sum up, the literature on SSB taxes is extensive, but still has some gaps (see also Allcott et al., 2019a; Andreyeva et al., 2022; and Cawley et al., 2019b for recent reviews). Specifically, evidence on the impacts of soda taxes on consumption is mostly limited to in-home consumption, as most studies use data covering only retail sales. Other agents besides consumers and retailers, like producers, importers, and workers, have received little attention. With this study we contribute with evidence on the impacts of soda taxes for producers/importers and their workforce, and provide insights about total consumption effects (both in- and out-of-home), thanks to having data on domestic sales of the universe of soda producers and importers.

[1] According to industry data, Portuguese manufacturers reduced the sugar content of some drinks, even though this was an ongoing trend even before the tax was introduced. The change in the caloric content per 100 milliliters of non-alcoholic beverages sold in Portugal was −11% from 2016 to 2017 (Goiana-da-Silva et al., 2020; Grupo de Trabalho, 2018).

[2] There is evidence of full shifts of other taxes to consumer prices (Lyssiotou and Savva, 2021). Rozema (2018) studies the impact of taxes on cigarettes to understand how the burden of these taxes not borne by consumers is shared between upstream and downstream firms. Using Nielsen Homescan data, the author suggests that taxes are passed through to both wholesale and retail prices, with downstream firms bearing no more than one-third of the tax burden.

The Portuguese soda tax was first mentioned in the Portuguese media on May 5, 2016, in the newspaper Expresso. It received extensive media coverage until it was approved by the Parliament in December 2016 (Decree-Law no. 42/2016), and afterwards, especially after it was implemented in February 2017. From the start, it was clear that the soda tax “was here to stay”.

The soda tax is an excise tax on sugary drinks sold on Portuguese territory, regulated in the Código dos Impostos Especiais de Consumo. It is levied on producers and importers. The tax applies to non-alcoholic drinks with added sugar or sweeteners, including liquid or powder concentrates; drinks with 0.5–1.2% alcohol by volume are also taxed (e.g., mead, cider). Tax-exempt products include (1) milk-, soy-, or rice-based drinks, (2) fruit-, algae-, or veggie- based juice and nectar, as well as cereal- and nut-based drinks, and (3) drinks considered essential for special dietary needs.

Taxed drinks are grouped according to their sugar content in grams per liter, and different tax rates apply to each category (i.e., multi-tier tax). Initially, in 2017, there were two tiers, with the sugar threshold at 80 grams per liter and the lower (upper) tier tax rate at about 8 (16) euro cents per liter. In 2018, both tax rates were slightly raised. In 2019, the lower tier was divided in three, further differentiating drinks according to the amount of sugar they contain. The two new lower tiers had their tax rate reduced, and the most sugary drinks (> 80 grams of sugar per liter) had their tax rate aggravated. Since 2019, the tax rates are 1 euro cent per liter for drinks with less than 25 grams of sugar per liter, 6 cents for drinks with 25 grams or more and less than 50 grams of sugar per liter, 8 cents for drinks with 50 grams or more and less than 80 grams of sugar per liter, and 20 cents for drinks with 80 grams or more sugar per liter (Table 1). Different tax rates apply to concentrates in liquid or powder form. The usual 23% VAT adds up to the soda tax. For reference, in Portugal, comparably with the UK, for example, the tax rate on the most sugary drinks is fairly aligned with WHO’s recommendation that soda taxes raise retail soda prices by at least 20% (WHO, 2016).

Table 1

Portuguese SSB tax rates over the years
Sugar content	2017	2018	2019
< 25 g/l	8.22 euro cents/l	8.34 euro cents/l	1 euro cent/l
>= 25 g/l, < 50g/l			6 euro cents/l
>= 50 g/l, < 80g/l			8 euro cents/l
>= 80 g/l	16.46 euro cents/l	16.69 euro cents/l	20 euro cents/l
Notes: g/L = grams per liter. Usual 23% VAT adds to the soda tax.

We use rich administrative data from the Central Balance Sheet Harmonized Panel (CBHP), provided by Banco de Portugal, for the years 2012–2019. The data cover the entire private sector in Portugal and include information on firms’ workforce, as well as accounting data from the balance sheet and the P&L statement. This dataset results from yearly compulsory tax declarations (IES – Informação Empresarial Simplificada) that are submitted electronically to the Portuguese Ministry of Finance. The Tax Authority then sends this information to the Institute of Registration and Notary Affairs (IRN) that is in charge of sending the files to Banco de Portugal, to implement several cross-checks with other administrative sources and assure the quality of the data.3

There are 19 SSB producers/importers and 27 producers/importers of bottled water, that we identify based on firms’ main economic activity —i.e., the activity accounting for the largest share of turnover (i.e., sales of goods and services). In the Portuguese Classification of Economic Activities, Revision 3, the relevant codes are CAE 11072 – manufacture of soda and other non-alcoholic beverages, and CAE 11071 – bottling of spring and mineral water. This is the (near) universe of SSB and bottled water producers/importers in Portugal, all of which are private firms not listed in the stock market.4

