from South Africa Indirect Taxation and Gender Equity ...iks.ukzn.ac.za/sites/default/files/Indirect Taxation and Gender Equity... · introduced (F. C. v. N. Fourie and A. Owen 1993;

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=rfec20

Download by: [UNIVERSITY OF KWAZULU-NATAL] Date: 27 June 2017, At: 02:40

Feminist Economics

ISSN: 1354-5701 (Print) 1466-4372 (Online) Journal homepage: http://www.tandfonline.com/loi/rfec20

Indirect Taxation and Gender Equity: Evidencefrom South Africa

Daniela Maria Casale

To cite this article: Daniela Maria Casale (2012) Indirect Taxation and Gender Equity: Evidencefrom South Africa, Feminist Economics, 18:3, 25-54, DOI: 10.1080/13545701.2012.716907

To link to this article: http://dx.doi.org/10.1080/13545701.2012.716907

Published online: 05 Sep 2012.

Submit your article to this journal

Article views: 325

View related articles

Citing articles: 1 View citing articles

http://www.tandfonline.com/action/journalInformation?journalCode=rfec20

http://www.tandfonline.com/loi/rfec20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/13545701.2012.716907

http://dx.doi.org/10.1080/13545701.2012.716907

http://www.tandfonline.com/action/authorSubmission?journalCode=rfec20&show=instructions

http://www.tandfonline.com/action/authorSubmission?journalCode=rfec20&show=instructions

http://www.tandfonline.com/doi/mlt/10.1080/13545701.2012.716907

http://www.tandfonline.com/doi/mlt/10.1080/13545701.2012.716907

http://www.tandfonline.com/doi/citedby/10.1080/13545701.2012.716907#tabModule

http://www.tandfonline.com/doi/citedby/10.1080/13545701.2012.716907#tabModule

IN D I R E C T TA X A T I O N A N D GE N D E R EQ U I T Y :EV ID E N C E F R O M SO U T H AF R I C A

Daniela Maria Casale

ABSTRACT

This study adds to the growing literature on the distributional effects of indirector consumption taxes in developing countries by exploring whether these taxeshave differential gender outcomes. Using data from Statistics South Africa’s2000 Income and Expenditure Survey, the study investigates differences in taxincidence between ‘‘female-type’’ and ‘‘male-type’’ households, classifiedaccording to their demographic and economic attributes. The results suggestthat zero-rating a well-targeted selection of basic foodstuffs and fuel forhousehold use is important in protecting female-type households, especiallythose in the lowest quintiles and with children, from bearing an otherwisedisproportionate share of the tax burden. In contrast, high taxes on alcohol,tobacco, and fuel for private transport result in a larger incidence on male-typehouseholds. The study also suggests ways in which the indirect tax structurecould be refined to further reduce the large gender (and income) inequitiesthat exist in South Africa.

KEYWORDS

Indirect taxes, incidence, gender equity, South Africa

JEL Codes: D63, H22, J16

INTRODUCTION

Equity in taxation is one of the three main criteria, along with efficiencyand ease of administration, by which a tax is conventionally evaluated in taxpolicy analysis. Attempts to quantify how equitable particular taxes are haveresulted in a relatively large body of literature, including a number ofstudies on developing countries, that specifically examine the incidence ofindirect taxes (Richard M. Bird and Barbara D. Miller 1989; EhtishamAhmad and Nicholas Stern 1991; Stephen D. Younger 1993; John Gibson1998; Stephen D. Younger, David E. Sahn, Steven Haggblade, and Paul A.Dorosh 1999; Harivelo Rajemison, Steven Haggblade, and Stephen D.Younger 2003; David E. Sahn and Stephen D. Younger 1998, 2003). Thisliterature explores the important question of who ultimately bears theeconomic burden of taxes on goods and services. The focus of most of these

Feminist Economics 18(3), July 2012, 25–54

Feminist Economics ISSN 1354-5701 print/ISSN 1466-4372 online � 2012 IAFFEhttp://www.tandf.co.uk/journals

http://dx.doi.org/10.1080/13545701.2012.716907

studies, however, has been on the incidence of indirect taxes by incomeclass – in other words, on how progressive or regressive the indirect taxsystem is. This study extends this work by examining equity and indirecttaxes in a developing country context from a gender perspective, using datafrom Statistics South Africa’s 2000 Income and Expenditure Survey (IES;SSA 2000a). In addition, the paper attempts to draw out the implications ofsome of the key findings in the South African case for other countries of asimilar level of economic development.1

In understanding the gendered incidence of indirect taxes, it is useful todraw on the distinction Janet Stotsky (1997a, 1997b) makes between explicitand implicit gender bias in tax systems. Explicit bias arises due to specificprovisions in the tax laws that treat women and men differently (forexample, if men and women are taxed at different marginal rates on theirpersonal income). Implicit bias occurs when provisions of the tax law have adifferential impact on women and men due to gendered social or economicbehavior, even though the tax law contains no explicit bias. While it is notcommon to find explicit bias in the laws governing indirect taxes, implicitbias may arise because men and women (and their households) spend theirincomes on different types of goods or on goods that are taxed differently.For example, if women spend a greater proportion of their budgets on foodthan men, and food is taxed at a higher rate than all other goods, then therewill be an implicit bias against women.

A formidable methodological challenge in studies of indirect taxincidence is how to identify the individual incidence of a tax (andtherefore whether an implicit bias exists against a particular group ofindividuals) when expenditure often occurs and expenditure data arealmost always collected at the household level. In the absence of individual-level data, the general approach in tax incidence studies has been either toassume equal sharing of income and expenditure in the household or toimpose another sharing rule, and then to analyze the incidence of the taxon the poor compared with the rich. In this study, I use a somewhatdifferent approach that may be considered more suitable for an analysis ofthe gendered impact of indirect taxes, which is based on the approach usedin Caren Grown and Imraan Valodia (2010). I classify households in SouthAfrica as being more ‘‘female’’ or more ‘‘male’’ according to theirdemographic and economic attributes,2 and I compare the incidence ofindirect taxes across these household types. For example, I compare the taxincidence in households that are headed by a woman or households inwhich only women are employed to the incidence in households headed bya man or households in which only men are employed. I also look at theincidence on these households across the expenditure distribution andaccording to whether or not children live in the household, as I aminterested in the tax burden on more vulnerable types of households inparticular.

ARTICLE

26

The South African case study is well suited to this methodology, as a highproportion of women in South Africa head their own households, do notlive with adult men, or do not live with men who earn income fromemployment. These demographic patterns are reflective of a number offactors that are specific to South Africa’s socioeconomic and politicallandscape during the apartheid and post-apartheid periods, including highrates of circular labor migration, low marriage rates, and very high levels ofunemployment. Just over 40 percent of South Africans live in householdsheaded by a woman, and almost 50 percent live in households that have noearnings from men in employment. Women who live without income-earning men or who head their own households are also among thepoorest in South Africa and are more likely to be living with children (mycalculations using data from the IES [SSA 2000a]).