The 19 SSB firms constitute our treatment group, as they produce/import drinks subject to the soda tax, i.e., drinks with added sugar or sweeteners, including liquid or powder concentrates (Section 2.1.) The 27 water bottling firms constitute our primary comparison group (for a similar approach see Alsukait et al., 2020; Etilé et al., 2021; Gonçalves and Pereira dos Santos, 2020; Taylor et al., 2019). We believe that bottled water firms are a suitable comparison group, as the soda tax is unlikely to affect bottled water consumption or water bottling firms, especially in the context of Portugal, for three main reasons. First, the water bottling and SSB industries are very similar in terms of inputs (except for sugar) and cost structures (e.g., packaging, marketing, logistics). So, they are likely to be similarly impacted by other shocks and trends (e.g., substitution of plastic for more sustainable packaging). Second, the Portuguese bottled water market is highly fragmented, consisting of many firms/brands (DGEG, 2022). In our data, in 2015, mean market share of bottled water firms was 4% (maximum at 19%), while the mean market share of SSB firms was 9% (maximum at 61%). More importantly, the largest SSB firms have very low market shares in the bottled water market, limiting strategic manipulation of prices, marketing, and other business aspects (Gonçalves and Pereira dos Santos, 2020). The third reason is that studies on the impacts of soda taxes in different jurisdictions (e.g., Philadelphia, France, Saudi Arabia) have found no substitution between soda and bottled water consumption (Alsukait et al., 2020; Capacci et al., 2019; Cawley et al., 2019a; Seiler et al., 2021). Nevertheless, in sensitivity analyses we consider alternative comparison groups (Section 4.2.).

In total, our main analyses include 46 firms. The panel is unbalanced, as 5 (2) firms enter the treatment (comparison) group between 2012 and 2019, and 3 (0) firms exit the treatment (comparison) group during that period. We confirm our results on the balanced panel as a robustness check.

Regarding the time dimension, we consider 2016 to be the first treatment year, because the tax was first publicly discussed early that year and approved by the Parliament in December (Section 2.1.). Doing so enables us to catch any anticipation effects, as firms may have adapted aspects of their business before the tax was implemented in February 2017 (Taylor et al., 2019).

Descriptive statistics are displayed in Table 2, for all outcomes considered: domestic and exported sales, profits (proxied by net income), total income revenues, and total expenses, cash, receivables (money owed to the firm for goods/services delivered but not yet paid for by customers), and liabilities, number of employees working in R&D, total number of employees, and average wage, and finally, corporate income tax payments. They are calculated separately for SSB and bottled water firms, in the pre- (2012-15) and post-tax periods (2016-19). In Table A1 in the Appendix, we present the results of balance tests, comparing the means of the outcome variables for SSB and bottled water firms in 2015 (last year of the pre-treatment period). Results indicate that the two groups of firms are very similar.

Table 2

Descriptive statistics
	Bottled water firms (comparison)					SSB firms (treatment)
	Obs.	Mean	Std. Dev.	Min.	Max.	Obs.	Mean		Std. Dev.		Min.		Max.
Pre-tax
Domestic sales	96	7,570,468	8,984,236	0	32,200,000	48	30,600,000		68,400,000		0		233,000,000
Exported sales	96	195,539	411,316	0	2,851,476	48	8,638,491		21,500,000		0		79,900,000
Net income	96	346,882	2,653,941	-5,088,466	11,400,000	48	473,910		3,539,913		-7,839,936		14,800,000
Total income	96	8,752,184	10,300,000	0	38,600,000	48	42,300,000		96,400,000		0		320,000,000
Total expenses	96	8,405,302	9,065,132	26,204	30,100,000	48	41,800,000		94,200,000		0		310,000,000
Cash	96	220,444	449,128	71	3,100,799	48	590,814		2,343,760		0		15,700,000
Receivables	96	1,646,639	1,955,405	0	9,828,467	48	7,348,327		15,900,000		0		64,300,000
Liabilities	96	12,100,000	22,100,000	8,095	123,000,000	48	38,400,000		109,000,000		86,272		412,000,000
Income tax	96	-75,090	1,347,197	-12,600,000	2,187,857	48	138,366		854,436		-2,762,697		2,971,260
Average wage	96	13,013	5,573	0	31,010	48	13,035		10,189		0		62,251
# Employees	96	55	67	0	315	48	142		325		0		1,212
# Employees in R&D	34	0.18	0.46	0	2	18	2		5		0		13
Post-tax
Domestic sales	99	8,739,701	10,700,000	0	48,200,000	54		32,100,000		76,700,000		0	260,000,000
Exported sales	99	150,200	259,005	0	968,772	54		5,271,218		10,800,000		0	39,000,000
Net income	99	1,109,984	3,393,949	-2,133,503	15,900,000	54		939,923		2,971,643		-1,092,650	13,000,000
Total income	99	10,200,000	12,400,000	0	55,900,000	54		39,100,000		90,600,000		0	317,000,000
Total expenses	99	9,052,056	10,300,000	4,912	40,000,000	54		38,100,000		88,300,000		0	309,000,000
Cash	99	560,315	1,000,692	0	4,797,750	54		469,200		1,020,181		0	5,080,177
Receivables	99	1,594,349	2,182,461	0	14,800,000	54		11,200,000		23,300,000		0	82,700,000
Liabilities	99	9,655,651	16,500,000	1,437	89,000,000	54		36,000,000		105,000,000		2,776	404,000,000
Income tax	99	211,702	859,451	-1,057,868	5,623,661	54		320,996		1,009,489		-269,367	4,515,820
Average wage	99	12,738	6,817	0	29,780	54		16,559		25,077		0	159,987
# Employees	99	53	68	0	333	54		130		330		0	1,264
# Employees in R&D	34	0.09	0.29	0	1	21		2		4		0	10
Note: All variables measured in Euros, except for numbers of employees.