BACKGROUND

An analysis of the equity implications of indirect taxes is important for anumber of reasons that have led in recent years to a growing literature onthe welfare outcomes of these taxes in developing countries (see forexample Bird and Miller [1989] on Jamaica; Ahmad and Stern [1991] onPakistan; Younger [1993] on Ghana; Gibson [1998] on Papua New Guinea;Harold Alderman and Carlo del Ninno [1999] on South Africa; Youngeret al. [1999] and Rajemison, Haggblade, and Younger [2003] onMadagascar; Sahn and Younger [2003] on Cote d’Ivoire, Guinea,Madagascar, and Tanzania). Indirect taxes make up a large portion ofgovernment tax revenue, particularly in developing countries, necessitatinga clearer understanding of how these taxes affect income distribution. InSouth Africa, indirect taxes make up approximately 40 percent of totalgovernment tax revenue, although even this is low by developing countrystandards where the share is generally between 50 and 60 percent (KathleenBarnett and Caren Grown 2004). In addition, there is a global pattern ofindirect taxes increasing as a share of total government revenue, as itbecomes more difficult to tax companies and individuals due to theincreased mobility of labor and capital (Barnett and Grown 2004; JoshuaAizenman and Yothin Jinjarak 2006). It is important therefore tounderstand whether and how the collection of a large share ofgovernment revenue affects men and women differently.

Furthermore, the indirect tax base is much wider compared to the directtax base. While personal income taxes (PIT) and other direct taxes affectonly a small percentage of the population in developing countries, indirecttaxes (because they are levied on consumption) will affect most individuals.This point is particularly pertinent to a gendered analysis of taxes in SouthAfrica. Unemployment rates among women have been around 50 percentsince 2000 (using a broad definition that includes the non-searching

INDIRECT TAXATION AND GENDER EQUITY

27

unemployed). Women are far more likely to be engaged in informal paidand unpaid work than men; and even where women are employed in theformal sector of the economy, their earnings are less likely to be above theincome tax threshold than men’s (Daniela Casale and Dorrit Posel 2005).Using 2005 labor force data, Debbie Budlender (2009) estimates that 73percent of employed women compared with 65 percent of employed menfell outside of the tax net, and of the PIT paid, women’s contributionsaccounted for only 30 percent. These findings suggest that the genderedeffects of tax collection in South Africa will be more far-reaching throughindirect taxes.

In South Africa, a few studies have tried to model the impact of themain component of indirect taxes, value-added tax (VAT), on welfareoutcomes, using data mostly from the early 1990s when VAT was firstintroduced (F. C. v. N. Fourie and A. Owen 1993; Alderman and delNinno 1999; Delfin S. Go, Marna Kearney, Sherman Robinson, andKaren Thierfelder 2005). A more extensive study by Ingrid Woolard,Charles Simkins, Morne Oosthuizen, and Christopher Woolard (2005)examines the incidence of VAT, excise duties, and the fuel levy usingdata from 2000 as part of a general report on total tax incidence for theSouth African National Treasury. However, none of this researchexplores the impact of indirect taxes from a gender perspective. Thisstudy begins to fill this gap by investigating whether the households inwhich women live and control resources, especially those that are in thepoorest quintiles and include children, bear a differential incidence ofindirect taxes.

A differential gender incidence of indirect taxes is unlikely to result fromexplicit bias, as there are very rarely statutory differences in indirect taxrates (tax authorities do not levy different VAT rates by group, whereas PITrates do often differ by group). However, implicit bias is likely to exist inconsumption taxes, as people have different spending patterns andtherefore will bear the burden of the tax in different proportions. Thisbias is particularly relevant for gender equity, as much research now showsthat men and women have different spending priorities when they controlresources (see John Hoddinott and Lawrence Haddad [1995] on CoteD’Ivoire; Shelly J. Lundberg, Robert A. Pollak, and Terence J. Wales [1997]on the United Kingdom; Shelley A. Phipps and Peter S. Burton [1998] onCanada; Cheryl Doss [2006] on Ghana).

An implicit bias against a particular type of gendered household would beat odds with the equity principle inherent in tax policy evaluation.However, the view taken in this paper is that tax policy at the very leastshould not disadvantage women and could be used to transform genderrelations by reducing pre-tax inequities. This seems particularly importantin a country such as South Africa that, despite being classified as a middle-income country with a relatively high gross domestic product (GDP) per

ARTICLE

28

capita (US$10,291 in 2009), has extremely high rates of poverty thatdisproportionately affect women and their households (Dorrit Posel andMichael Rogan 2009).

Of course, tax policy has a number of objectives other than ensuringequity, including that taxes should promote economic efficiency and beadministratively easy to collect in order to maximize revenue collection.Furthermore, taxes are often used to discourage negative externalities; the‘‘sin taxes’’ on alcohol and tobacco are typical examples. Therefore theresults of the incidence analysis in this paper need to be evaluated in thisbroader policy context. Although South Africa, unlike many otherdeveloping countries, has been able to maintain a relatively high tax/GDP ratio of around 28 percent in recent years, above even thegovernment’s own stated target (Budlender 2009), it is still necessary tobe mindful of the revenue implications of tax policy. Along these lines, thepolicy simulations presented in this paper also consider the fiscal impact ofthe proposed changes.

METHODOLOGY AND DATA

Tax incidence and rate structure

The methodology most commonly used for calculating the incidence ofindirect taxes involves estimating the amount of tax paid by householdsindirectly through information on their spending behavior. Mostcountries, including South Africa, conduct household surveys in whichdetailed data are collected on households’ expenditure patterns. Thepost-tax expenditure values available in these surveys are used withcorresponding tax rate and price information for the year in question tocalculate the amount of tax paid by each household on each consumptionitem.

Assuming that the tax is shifted forward entirely onto the consumer, theamount of tax paid per item can be calculated as follows where the tax is advalorem:

taxpaidVij ¼ ratej�ðexpendij=ð1þ ratejÞÞ;

where ratej is the tax rate on item j and expendij is the reported expenditurefor household i on item j. For a unit tax, the amount of tax paid by thehousehold per item is calculated as:

taxpaidSij ¼ ðexpendij=pricejÞ � dutyj ;

where dutyj is the per unit duty on item j and pricej is the retail price of thatitem.


29

Tax incidence is then usually calculated as the percentage of totalhousehold consumption expenditure spent on the tax for that item, or intotal. The convention in the literature on tax incidence in developingcountries is to use consumption expenditure rather than income as thebase, as it is a better measure of well-being if households engage inconsumption smoothing (that is, borrow or use savings to maintainconsumption in the face of transitory income changes; Grown and Valodia2010). A more practical reason for using consumption expenditure as thebase in this study is that consumption information is considered morereliable than the income information in the dataset I use (Charles Simkins2004).

To calculate the tax incidence for South Africa, I draw expenditure datafrom the 2000 IES (SSA 2000a).3 The IES, which is predominantly used toupdate the consumer price index weights, is conducted every five yearsamong a nationally representative sample of about 30,000 households. Itcontains very detailed information on the spending patterns of households,with data collected on around 500 expenditure items through face-to-faceinterviews. According to the official report released with the data,respondents were selected as follows: ‘‘The person (or persons)responding in this interview should be a member/members of thehousehold who is/are likely to do the purchases for the household orknow the answers to our questions’’ (SSA 2002a: 91).

The tax rate and price information that I used to calculate the taxincidence per item came from various government sources: Department ofFinance’s (2000) Budget Review; South African Revenue Services’ (2007)Value-Added Tax Guide for Vendors; and the 2000 Survey on Retail Prices (SSA2000b). In 2000–1, just over 42 percent of total tax revenue in South Africawas collected through indirect taxes. The main component was VAT, whichaccounted for 24.7 percent of total tax revenue, with much smaller sharesfrom excise duties (4.8 percent), the fuel levy (7.5 percent), and customsduties or trade taxes (2.9 percent; Department of Finance 2000). This studydeals only with the incidence of VAT, excise duties, and the fuel levy.

In South Africa, VAT is a single-rate tax levied on the value added at eachstage in the production process. Suppliers can claim back the VAT onintermediate goods, which means that VAT is effectively a consumption tax,as it is paid by consumers at the final stage of production (on both locallyproduced and imported goods). VAT was introduced at 10 percent in 1991to replace the general sales tax on goods and services and was subsequentlyincreased to 14 percent in 1993, where it has remained since. To reduce theburden of this tax on the poor, the government had zero-rated (that is,eliminated taxes on) nineteen basic food items by 1993, among them avariety of grains, dairy products, fruits, and vegetables.4 In addition,paraffin, goods that are subject to the fuel levy (that is, petrol and diesel),international transport services, farming inputs, sales of going concerns,

ARTICLE

30

and certain government grants were zero-rated. In particular, thegovernment zero-rated paraffin/kerosene, used predominantly by thepoor as fuel for cooking, lighting, and heating, in 2001 with the aim offurther alleviating the burden of VAT on the poor, especially poor women,who are generally responsible for these household chores (Budlender2009).5

There are also a number of goods and services that are VAT exempt:residential rental and accommodation; educational services (includingcreches); public road and rail transport; non-fee related financial services;and medical aid and medicine/medical services provided by public healthinstitutions. Unlike with zero-rated goods, suppliers of VAT-exempt goodsare not able to claim back the input VAT. This implies that, to the extentthat the inputs attract VAT themselves, some of the VAT may be passed onto the final consumer. An effective rate would be somewhere between 0 and14 percent. However, in the empirical work, these goods are rated at 0percent, given that the largest input cost in these sectors is likely to belabor.6

South Africa levies specific unit excise duties on sorghum meal, tobaccoproducts, and non-alcoholic and alcoholic beverages. It is important to notehere that the unit taxes on tobacco and alcoholic beverages are particularlyhigh with effective rates (based on 2000 retail prices) ranging from 26percent on spirits and 35 percent on cigarettes to 56 percent on cigars, forexample. The fuel levy is also a unit tax, levied at 110.1 cents per liter ofpetrol and 89.4 cents per liter of diesel in 2007. For this study, I calculatethe incidence of the fuel levy on petrol and diesel for household use andprivate transport only. I do not estimate the impact of a transfer of the fuellevy onto the consumer where fuel is an input in other productionprocesses. However, I follow Grown and Valodia (2010) and make a rule-of-thumb adjustment for the public transport sector, where I assume that thetotal amount of the fuel levy is passed on to the consumer and that fuelconstitutes 30 percent of input costs in this sector.

Identifying the gender incidence of indirect taxes

As noted earlier, the biggest methodological challenge in estimating thegender incidence of indirect taxes is how to reconcile individual andhousehold-level concepts. This challenge occurs because gender is anindividual attribute, but expenditure information is collected and oftenoccurs at the household level (especially where spending is on indivisible/public-type goods). It is common practice in the literature estimating theindirect tax incidence on individuals by race or by income, for example, toassume equal sharing of expenditure in the household, and in turn equalsharing of the burden of taxation across individuals in the household.However, this method is especially unsatisfactory for a study of the gender


31

impact of taxes, given that the intrahousehold allocation of resourcescannot be assumed to be equal.

The alternative method I use here involves classifying households as beingeither more ‘‘female’’ or more ‘‘male’’ according to certain demographicor economic characteristics, and analyzing the tax incidence on theindividuals within these households.7 I use three definitions to classifyhouseholds as being ‘‘male-type’’ or ‘‘female-type.’’ The first takes intoaccount only the presence of adult men and women in the household. Thesecond and third try to take into account gendered spending power in thehousehold by adding the dimension of control over resources, measuredthrough employment status and household headship, respectively. Table 1presents the distribution of individuals across the resulting householdtypes.

Table 1 Distribution of individuals across household categories and by quintile(percentages)

Quintile

All Q1 Q2 Q3 Q4 Q5

Presence of adultsAdult men are themajority

21.9 15.2 17.5 20.7 22.8 23.8 100

Adult women arethe majority

42.0 26.6 24.2 20.4 17.5 11.3 100

Equal number ofadults

36.1 15 16.7 19.2 21.3 27.8 100

100Employment status

Only men employed 26.4 12.2 15.4 20.8 26.7 24.9 100Only womenemployed

21.6 20.3 22.9 22.7 20.2 13.9 100

Men and womenemployed

24.2 9.8 12.5 18.6 23.1 36 100

No employedmembers

27.8 35.8 28.8 18.5 10.7 6.3 100

100Headship

Headed by a man 59.1 14.7 15.9 19.1 22.6 27.7 100Headed by a woman 40.9 27.4 26 21.5 16.3 9 100

100Average per

capita monthlyexpenditure perquintile in 2000

R66.26 R140.37 R250.16 R531.38 R2,585.30

Notes: The average South African Rand to US$ exchange rate for 2000 was just below R7. The averagefor 2009 was R8.44.

Source: Own calculations from IES (SSA 2000a).

ARTICLE

32

The first definition, which uses the presence of adult men and women(age 18 years and older) to classify households by gender, results in threecategories: ‘‘adult women are the majority’’ (where adult womenoutnumber adult men); ‘‘adult men are the majority’’ (where adult menoutnumber adult women); and households with an ‘‘equal number ofadults.’’ In South Africa, 42 percent of individuals live in households inwhich adult women are the majority, and 22 percent live in households inwhich adult men are the majority, with the remaining 36 percent living inhouseholds with an equal number of adult men and women.

The employment status definition classifies households into fourcategories: ‘‘only women employed’’ (households containing at least oneemployed adult woman and no employed adult men); ‘‘only menemployed’’ (households containing at least one employed adult man andno employed adult women); ‘‘women and men employed’’ (householdswith at least one employed adult man and one employed adult woman);and ‘‘no employed members.’’8 There are more individuals living inhouseholds with only men employed than in households with only womenemployed (26.4 versus 21.6 percent), with another 24.2 percent living inhouseholds with men and women employed. With South Africa’s (non-searching inclusive) unemployment rate at around 40 percent for the lastdecade, it is not surprising that the largest proportion of individuals inSouth Africa (27.8 percent) live in households with no employed members.For the vast majority of these households, either pensions and grants(predominantly through the government’s social welfare program) orremittances from migrant workers form the main sources of income.

The headship classification categorizes households as either headed by aman or headed by a woman. The following excerpt from the official reporton the IES provides the definition of headship used in the survey:

At Statistics SA we have a clear definition of a household head.Respondents may have a different idea of what ‘‘household head’’means, and you must explain to them what Stats SA wants. The head isthe person in whose name the dwelling is registered. It may be theperson who owns the dwelling, or is responsible for the rent, or getsthe dwelling through their work, or through their relationship to theowner. If two or more persons have equal claim to be head of thehousehold, or if people state that they are joint heads or that thehousehold has no head, then choose the eldest as the head. (SSA2002a: 90)

According to this definition, 41 percent of individuals were classified asliving in households headed by a woman in 2000.