[3] This dataset was used by other papers including, inter alia, Caliendo et al. (2020) and Muñoz (2023).

[4] Coca-Cola bottler for most European countries is listed on the London Stock Exchange. For the Portuguese market, they contract with a private domestic firm in Portugal that is in our data.

4.1. Event study specification

To identify the causal effects of the soda tax on the various firm-level outcomes, we estimate a series of event studies, using the following specification:

$${y}_{it}= \sum _{t=2012}^{2014}{\beta }_{t} SS{B}_{i}\times yea{r}_{t}+ \sum _{t=2016}^{2019}{\beta }_{t} SS{B}_{i}\times yea{r}_{t}+{\alpha }_{i}+ {\gamma }_{t}+ {\epsilon }_{it} \left(1\right)$$

Where ${y}_{it}$ denotes one of the outcomes considered (e.g., domestic sales), for firm $i$ in year $t$, $SS{B}_{i}$ denotes the treatment indicator, which is equal to one if the firm is a SSB producer/importer and zero if it is a bottled water firm, and $yea{r}_{t}$ are a set of year indicators. The coefficients of interest are the ${\beta }_{t}$, which give the average change in the outcome of SSB firms, compared to bottled water firms, between year $t$ and 2015, the reference year (last year of the pre-treatment period). Lastly, ${\alpha }_{i}$ and ${\gamma }_{t}$ are firm and year fixed effects, respectively, and ${\epsilon }_{it}$ is the random error term. The standard errors are clustered at the firm level (Bertrand et al., 2004).5

For comparison, we also estimate difference-in-differences (DiD) models, following a specification similar to the one above, but with only one interaction term, between the treatment indicator, $SS{B}_{i}$, and a post-treatment indicator equal to one for years 2016–2019 and zero otherwise. We favor the event study specification because it allows us to test if the outcomes of SSB and bottled water firms follow similar patterns in the pre-tax period (see below), and it provides estimations of both short- and medium-run effects (instead of an average effect over the entire post-tax period).

Our identification strategy relies on the assumption that each outcome would have followed a common trend for both SSB and bottled water firms, had the soda tax not been introduced. We can split this assumption in two. First, the parallel trends assumption states that prior to the tax implementation, each outcome follows comparable trends for both SSB and bottled water firms. The event study design allows us to formally test this assumption (Roth, 2022). Throughout the results section, we show the estimates of the ${\beta }_{t}$ coefficients from Eq. (1) for the pre-treatment period, along with the 95% and 90% confidence intervals. For all outcomes, those estimates are small and not statistically different from zero, indicating that the parallel trends assumption holds. Second, the common shocks assumption states that other events occurring during or after the soda tax was introduced affect the outcomes of both groups of firms in a similar manner. The fact that the parallel trends assumption appears to hold, combined with the great similarity between the SSB and water bottling industries (Section 3), makes the common shocks assumption reasonable. We are not aware of any relevant event occurring between 2016 and 2019, apart from the soda tax, that was likely to affect only one of the two industries.

4.2. Empirical challenges and robustness checks

To cope with the right skewness of some outcome variables, we use the inverse hyperbolic sine (IHS) transformation. This transformation is increasingly popular amongst econometricians as it allows to retain nonpositive values for the analyses, as opposed to a log transformation, for which nonpositive values are not defined (Bellemare and Wichman, 2020). The IHS transformation depends on the scale of the variables: for large numbers, the transformation is close to a natural log transformation, while for small magnitudes, it almost does not modify the variable. Following Aïhounton and Henningsen (2019), we multiply each outcome variable by a scaling factor. No matter the scaling factor, zero values remain zeros, but we can move the non-zero values “closer to” or “further away from” the zero values. For each variable, we test 9 scaling factors: ${10}^{-8}, {10}^{-6}, {10}^{-4}, {10}^{-2}, {10}^{0},{10}^{2}, {10}^{4},{10}^{6},{10}^{8}$. As advised by Aïhounton and Henningsen (2019), we use the within R² as a criterion to select the most suited scaling factor for each outcome variable. Bellemare and Wichman (2020) point out that one should not directly interpret IHS coefficient estimates as percent changes when the mean of the IHS-transformed outcome variable is below 10. Doing so could lead to over- or understatements of the effects’ magnitudes. So, in addition to the coefficients, we report the retransformed marginal effects (i.e., marginal effects on the original scale of the dependent variable) based on the recent work of Norton (2022). As a robustness check, in the Appendix we also present results when using the natural logarithm (ln) transformation, as well as a ln(y + 1) transformation.