Compared with the other countries for which a similar analysis wasperformed in Grown and Valodia (2010),9 a much higher proportion of


33

South Africa’s population lives in what I refer to as female-type households –those in which only women are employed, in which adult women are themajority, or that are headed by a woman. A number of factors drive thisphenomenon in South Africa. There are particularly low levels of marriageand partnership and high rates of extramarital childbearing in SouthAfrica. These trends partly reflect the destabilizing effect that apartheidpolicies had on the institution of marriage among black South Africans(Victoria Hosegood, Nuala McGrath, and Tom Moultrie 2009). They arealso a result of the more recent financial constraints faced by many youngSouth African men who cannot afford to get married and make bridewealthpayments (Dorrit Posel and Daniela Casale 2009). The high incidence oflabor migration in South Africa, particularly among men, is also acontributing factor and often results in a woman in the household beingreported as the de facto resident head. Historically, the apartheidsettlement laws prohibited Africans from settling permanently at theirplaces of employment, resulting in a system of circular labor migrationamong mainly African men who migrated to urban and industrial areaswhile African women were left to tend to the homestead in designated ruralareas. While these laws were abolished in the late 1980s, their effects persist,and labor migration remains an important feature of the South Africanlabor market (Dorrit Posel and Daniela Casale 2003). In 2002, for example,about 20 percent of households headed by a woman reported at least onenon-resident household member as a man doing migrant work, accordingto data from the Labour Force Survey of that year (SSA 2002b). In addition,economic growth and employment creation have been lackluster in thepost-apartheid period since 1994, with unemployment rates rising toalarming levels. As a result, even when women do live with men, these menmay not have employment.

Also evident from Table 1 is that the female-type households and thosewith no employed members tend to be concentrated in the lower quintilesof the expenditure distribution. In contrast, male-type households,households with both women and men employed, and equal-numberedadult households are more heavily concentrated at the upper end of theexpenditure distribution. This pattern implies that any tax policy that haspositive gender equity implications will also have strong income equityoutcomes.

For South Africa, there is a large overlap across the three genderedhousehold classifications, which is evident from the cross-tabulations inTable 2. For example, just over 80 percent of households headed by awoman fall into two employment status categories: 40.7 percent are in the‘‘only women employed’’ category, and 40.6 percent are in the ‘‘noemployed members’’ category. The majority of households headed by awoman (71 percent) are in the category of households in which adultwomen are the majority. About 73 percent of households in which adult

ARTICLE

34

women are the majority fall into the categories of ‘‘only women employed’’(36.6 percent) and ‘‘no employed members’’ (35.9 percent). This highcorrelation suggests therefore that the tax incidence results will tell asimilar story regardless of the gendered household definition that is used.

A key assumption in this methodology is that headship or employmentstatus sufficiently captures control over resources in female-type households.Data on decision making from the National Income Dynamics Study, arecently released household survey, provide some evidence to support thisassertion (Southern Africa Labour and Development Research Unit[SALDRU] 2008). Some 95 percent of women who reported headingtheir households also reported being the main decision maker for day-to-dayhousehold expenditures (groceries, for example), and 91 percent reportedbeing the main decision maker for large, unusual purchases (such asappliances or furniture). In addition, over 90 percent of women headsreported making decisions about who is allowed to live in the household andabout where the household should live.

These findings are reflected in statistically significant differences inspending behavior between male-type and female-type households in thedata I use for this analysis. Figure 1 presents budget shares only forhouseholds headed by a man and households headed by a woman, assimilar results are found in comparisons between ‘‘only men employed’’

Table 2 Cross-tabulations of individuals by household classification

a) Employment status by headshipHeaded by a man Headed by a woman

Only men employed 39.5 7.4Only women employed 8.4 40.7Men and women employed 33.2 11.2No employed members 18.8 40.6

100 100b) Presence of adults by headship

Headed by a man Headed by a womanAdult men are the majority 30.3 9.8Adult women are the majority 21.3 71.8Equal number of adults 48.4 18.4

100 100

c) Employment status by presence of adultsAdult men are

the majorityAdult women are

the majorityEqual number

of adultsOnly men employed 46.3 10.4 32.9Only women employed 9.1 36.6 11.9Men and women employed 22.2 17.1 33.7No employed members 22.4 35.9 21.5

100 100 100



35

Figure 1 Expenditure shares in households headed by a woman and householdsheaded by a manNotes: – Denotes subcategory. All gender differences are statistically significant at the1 percent level.Source: Own calculations from IES (SSA 2000a).

ARTICLE

36

and ‘‘only women employed’’ households as well as in comparisonsbetween households in which men are the majority and households inwhich women are the majority. I allocate the approximated 500consumption items detailed in the IES to twenty-five main categorieslargely based on the United Nations Classification of IndividualConsumption According to Purpose (COICOP), but with somemodification to highlight potential gender differences. The results revealthat households headed by a woman spend a larger share of their budgetson food (all three subcategories of basic, sugar/confectionary, and otheritems), children’s clothing and footwear, personal care items (bothessential and non-essential), fuel for household use, public transport, andeducation. In contrast, households headed by a man spend a larger share oftheir budgets on housing, meals out, alcoholic beverages, tobacco, adultclothing, private transport, fuel for private transport, medical expenses,communication, and recreation. These differences in budget shares arelargely consistent with the broader international literature on how controlover resources results in gendered spending patterns (Hoddinott andHaddad 1995; Lundberg, Pollak, and Wales 1997; Phipps and Burton 1998;Doss 2006). The differences suggest that there will be differential taxburdens by gender in South Africa.

Before presenting the tax incidence results, however, it is important toreiterate the main caveat of this analysis – that is, I am unable to estimate anindividual incidence for men and women because I do not know theintrahousehold allocation of expenditure and consumption. So while male-type households may bear a higher burden of a particular tax, women livingin those households might also bear part of the tax burden throughreduced consumption opportunities. Table 3 shows the distribution of menand women across household types. Although the majority of women liveeither in households in which only women are employed or no adult isemployed, 21 and 23 percent live in households in which only men areemployed or both men and women are employed, respectively. However, inthe same vein, 16 percent of men live in households in which only womenare employed, and 26 percent of men live in households in which no adultis employed, and insofar as these households bear a lower incidence oncertain goods, the men in these households will also benefit. This analysisenables me to identify how tax incidence differs when households consistmostly of women or when women have greater decision-making or earningpower in the household compared to when men have a greater presence inthe household or are in control of resources.

INDIRECT TAX INCIDENCE RESULTS

Table 4 reports the indirect tax incidence results for the differenthousehold types using the three gendered household classifications. Due


37

to the strong correlations across household categories reported above, theresults are generally consistent regardless of which household classificationI use. Total indirect tax incidence is lower in female-type households thanin male-type households by around a full percentage point on a base ofapproximately 8 percent. This result holds for the different types of taxes aswell – that is, VAT, excise duties, and the fuel levy. The pattern of incidenceamong households with no employed members is similar to the patternamong female-type households, while the households with both men andwomen employed and the equal-numbered adult households resemble themale-type households in their tax incidence.

While there are statistically significant gender differences for all threetypes of tax, the largest gender differentials are reported for the exciseduties and the fuel levy – taxes that can be justified on negative externalitygrounds. The higher incidence of indirect taxes in male-type households isbeing driven predominantly by the larger expenditure in these householdson alcohol and tobacco and on fuel for private transport. The genderdifference in the incidence of the fuel levy would have been even morepronounced had I not adjusted for the transfer of the fuel levy toconsumers in the public transport sector because female-type householdsare relatively more intensive users of public transport, while male-typehouseholds are relatively more intensive users of private transport.

It is possible that because female-type households and those withno employed members are among the least well off in South Africa

Table 3 Distribution of men and women across household categories

Men Women

Number(1,000s) Percentage

Number(1,000s) Percentage

Presence of adultsAdult men are the majority 6,795 32.82 2,681 11.9Adult women are the majority 6,099 29.46 12,040 53.45Equal number of adults 7,812 37.73 7,804 34.65

100 100Employment statusOnly men employed 6,734 32.52 4,672 20.74Only women employed 3,406 16.45 5,950 26.41Men and women employed 5,188 25.06 5,285 23.46No employed members 5,379 25.98 6,618 29.38

100 100HeadshipHeaded by a man 14,085 68.04 11,461 50.9Headed by a woman 6,616 31.95 11,056 49.1

100 100


ARTICLE

38

(see Table 1), the results simply reflect the nature of spending across theincome classes rather than the gender categories. However, disaggregatingthe results by expenditure quintile indicates that female-type householdsand households with no employed members bear a lower indirect taxincidence than male-type households, households with both employed menand women, and equal-numbered adult households, regardless of whichexpenditure quintile the households are in (Table 5). This result holds fortotal indirect taxes and, with few exceptions, for the different types of taxes.In particular, female-type households across the distribution bear a lowerburden of the excise taxes on alcohol and tobacco than male-typehouseholds. This finding is consistent with the idea that men have agreater preference for these goods, or that they have greater socialpermission to consume them, rather than the idea that these are luxurygoods that female-type households cannot afford. These results confirmthat even after controlling for expenditure class, there are gendereddifferences in spending patterns, underscoring the importance of an

Table 4 Overall incidence by household category (tax as a percentage ofexpenditure)

Total Tax VAT Excise Tax Fuel Tax

HeadshipHeaded by a man 9.06*** 7.17*** 0.96*** 0.94***

(0.011) (0.007) (0.006) (0.006)Headed by a woman 7.99 7.08 0.44 0.48

(0.011) (0.008) (0.005) (0.005)Employment categoriesOnly men employed 9.36*** 7.36*** 1.12*** 0.88***

(0.018) (0.011) (0.010) (0.009)Only women employed 8.14 7.05 0.45 0.64

(0.016) (0.012) (0.007) (0.007)Men and women employed 9.15*** 7.13*** 0.89*** 1.14***

(0.018) (0.012) (0.009) (0.011)No employed members 7.84*** 6.99*** 0.49*** 0.37***

(0.013) (0.010) (0.006) (0.005)Presence of adultsAdult men are the majority 9.23*** 7.29*** 1.1*** 0.84***

(0.020) (0.012) (0.012) (0.011)Adult women are the majority 8.13 7.07 0.47 0.59

(0.011) (0.008) (0.005) (0.005)Equal number of adults 8.84*** 7.12*** 0.85*** 0.88***

(0.014) (0.010) (0.007) (0.008)

Notes: Standard errors are in parentheses. ***, **, * denote statistical significance at the 1, 5, and 10percent levels, respectively. The reference category for the t-tests of differences in means is thefemale-type household for each classification. For example, the tax incidence of 9.06 percent forhouseholds headed by a man is greater than the tax incidence of 7.99 percent for householdsheaded by a woman, and this difference is significant at the 1 percent level.



39

Tab

le5

Inci

den

ceb

yh

ou

seh

old

cate

gory

and

qu

inti

le(t

axas

ap

erce

nta

geo

fex

pen

dit

ure

)

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Hea

dshi

p

Qu

inti

leH

eade

dby

am

anH

eade

dby

aw

oman

17.

81**

*(0

.024

)6.

7***

(0.0

17)

0.8*

**(0

.014

)0.

31**

*(0

.007

)6.

87(0

.017

)6.

34(0

.015

)0.

27(0

.007

)0.

26(0

.005

)2

8.76

***

(0.0

23)

7.34

***

(0.0

14)

0.98

***

(0.0

15)

0.44

***

(0.0

09)

8 (0.0

19)

7.2

(0.0

14)

0.47

(0.0

11)

0.33

(0.0

05)

39.

5***

(0.0

23)

7.7*

*(0

.014

)1.

13**

*(0

.015

)0.

67**

*(0

.011

)8.

61(0

.021

)7.

65(0

.016

)0.

47(0

.011

)0.

48(0

.010

)4

9.85

***

(0.0

25)

7.55

***

(0.0

16)

1.15

***

(0.0

14)

1.16

***

(0.0

16)

8.97

(0.0

30)

7.7

(0.0

22)

0.61

(0.0

15)

0.66

(0.0

15)

58.

97**

*(0

.026

)6.

65**

*(0

.018

)0.

76**

*(0

.012

)1.

56**

*(0

.015

)8.

17(0

.054

)6.

49(0

.037

)0.

44(0

.020

)1.

24(0

.033

)

Em

ploy

men

tca

tego

ries

Qu

inti

leO

nly

men

empl

oyed

On

lyw

omen

empl

oyed

18.

17**

*(0

.039

)6.

98**

*(0

.025

)0.

85**

*(0

.024

)0.

35(0

.012

)6.

9(0

.027

)6.

27(0

.023

)0.

29(0

.010

)0.

33(0

.009

)2

8.95

***

(0.0

38)

7.4*

**(0

.023

)1.

08**

*(0

.025

)0.

47**

*(0

.015

)8.

2(0

.030

)7.

27(0

.021

)0.

52(0

.019

)0.

41(0

.008

)3

9.64

***

(0.0

33)

7.78

***

(0.0

20)

1.24

***

(0.0

23)

0.62

(0.0

14)

8.72

(0.0

28)

7.59

(0.0

22)

0.49

(0.0

14)

0.64

(0.0

17)

49.

92**

*(0

.037

)7.

53**

*(0

.023

)1.

31**

*(0

.021

)1.

08**

*(0

.023

)8.

73(0

.038

)7.

43(0

.028

)0.

53(0

.017

)0.

77(0

.021

)5

9.36

***

(0.0

18)

6.97

***

(0.0

29)

0.99

***

(0.0

23)

1.4*

**(0

.024

)8.

08(0

.057

)6.

41(0

.041

)0.

41(0

.020

)1.

25(0

.029

)

(con

tin

ued

)

ARTICLE

40

Tab

le5

(Con

tin

ued

)

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Qu

inti

leM

enan

dw

omen

empl

oyed

No

empl

oyed

mem

bers

17.

95**

*(0

.047

)6.

73**

*(0

.030

)0.

9***

(0.0

28)

0.33

(0.0

12)

7 (0.0

20)

6.39

***

(0.0

16)

0.39

***

(0.0

09)

0.23

***

(0.0

05)

29.

24**

*(0

.041

)7.

45**

*(0

.023

)1.

23**

*(0

.027

)0.

57**

*(0

.014

)7.

82**

*(0

.020

)7.