SSB and bottled water firms are identified based on their main economic activity, i.e., the activity accounting for the largest share of turnover (Section 3). Since the SSB and bottled water industries are arguably similar, it is possible that a firm’s main activity is the manufacture/importation of SSB, but part of its turnover comes from bottled water, or vice versa. Such a firm would compromise the parallel trends assumption. We address this potential issue in three ways. First, we exclude from the analyses the only firm whose main economic activity switched from manufacture/importation of SSB to bottled water during the period of analysis (a small firm with only 5 workers). Importantly, the average share of turnover (sales of goods and services) generated by the main economic activity of the remaining 46 firms is 96.7%, and that share is below 90% for only four companies. Second, we conduct a robustness check where we drop those four firms that generate less than 90% of their turnover from their main economic activity. Lastly, we rely on the following reasoning. Since the soda tax rate is defined at the product level, while our data are at the firm level, and given that SSB firms produce more than one beverage, there is heterogeneity in treatment intensity within the treatment group. Depending on their product mix, some firms are more impacted by the soda tax than others (e.g., those producing the sweetest drinks).6 In this context, the presence of a few firms in the treatment group that generate a small share of their turnover from bottled water merely “dilutes” the treatment effects, but does not harm our identification strategy.

We also seek to validate our main findings using alternative comparison groups, namely firms that produce/import (a) fruit juice, (b) milk (c) wine, and even (d) perfumes, cosmetics and hygiene products (CAE codes 10320, 10510, 1102, 20420). These are industries that share some features with the SSB industry (e.g., carton packages, glass or plastic bottles, logistics), albeit less than bottled water does.

Lastly, since the tax only applies to soda sold in Portugal, SSB firms that sell a larger share of their products in Portugal are more impacted by the soda tax than those that export a large part of their products. Based on this reasoning, we create a treatment intensity variable by dividing pre-treatment sales (in 2015) by total sales in the same year. By construction, a SSB firm exporting 40% of its sales has a treatment intensity of 0.6, while a SSB firm exporting all its products has a treatment intensity of 0 – the same as a water bottling firm. We repeat the analyses using this treatment intensity variable instead of the binary treatment indicator. This complements our placebo test whereby we assess the impacts of the tax on exports (section 5.1).

[5] Prices within sectors of activity tend to be correlated because they face similar shocks (Cameron and Miller, 2015). Clustering standard errors by sector would, however, “overfit” the estimated residuals. This is similar to the problems faced by, inter alia, Cawley and Frisvold (2017), Harju et al. (2018), and Cotropia and Rozema (2018).

[6] Unfortunately, we cannot use this variation given that our firm-level data are anonymized. Moreover, we do not know the product mix of each firm and therefore we cannot exploit border discontinuities among tax tiers to understand and analyze reformulation.

5.1. Domestic and exported sales

Figure 1 shows the estimated impacts of the soda tax on domestic and exported sales. Results indicate that domestic sales of SSB, vis-à-vis bottled water, decreased after the soda tax was implemented, and the effect exacerbates over the years. Considering the estimated marginal effect from the DiD specification, which gives the average impact over the entire treatment period, the introduction of the soda tax caused a EUR 2.1 million decrease in domestic sales per firm per year, which represents 6.8% of mean domestic sales among SSB firms in pre-treatment years (Tables 2 and A2). This estimated decrease in sales is closer to the lower range of estimates from the literature (as reported in the meta-analysis by Andreyeva et al., 2022). Since previous studies are mostly limited to retail sales, i.e., in-home consumption (Section 1), this may suggest that out-of-home consumption is less responsive to the soda tax (either due to a more inelastic demand, or because of lower pass-through to consumer prices; Grupo de Trabalho, 2018).

We also show the results of a placebo check by measuring the impacts of the soda tax on exported sales. Since only SSB sold on Portuguese territory are subject to the tax, exported sales are not expected to respond to the soda tax, which is what we find (results from the event study in Fig. 1 and DiD results in Table A2). These precise zero results lend further credibility to the use of bottled water as the comparison group.

5.2. Firms’ financial health and workforce

We find negative impacts of the soda tax on net income (total revenues minus total expenses) of SSB firms, compared to bottled water firms, statistically significant in 2016 (p < 0.1), 2017 (p < 0.05), and 2019 (p < 0.01) (Fig. 2). Like with domestic sales, the negative impacts exacerbate over the years. However, results from the DiD specification are not statistically different from zero (Table A3). We also explore whether the event study result is driven by impacts on total revenues or total expenses, but lack of estimation precision precludes any conclusions (Roth, 2022).

The soda tax impacted negatively SSB firms’ cash account, and positively their receivables account (Fig. 3). These effects are statistically significant in 2016 and 2018 for cash, and in 2016, 2017, and 2018 for receivables (p < 0.1), but they are not statistically significant when considering all post-tax years together (DiD results in Table A4). Nevertheless, these results suggest that the tax may have hindered SSB firms’ capacity to convert receivables into cash. A possible explanation is loss of negotiation power of manufacturers vis-à-vis retailers and other clients, that could translate into longer payment delays (Grupo de Trabalho, 2018). Furthermore, the soda tax significantly increased SSB firms’ liabilities, compared to bottled water firms’, starting in 2018 (Fig. 3). The estimated marginal effect from the DiD specification indicates an average increase in liabilities of EUR 8.5 million per firm per year, which represents 22% of mean liabilities among SSB firms in pre-treatment years (Tables 2 and A4). For instance, SSB firms may have had to contract debt to face costs associated with reformulation and relabeling, as well as early departure of products from shelves due to discontinuity (Grupo de Trabalho, 2018). Overall, the soda tax appears to have harmed SSB firms’ financial health.

We do not find impacts of the soda tax on SSB firms’ workforce, namely average wages, numbers of employees, or numbers of employees working in R&D (Fig. 4, Table A5). This suggests that reformulation activities (see Section 2.2) were undertaken without restructuring the employee base towards higher R&D capacity. It may be that firms had already built capacity to develop new recipes (industry reports suggest reformulation activities have been going on since before the tax was announced), or that firms outsource this activity. For example, multinational brands may develop new recipes in other countries. Despite the suggestive evidence that the soda tax harmed firms’ profitability (captured here by net income), in Portugal firms face great hurdles to fire employees or decrease wages (Martins & Portugal, 2019), which may be one reason to explain why there are no impacts of the soda tax on employment or average wages.

5.3. Corporate income tax payments

Lastly, we find that starting in 2017, corporate income tax payments by SSB firms are significantly reduced by the soda tax (Fig. 5). However, considering the full post-tax period, the marginal effect from the DiD specification is both statistically and economically zero (Table A4). Taking the point estimate at face value, a back-of-the envelope calculation gives 210,735 euros of forgone corporate income tax over the period 2017–2019 (-4,683 euros times 45 SSB firm-year observations). Official estimates from the Portuguese Ministry of Finance claim that the soda tax generated revenue of EUR 71.4, 72.5, and 60.1 million in 2017, 2018, and 2019, respectively. Hence, we conclude that the soda tax had a large positive impact on Portuguese public finances, despite possible reductions in corporate income tax payments.

5.4. Robustness checks

Our baseline results are robust to a series of checks regarding potential outliers and our methodological decisions, namely (1) excluding extreme values of the outcome variables (1% winsorization), (2) dropping the largest firm of the dataset (a SSB producer that employs more than 1,000 employees and has turnover more than ten times larger than mean turnover), (3) dropping the four firms that generate less than 90% of their turnover from their main economic activity, (4) restricting the analyses to the balanced panel, and (5) using the ln(y) or ln(y + 1) transformation instead of the IHS transformation (Tables A6-A7 in the Appendix).7

In addition, we show that our baseline results do not depend on one specific comparison industry. Using alternative comparison groups, namely firms producing (a) fruit juice, (b) milk, (c) wine, or (d) perfumes, cosmetics and hygiene products, instead of bottled water firms, consistently produces comparable results (Table A8).8 Lastly, results using the (continuous) treatment intensity variable, instead of the binary one, are highly similar, once again supporting the validity of our identification strategy (Table A9).

[7] As can be seen in Table A7 by comparing Panels A. and B., the number of zeros in the sample is limited.

[8] Each group is considered separately. Event study graphs considering all groups provide very similar results to baseline and are available from the authors upon request.

This study contributes with evidence on the impacts of soda taxes for producers/importers and their workforce, and provides insights about total consumption effects (both in- and out-of-home) —two important gaps that were identified in the literature on the impacts of soda taxes. Using novel firm-level data on the near-totality of SSB firms in Portugal, and relying on a comparator group of bottled water firms to identify causal effects, we find a significant 6.8% decrease in domestic sales. There is also some evidence of negative impacts of the soda tax on firms’ financial health, namely a decrease in net income and an increase in liabilities. There is no evidence that employment in the soda manufacturing industry was affected. Any reformulation activity that may or may not have been induced by the introduction of the soda tax did not translate into higher numbers of employees working in R&D. The Portuguese soda tax seems to have been lucrative from a public finance standpoint, with virtually no impacts on forgone corporate income tax payments.

By looking at SSB producers/importers’ domestic sales, this study complements the existing body of literature, which mostly considers retail sales (i.e., mainly in-home soda consumption). However, a distinction between the impacts of soda taxes on in-home and out-of-home consumption is not possible here; we only provide the aggregate picture. Another limitation is that sales mix together a quantity effect and a price effect, which would also be important to separate. Lastly, it is possible that we do not include the universe of SSB producers/importers because firms are identified based on their main economic activity; in any case, any excluded firm is likely to be small.