12**

*(0

.015

)0.

44**

(0.0

11)

0.26

***

(0.0

05)

39.

5***

(0.0

35)

7.7*

**(0

.022

)1.

04**

*(0

.023

)0.

76**

*(0

.018

)8.

56**

*(0

.029

)7.

63(0

.020

)0.

59**

*(0

.016

)0.

34**

*(0

.010

)4

10.0

7***

(0.0

34)

7.78

***

(0.0

22)

1.01

***

(0.0

19)

1.27

***

(0.0

24)

8.96

**(0

.047

)7.

67**

*(0

.033

)0.

72**

*(0

.027

)0.

56**

*(0

.020

)5

8.69

***

(0.0

34)

6.4

(0.0

23)

0.61

***

(0.0

13)

1.68

***

(0.0

22)

8.72

***

(0.0

80)

6.73

***

(0.0

57)

0.56

***

(0.0

29)

1.43

***

(0.0

47)

Pre

sen

ceof

adu

lts

Qu

inti

leA

dult

men

are

the

maj

orit

yA

dult

wom

enar

eth

em

ajor

ity

17.

67**

*(0

.039

)6.

68**

*(0

.027

)0.

76**

*(0

.024

)0.

23**

*(0

.008

)7.

07(0

.018

)6.

4(0

.015

)0.

37(0

.008

)0.

3(0

.006

)2

8.8*

**(0

.040

)7.

37**

*(0

.023

)1.

01**

*(0

.028

)0.

42**

*(0

.014

)8.

1(0

.019

)7.

25(0

.014

)0.

47(0

.010

)0.

38(0

.006

)3

9.52

***

(0.0

38)

7.76

***

(0.0

23)

1.14

***

(0.0

26)

0.62

***

(0.0

16)

8.64

(0.0

22)

7.58

(0.0

16)

0.52

(0.0

12)

0.54

(0.0

10)

410

.05*

**(0

.043

)7.

58**

*(0

.025

)1.

37**

*(0

.027

)1.

1***

(0.0

28)

9.13

(0.0

29)

7.64

(0.0

21)

0.61

(0.0

13)

0.87

(0.0

17)

59.

52**

*(0

.053

)6.

92**

*(0

.033

)1.

09**

*(0

.028

)1.

5***

(0.0

31)

8.2

(0.0

43)

6.4

(0.0

32)

0.42

(0.0

14)

1.38

(0.0

25)

(con

tin

ued

)


41

Tab

le5

(Con

tin

ued

)

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Qu

inti

leE

qual

nu

mbe

rof

adu

lts

17.

49**

*(0

.029

)6.

59**

*(0

.021

)0.

62**

*(0

.016

)0.

28**

*(0

.008

)2

8.51

***

(0.0

28)

7.21

(0.0

18)

0.93

***

(0.0

18)

0.37

(0.0

08)

39.

41**

*(0

.027

)7.

75**

(0.0

17)

1.05

***

(0.0

18)

0.62

***

(0.0

14)

49.

64**

*(0

.032

)7.

57(0

.021

)1.

04**

*(0

.017

)1.

03**

*(0

.019

)5

8.75

***

(0.0

32)

6.56

***

(0.0

22)

0.63

***

(0.0

12)

1.56

***

(0.0

19)

Not

es:S

tan

dar

der

rors

are

inp

aren

thes

es.*

**,*

*,*

den

ote

stat

isti

cals

ign

ifica

nce

atth

e1,

5,an

d10

per

cen

tle

vels

,res

pec

tive

ly.T

he

refe

ren

ceca

tego

ryfo

rth

et-t

ests

of

dif

fere

nce

sin

mea

ns

isth

efe

mal

e-ty

pe

ho

use

ho

ldin

that

qu

inti

lefo

rea

chcl

assi

fica

tio

n.

So,

for

exam

ple

,ta

xin

cid

ence

inh

ou

seh

old

sh

ead

edb

ya

wo

man

inq

uin

tile

on

eis

test

edag

ain

stta

xin

cid

ence

inh

ou

seh

old

sh

ead

edb

ya

man

inq

uin

tile

on

e.

Sou

rce:

Ow

nca

lcu

lati

on

sfr

om

IES

(SSA

2000

a).

ARTICLE

42

analysis of taxes from a gender perspective that goes beyond standardincome/expenditure class comparisons.10

While most of the gender difference in tax incidence is being driven by ahigher incidence of excise duties and the fuel levy in male-type households,there is also a small but statistically significant difference in VAT incidence(7.17 percent in households headed by a man compared with 7.08 percentin households headed by a woman, for example). It is interesting to explorethe VAT incidence further, given that it is the largest component of totalindirect tax incidence and that the estimated incidence encompasses avariety of zero-ratings and exemptions. In particular, useful lessons can belearned from simulating the simple counterfactual – that is, what the VATincidence would be if no items were zero-rated or exempt from the tax(assuming no knock-on effects of price changes on consumption shares).

Figure 2 shows the simulated effect on the VAT incidence for householdsheaded by a man and households headed by a woman across thequintiles.11 Without any zero-rating or exemptions, the VAT incidencewould have been substantially higher and clearly regressive, and theincidence in households headed by a woman would have been greater than

Figure 2 VAT incidence with and without zero-rating or exemptions, by householdheadshipSource: Own calculations from IES (SSA 2000a).


43

in households headed by a man. The net effect of zero-ratings andexemptions reduces the incidence substantially across all the quintiles, butparticularly in the lowest quintile where incidence is halved (from around12 to 6 percent of expenditure). The tax is transformed from being highlyregressive to one where the burden falls mainly on the middle quintiles.Also, the net effect of zero-ratings and exemptions benefits householdsheaded by a woman relatively more, and particularly those at the lower endof the expenditure distribution.

As the budget shares presented in Figure 1 suggest, the benefit tohouseholds headed by a woman derives mainly from the zero-rating of thebasket of basic foodstuffs and paraffin (the latter falls under the ‘‘fuel forhousehold use’’ category in Figure 1), where the relative genderdifferences are largest, and to a lesser extent from the exemptions onpublic transport and education. In contrast, the exemptions offered onhousing/rentals, medical expenditure, and financial services (the latterfalls under the category ‘‘miscellaneous’’) benefit households headed by aman relatively more, although not to the same extent that the zero-ratingsof the basket of basic foodstuffs and paraffin and the exemptions on publictransport and education benefit households headed by a woman. Thesefindings imply that the zero-rating of a well-targeted selection of items –particularly key foodstuffs, which make up a larger portion of the budgetsof poor female-type households, and fuel for household use, which iscommonly used by poor women in many African countries for domesticchores – can have far-reaching effects.12

However, even after taking into account the VAT exemptions and zero-ratings outlined above, there are still implicit gender biases in the indirecttax structure by consumption item. The disaggregated tax incidence resultsfor households headed by a man and households headed by a womandescribed in Figure 3 mirror the budget share patterns for thesehouseholds in Figure 1. Although female-type households bear a lowerindirect tax incidence overall, they still bear a greater tax incidence on food(including other non-basic or non–zero-rated items as well as sugar/confectionary items); non-alcoholic beverages; utilities; children’s clothing;personal care items (both essential and non-essential); fuel for householduse (although paraffin is zero-rated, VAT on firewood, charcoal, andcandles still results in a differential burden); furniture, householdequipment, and maintenance; and education (although education isexempt, textbooks and stationery attract VAT in this category). In additionto alcoholic beverages, tobacco, and fuel for transport, households headedby a man bear a greater tax incidence on housing, meals out, adultclothing, private transport, medical expenditure (as private healthcareattracts VAT), communication, and recreation.