To conclude, governments still pondering the introduction or revision of their soda taxes should consider all economic agents; not just consumers and retailers, who have received most of the attention, but also producers/importers and workers. Our results inform about some of the impacts of soda taxes on the latter.

Acknowledgements

The authors thank Banco de Portugal for access to the data and technical support. We are also indebted to Pedro Bação, Ronald Bachmann, Pedro Pita Barros, Marc F. Bellemare, Pierre Dubois, Eleonora Fichera, Pedro Freitas, Christos Kotsogiannis, Pedro Martins, Luís Catela Nunes, Susana Peralta, Miguel Portela, as well as participants at the Tax Administration Research Centre (TARC) – University of Essex Seminar, the EuHEA Seminar Series, and the 2022 Linked Employer-Employee Data Workshop for comments and suggestions. All errors are our own.

Aguilar, A., Gutierrez, E., & Seira, E. (2021). The effectiveness of sin food taxes: Evidence from Mexico. Journal of Health Economics, 77, 102455.
Aïhounton, G. B. D., & Henningsen, A. (2019). Units of Measurement and the Inverse Hyperbolic Sine Transformation. The Econometrics Journal, 24(2), 334–351.
Allcott, H., Lockwood, B. B., & Taubinsky, D. (2019a). Should we tax sugar-sweetened beverages? An overview of theory and evidence. Journal of Economic Perspectives, 33(3), 202–27.
Allcott, H., Lockwood, B. B., & Taubinsky, D. (2019b). Regressive sin taxes, with an application to the optimal soda tax. The Quarterly Journal of Economics, 134(3), 1557–1626.
Alsukait, R., Wilde, P., Bleich, S. N., Singh, G., & Folta, S. C. (2020). Evaluating Saudi Arabia’s 50% carbonated drink excise tax: Changes in prices and volume sales. Economics and Human Biology, 38, 100868.
Andreyeva, T., Marple, K., Marinello, S., Moore, T. & Powel, L. (2022). Outcomes following taxation of sugar-sweetened beverages. A systematic review and meta-analysis. JAMA Network Open 5(6). Arteaga, J. C., Flores, D., & Luna, E. (2017). The effect of a soft-drink tax in Mexico: a time series approach.
Arteaga, J. C., Flores, D., & Luna, E. (2021). The effect of a soft drink tax in Mexico: evidence from time-series industry data. Australian Journal of Agricultural and Resource Economics 65: 349–366.
Bellemare, M. F., & Wichman, C. J. (2020). Elasticities and the inverse hyperbolic sine transformation. Oxford Bulletin of Economics and Statistics, 82(1), 50–61.
Berardi, N., Sevestre, P., Tepaut, M., & Vigneron, A. (2016). The impact of a ‘soda tax’ on prices: evidence from French micro data. Applied Economics, 48(41), 3976–3994.
Bertrand, M., Duflo, E., & Mullainathan, S. (2004). How much should we trust differences-in-differences estimates?. The Quarterly Journal of Economics, 119(1), 249–275.
Bollinger, B., & Sexton, S. (2018). Local excise taxes, sticky prices, and spillovers: evidence from Berkeley's soda tax. Available at SSRN: http://dx.doi.org/10.2139/ssrn.3087966.
Capacci, S., Allais, O., Bonnet, C., & Mazzocchi, M. (2019). The impact of the French soda tax on prices and purchases. An ex post evaluation. PloS one, 14(10), e0223196.
Castelló, J. V., & Casasnovas, G. L. (2020). Impact of SBB taxes on sales. Economics & Human Biology 36: 100821.
Caliendo, L., Mion, G., Opromolla, L. D., & Rossi-Hansberg, E. (2020). Productivity and organization in Portuguese firms. Journal of Political Economy, 128(11), 4211–4257.
Cawley, J., & Frisvold, D. E. (2017). The pass-through of taxes on sugar‐sweetened beverages to retail prices: the case of Berkeley, California. Journal of Policy Analysis and Management, 36(2), 303–326.
Cawley, J., Willage, B., & Frisvold, D. (2018a). Pass-through of a tax on sugar-sweetened beverages at the Philadelphia International Airport. Journal of the American Medical Association, 319(3), 305–306.
Cawley, J., Crain, C., Frisvold, D., & Jones, D. (2018b). The pass-through of the largest tax on sugar-sweetened beverages: the case of Boulder, Colorado. National Bureau of Economic Research (No. w25050).
Cawley, J., Frisvold, D., Hill, A., & Jones, D. (2019a). The impact of the Philadelphia beverage tax on purchases and consumption by adults and children. Journal of Health Economics, 67, 102225.
Cawley, J., Thow, A. M., Wen, K., & Frisvold, D. (2019b). The economics of taxes on sugar-sweetened beverages: a review of the effects on prices, sales, cross-border shopping, and consumption. Annual review of nutrition, 39, 317–338.
Cawley, J., Frisvold, D., & Jones, D. (2020). The impact of sugar-sweetened beverage taxes on purchases: evidence from four city-level taxes in the United States. Health Economics, 29(10): 1289–1306.
Cawley, J., Daly, M., & Thornton, R. (2022). The effect of beverage taxes on youth consumption and body mass index: Evidence from Mauritius. Health Economics, 31(6): 1033–1045.
Cotropia, C., & Rozema, K. (2018). Who benefits from repealing tampon taxes? Empirical evidence from New Jersey. Journal of Empirical Legal Studies, 15(3), 620–647.
Colchero, M. A., Salgado, J. C., Unar-Munguía, M., Hernandez-Avila, M., & Rivera-Dommarco, J. A. (2015). Price elasticity of the demand for sugar sweetened beverages and soft drinks in Mexico. Economics and Human Biology, 19, 129–137.
Colchero, M. A., Popkin, B. M., Rivera, J. A., & Ng, S. W. (2016). Beverage purchases from stores in Mexico under the excise tax on sugar sweetened beverages: observational study. BMJ, 352.
Colchero, M. A., Molina, M., & Guerrero-López, C. M. (2017). After Mexico implemented a tax, purchases of sugar-sweetened beverages decreased and water increased: difference by place of residence, household composition, and income level. The Journal of nutrition, 147(8), 1552–1557.
Cornelsen L., & Smith R. (2018). Viewpoint: Soda taxes — Four questions economists need to address. Food Policy 74: 138–142.
DGEG Direção-Geral de Energia e Geologia (2022). Exploração de águas minerais naturais —Engarrafamento. https://www.dgeg.gov.pt/pt/areas-setoriais/geologia/recursos-hidrogeologicos/exploracao-de-aguas-minerais-naturais/engarrafamento/
Dickson, A., Gehrsitz, M., Kemp, J. (2021) Does a Spoonful of Sugar Levy Help the Calories Go Down? An Analysis of the UK Soft Drinks Industry Levy. IZA DP No. 14528
Dubois, P., Griffith, R., ’ O'Connell, M. (2020). How well targeted are soda taxes?. American Economic Review, 110(11), 3661–3704.
Etilé, F., Lecocq, S., & Boizot-Szantai, C. (2021). The Incidence of Soft-Drink Taxes on Consumer Prices and Welfare: Evidence from the French Soda Tax. European Review of Agricultural Economics, 48(4), 915–939.
Fearne, A., Borzino, N., De La Iglesia, B., Moffatt, P., & Robbins, M. (2021). Using supermarket loyalty card data to measure the differential impact of the UK soft drink sugar tax on buyer behaviour. Journal of Agricultural Economics, 1–17.
Fichera, E., Mora, T., Lopez-Valcarcel, B. G., & Roche, D. (2021). How do consumers respond to “sin taxes”? New evidence from a tax on sugary drinks. Social Science & Medicine, 274, 113799.
Global Food Research Program UNC (2022). Sugary drink taxes around the world. https://www.globalfoodresearchprogram.org/wp-content/uploads/2022/05/Sugary_Drink_Tax_maps_upload.pdf
Goiana-da Silva, F., Nunes, A.M., Miraldo, M., Bento, A., Breda, J., Araújo, F.F., 2018. Using pricing policies to promote public health: the sugar sweetened beverages taxation experience in Portugal. Acta Med. Port. 31.
Gonçalves, J., & Pereira Dos Santos, J. (2020). Brown sugar, how come you taste so good? The impact of a soda tax on prices and consumption. Social Science & Medicine, 264, 113332.
Grogger, J. (2017). Soda taxes and the prices of sodas and other drinks: evidence from Mexico. American Journal of Agricultural Economics, 99(2), 481–498.
Grummon, A. H., Lockwood, B. B., Taubinsky, D., & Allcott, H. (2019). Designing better sugary drink taxes. Science, 365(6457), 989–990.
Grupo de Trabalho. (2018). Impacto do imposto especial sobre o consumo de bebidas açucaradas e adicionadas de edulcorantes. https://www.portugal.gov.pt/download-ficheiros/ficheiro.aspx?v=%3D%3DBAAAAB%2BLCAAAAAAABAAzM7I0AQCQMNv8BAAAAA%3D%3D
Guerrero-López, C.M., Molina, M., Colchero, M.A. (2017). Employment changes associated with the introduction of taxes on sugar-sweetened beverages and nonessential energy-dense food in Mexico. Preventive Medicine 105, S43–S49.
Harju, J., Kosonen, T., & Skans, O. N. (2018). Firm types, price-setting strategies, and consumption-tax incidence. Journal of Public Economics, 165, 48–72.
Law, C., Cornelsen, L., Adams, J., Penney, T., Rutter, H., White, M., & Smith, R. (2020a). An analysis of the stock market reaction to the announcements of the UK Soft Drinks Industry Levy. Economics & Human Biology, 38, 100834.
Law, C., Cornelsen, L., Adams, J., Pell, D., Rutter, H., White, M., & Smith, R. (2020b). The impact of UK soft drinks industry levy on manufacturers’ domestic turnover. Economics & Human Biology, 37, 100866.
Law, C., Smith, R., & Cornelsen, L. (2022). Place matters: Out-of-home demand for food and beverages in Great Britain. Food Policy 107: 102215.