Finally, Table 6 presents the tax incidence results according to whetherthere are children age 17 years or younger living in the household. The

ARTICLE

44

Figure 3 Tax incidence in households headed by a woman and households headedby a man, by consumption categoryNotes: – Denotes subcategory. All gender differences are statistically significant at the1 percent level, except for domestic and household services. The incidence on basicfoodstuffs and public transport is zero as these items are exempt and zero-ratedrespectively. The incidence on domestic and household services is only 0.01 percentas the majority of spending in this category is on women employed as privatedomestic workers in the home.Source: Own calculations from IES (SSA 2000a).


45

Tab

le6

Inci

den

ceb

yh

ou

seh

old

cate

gory

and

pre

sen

ceo

fch

ild

ren

(tax

asa

per

cen

tage

of

exp

end

itu

re)

Wit

hch

ildr

enW

itho

ut

chil

dren

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Tot

alT

axV

AT

Exc

ise

Tax

Fuel

Tax

Hea

dshi

pH

ead

edb

ya

man

8.93

***

(0.0

12)

7.17

***

(0.0

08)

0.87

***

(0.0

06)

0.89

***

(0.0

07)

9.52

***

(0.0

29)

7.18

***

(0.0

18)

1.26

***

(0.0

17)

1.09

***

(0.0

15)

Hea

ded

by

aw

om

an7.

92(0

.012

)7.

06(0

.009

)0.

41(0

.005

)0.

45(0

.005

)8.

48(0

.037

)7.

17(0

.027

)0.

61(0

.019

)0.

7(0

.017

)E

mpl

oym

ent

cate

gori

esO

nly

men

emp

loye

d9.

14**

*(0

.019

)7.

34**

*(0

.012

)0.

98**

*(0

.011

)0.

82**

*(0

.010

)9.

95**

*(0

.040

)7.

39**

*(0

.024

)1.

51**

*(0

.026

)1.

06**

*(0

.021

)O

nly

wo

men

emp

loye

d8.

11(0

.017

)7.

06(0

.013

)0.

45(0

.008

)0.

6(0

.008

)8.

37(0

.046

)7.

01(0

.037

)0.

5(0

.020

)0.

86(0

.021

)M

enan

dw

om

enem

plo

yed

9.11

***

(0.0

18)

7.14

***

(0.0

12)

0.86

***

(0.0

09)

1.11

***

(0.0

11)

9.39

***

(0.0

53)

7.02

***

(0.0

36)

1.03

***

(0.0

27)

1.33

***

(0.0

31)

No

emp

loye

dm

emb

ers

7.7*

**(0

.014

)6.

97**

*(0

.011

)0.

42(0

.006

)0.

31**

*(0

.005

)*8

.58

(0.0

42)

7.09

(0.0

28)

0.85

***

(0.0

24)

0.64

***

(0.0

17)

Pre

sen

ceof

adu

lts

Ad

ult

men

are

the

maj

ori

ty8.

99**

*(0

.022

)7.

31**

*(0

.014

)0.

88**

*(0

.012

)0.

79**

*(0

.013

)9.

68**

*(0

.040

)7.

24**

*(0

.023

)1.

5***

(0.0

26)

0.93

***

(0.0

19)

Ad

ult

wo

men

are

the

maj

ori

ty8.

1(0

.012

)7.

07(0

.009

)0.

47(0

.005

)0.

56(0

.005

)8.

34(0

.038

)7.

01(0

.030

)0.

52(0

.017

)0.

82(0

.018

)E

qu

aln

um

ber

of

adu

lts

8.73

***

(0.0

15)

7.09

***

(0.0

10)

0.82

***

(0.0

08)

0.82

***

(0.0

08)

9.33

***

(0.0

37)

7.22

***

(0.0

25)

0.96

***

(0.0

20)

1.15

***

(0.0

21)

Not

es:S

tan

dar

der

rors

are

inp

aren

thes

es.*

**,*

*,*

den

ote

stat

isti

cals

ign

ifica

nce

atth

e1,

5,an

d10

per

cen

tle

vels

,res

pec

tive

ly.T

he

refe

ren

ceca

tego

ryfo

rth

et-

test

so

fd

iffe

ren

ces

inm

ean

sis

the

fem

ale-

typ

eh

ou

seh

old

for

each

clas

sifi

cati

on

.So

,fo

rex

amp

le,t

axin

cid

ence

inh

ou

seh

old

sw

ith

chil

dre

nh

ead

edb

ya

wo

man

iste

sted

agai

nst

tax

inci

den

cein

ho

use

ho

lds

wit

hch

ild

ren

hea

ded

by

am

an.

Sou

rce:

Ow

nca

lcu

lati

on

sfr

om

IES

(SSA

2000

a).

ARTICLE

46

main gendered findings hold: Regardless of the presence of children in thehousehold, female-type households bear a lower incidence than male-typehouseholds for total indirect taxes and for the different types of taxes. Totake one example, the total tax incidence among households headed by awoman that include children is 7.92 percent, while the total tax incidencefor households headed by a man that include children is 8.93 percent(similarly, there is just over a one percentage point difference betweenhouseholds headed by a woman that do not include children andhouseholds headed by a man that do not include children). In otherwords, the figures suggest that the gendered difference in tax incidence isnot simply due to households headed by a woman being more likely toinclude children. An analysis by consumption item finds that householdsheaded by a woman that include children still bear a higher burden onfood, children’s clothing, essential and non-essential personal care items,fuel for household use, and furniture, equipment, and householdmaintenance items; and households headed by a man that includechildren still bear a higher incidence of taxes on alcoholic beverages,tobacco, fuel for transport, housing, meals out, adult clothing, privatetransport, communication, and recreation.

Nonetheless, it is also interesting to note the differences withinhousehold categories in the total indirect tax burden according towhether households contain children or not (although these differencesare somewhat smaller than those summarized above by gender). The taxincidence of 7.92 percent among households headed by a woman thatinclude children is lower than the 8.48 percent among households headedby a woman that do not include children (similarly, there is about a half apercentage point difference among households headed by a manaccording to the presence of children). Households with children,whether headed by a man or a woman, bear a higher incidence onhousing, food, children’s clothing, personal care items (especiallynecessities and diapers), fuel for household use, education, furniture,equipment, and household maintenance items when compared tohouseholds without children, whether headed by a man or a woman. Thehouseholds without children bear a higher incidence on meals out, non-alcoholic and alcoholic beverages, tobacco, non-essential personal careitems, adult clothing, transport, fuel for transport, medical expenditure,communication, and recreation. This finding reflects that living withchildren affects the way both men and women prioritize their spendingwhen they control resources.