Lawman, H.G., Bleich, S.N., Yan, J., LeVasseur, M.T., Mitra, N., Roberto, C.A. (2019). Unemployment claims in Philadelphia one year after implementation of the sweetened beverage tax. PLoS One 14 (3) e0213218.
Léger, P. T., & Powell, L. M. (2021). The impact of the Oakland SSB tax on prices and volume sold: A study of intended and unintended consequences. Health Economics, 30(8): 1745–1771.
Leider, J. & Powell, L. M. (2022). Longer-term impacts of the Oakland, California, sugar-sweetened beverage tax on prices and volume sold at two-year post tax. Social Science & Medicine 292: 114537.
Lyssiotou, P., & Savva, E. (2021). Who pays taxes on basic foodstuffs? Evidence from broadening the VAT base. International Tax and Public Finance, 28, 212–247.
Martins, F., & Portugal, P. (2019) How did the downward wage rigidity shape unemployment during the crisis?. In Portuguese economic growth: A view on structural features, blockages and reforms (pp. 95–103), Banco de Portugal.
Muñoz, M. (2023) Trading non-tradables: The implications of Europe’s job posting policy. Conditionally accepted at The Quarterly Journal of Economics.
Nakamura, R., Mirelman, A. J., Cuadrado, C., Silva-Illanes, N., Dunstan, J., & Suhrcke, M. (2018). Evaluating the 2014 sugar-sweetened beverage tax in Chile: an observational study in urban areas. PLoS medicine, 15(7), e1002596.
Norton, E. (2022). The inverse hyperbolic sine transformation and retransformed marginal effects. NBER Working Paper 29998. http://www.nber.org/papers/w29998
O’Connell, M., & Smith, K. (2021). Optimal sin taxation and market power. IFS Working Paper No. W21/30. https://www.econstor.eu/handle/10419/242929.
Rojas, C., & Wang, E. Y. (2021). Do taxes for soda and sugary drinks work? Scanner data evidence from Berkeley and Washington. Economic Inquiry 59: 95–118.
Roth, J. (2022). Pretest with Caution: Event-Study Estimates after Testing for Parallel Trends. American Economic Review: Insights 4(3), 305–322.
Rozema, Kyle. (2018). "Tax incidence in a vertical supply chain: Evidence from cigarette wholesale prices." National Tax Journal 71(3) 427–450.
Seiler, S., Tuchman, A., & Yao, S. (2021). The impact of soda taxes: Pass-through, tax avoidance, and nutritional effects. Journal of Marketing Research, 58(1), 22–49.
Sharma, A., Hauck, K., Hollingsworth, B., & Siciliani, L. (2014) The effects of taxing sugar-sweetened beverages across different income groups. Health Economics, 23(9): 1159–1184.
Silver, L. D., Ng, S. W., Ryan-Ibarra, S., Taillie, L. S., Induni, M., Miles, D. R., Poti, J. M. & Popkin, B. M. (2017). Changes in prices, sales, consumer spending, and beverage consumption one year after a tax on sugar-sweetened beverages in Berkeley, California, US: A before-and-after study. PLoS medicine, 14(4), e1002283.
Stacey, N. Mudara, C., Ng, S. W., van Walbeek, C., Hofman, K., & Edoka, I. (2019). Sugar-based beverage taxes and beverage prices: evidence from South Africa’s Health Promotion Levy. Social Science & Medicine 238: 112465.
Taylor, R. L., Kaplan, S., Villas-Boas, S. B., & Jung, K. (2019). Soda wars: The effect of a soda tax election on university beverage sales. Economic Inquiry, 57(3), 1480–1496.
WHO. (2016, October 11). WHO urges global action to curtail consumption and health impacts of sugary drinks. Retrieved from World Health Organization: https://www.who.int/news/item/11-10-2016-who-urges-global-action-to-curtail-consumption-and-health-impacts-of-sugary-drinks

No competing interests reported.

Appendix.docx

Download PDF

Journal Publication

published 22 Nov, 2023

Read the published version in International Tax and Public Finance →

Editorial decision: Major revision
23 Jun, 2023
Reviews received at journal
17 May, 2023
Reviewers agreed at journal
09 May, 2023
Reviewers invited by journal
06 May, 2023
Editor assigned by journal
02 May, 2023
Submission checks completed at journal
25 Apr, 2023
First submitted to journal
22 Apr, 2023

You are reading this latest preprint version

Not so sweet: Impacts of a soda tax on producers

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. The Portuguese soda tax

3. Data and descriptive statistics

4. Empirical strategy

4.1. Event study specification

4.2. Empirical challenges and robustness checks

5. Results

5.1. Domestic and exported sales

5.2. Firms’ financial health and workforce

5.3. Corporate income tax payments

5.4. Robustness checks

6. Discussion and concluding remarks

Declarations

Acknowledgements

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1