SIMULATIONS AND POLICY SUGGESTIONS

This study also considers further changes to the indirect tax structure. Inparticular, I estimate the distributional and revenue consequences of zero-


47

rating additional items that would benefit female-type households,including those in the poorer quintiles and those with children. Iperform these simulations for the following goods: (1) all other (non-confectionary) food items that are not currently zero-rated; (2) poultry inparticular; (3) children’s clothing and footwear;13 and (4) a basket of basicpersonal care items (toilet paper, toothpaste, toothbrushes, soap, tissues,contraception, and sanitary napkins). I chose the goods on the basis that(a) they are recurring expenditure items, (b) they do not impose anyobvious negative externalities, and (c) they make up a larger relative shareof the budget of female-type households (and those in the lower quintilesor with children) compared with male-type households.14 By definition, thislast criterion results in strong gender and income distributional outcomesfor all of the policy experiments; therefore, I am interested in which policychanges have the largest relative effect without resulting in undue revenuelosses. For comparison, I estimate the simulated effect of VAT rating itemsthat are currently zero-rated, such as basic food items and paraffin.

I present the results of the policy simulations in Table 7. The table showsthe percentage change to the base incidence for that household categoryfollowing the policy change (meaning either zero-rating the additional itemsor levying 14 percent VAT on goods that are currently zero-rated), as well asthe relative gains or losses. The findings, summarized by the ratios in the lastrow of each frame, suggest that the largest gender and income equity gainshave already been exhausted through the government’s current zero-ratingof basic food items and paraffin, which benefits female-type households andthose in the lowest quintiles the most in relative terms. Of the four policyexperiments that zero-rate further items, the zero-rating of children’sclothing and footwear provides the next largest gains in terms of bothgender and income equity. While similar gains are evident in the zero-ratingof basic personal care items and poultry, zero-rating children’s clothingwould seem to be the most attractive policy recommendation, especiallybecause of its almost perfect targeting of households with children.

Although the revenue loss to the fiscus from this policy change of R576million per annum in 2000 prices (see final row of Table 7) is notinsubstantial (it is more than double the loss incurred through the zero-rating of paraffin, for example), it amounts to a relatively small percentageof the total VAT intake (only 1.2 percent, using the budget estimate for2000). By comparison, the reduction in revenue from the zero-rating of allother non-confectionary foodstuffs would amount to a loss of over 10percent of the total VAT intake. Put differently, the loss of revenue fromthe zero-rating of children’s clothing is the equivalent of about 1 percent ofthe education budget for that year and only about 0.5 percent of the totalsocial services budget (Department of Finance 2000).

In the same manner as reducing taxes on ‘‘necessity’’ goods that areconsumed relatively more intensively by female-type households, one could

ARTICLE

48

Tab

le7

Eff

ect

on

tax

inci

den

cean

dgo

vern

men

tre

ven

ue

of

VA

T/

zero

-rat

ing

cert

ain

item

s

Bas

ein

cide

nce

(per

cen

tage

of

exp

end

itu

re)

Eff

ect

ofV

AT

-rat

ing

(per

cen

tage

chan

gein

tax

inci

den

ce)

Eff

ect

ofze

ro-r

atin

g(p

erce

nta

gech

ange

inta

xin

cid

ence

)

Bas

icfo

od

Par

affi

n

Oth

ern

on

-co

nfe

ctio

nar

yfo

od

item

sC

hil

dre

n’s

clo

thin

g

Bas

icp

erso

nal

care

item

sP

ou

ltry

Hea

ded

by

am

an9.

0624

.50

2.21

-20.

20-3

.09

-3.2

0-4

.30

Hea

ded

by

aw

om

an7.

9939

.67

3.50

-25.

41-4

.76

-4.5

1-6

.13

Rat

ioo

fth

ep

erce

nta

gech

ange

bet

wee

nh

ou

seh

old

sh

ead

edb

ya

man

and

ho

use

ho

lds

hea

ded

by

aw

om

an

1.62

1.59

1.26

1.54

1.41

1.42

Qu

inti

le1

7.28

60.0

35.

22-2

6.37

-5.9

1-5

.77

-7.2

9Q

uin

tile

28.

3641

.27

4.07

-25.

96-5

.26

-4.7

8-6

.76

Qu

inti

le3

9.11

29.0

92.

74-2

4.15

-3.9

5-3

.95

-5.6

5Q

uin

tile

49.

5618

.83

1.36

-21.

44-2

.72

-2.9

3-4

.07

Qu

inti

le5

8.82

8.39

0.23

-14.

17-1

.59

-1.5

9-1

.81

Rat

ioo

fth

ep

erce

nta

gech

ange

bet

wee

nQ

1–3

and

Q4–

54.

797.

572.

153.

513.

213.

35

Wit

hch

ild

ren

8.49

32.5

12.

71-2

3.20

-4.4

8-3

.89

-5.3

0W

ith

ou

tch

ild

ren

9.23

20.4

82.

06-1

8.53

-0.8

7-3

.25

-4.0

1R

atio

of

the

per

cen

tage

chan

geb

etw

een

ho

use

ho

lds

wit

han

dw

ith

ou

tch

ild

ren

1.59

1.32

1.25

5.16

1.20

1.32

Lo

sso

rga

into

the

fisc

us

per

year

(in

mil

lio

ns

of

Ran

ds;

2000

pri

ces)

3,87

622

9-4

,788

-576

-618

-761

Sou

rce:

Ow

nca

lcu

lati

on

sfr

om

IES

(SSA

2000

a).


49

estimate the effect of raising taxes on luxury or demerit items whereincidence falls more heavily on male-type households to finance the policychanges suggested above. However, one needs to be very cautious here,given that I am unable to estimate the gender incidence of indirect taxeswithin the household or measure any behavioral change following a taxpolicy adjustment. A tax increase based on static household incidenceresults may have unintentional negative effects. For example, raising theexcise duties on alcohol and tobacco, which are already very high, couldresult in negative gender effects within the household if income is takenaway from necessities to finance the inelastic demand for these items.

CONCLUDING REMARKS

This study suggests that in South Africa total indirect tax incidence is loweramong female-type households compared with male-type households, aresult that holds when controlling for both expenditure quintile and thepresence of children in the household. The findings imply that the indirecttax structure in South Africa is redistributive in a gender-equitable manner.Understanding what drives this result suggests important policy lessons forother countries at a similar level of development, especially those in whichgender outcomes in tax incidence are not as positive. In particular, thezero-rating of a well-targeted basket of basic food items and paraffin, whichare consumed relatively more by poor female-type households, has helpedto protect these households from carrying a very high and otherwisedisproportionate share of the indirect tax burden, as the simulations in thispaper have shown. The high taxes on alcohol and tobacco and the fuel levy,often introduced to reduce the negative externality effects of these goods,at the same time contribute to the heavier tax burden on male-typehouseholds.

Some implicit bias against households in which women control resourcesis visible only when the results are disaggregated into differentconsumption categories: female-type households bear a higher taxburden on certain goods such as other non–zero-rated foodstuffs andfuels for household use, basic personal care items, and children’s clothing.The policy experiments in this paper suggest that the zero-rating ofchildren’s clothing in particular may be a feasible recommendation as itproduces large gender and income distributional benefits, it almostperfectly targets households with children, and it has relatively smallrevenue implications compared with the other policy simulations.

However, any change to the indirect tax structure that benefits female-type households needs to be evaluated against the trade-off of introducingfurther horizontal inequity (and complexity) into the indirect tax system.Debates in the tax literature suggest that a broad-based tax that introducesthe fewest distortions possible is most desirable from both an efficiency and

ARTICLE

50

administrative simplicity point of view (Eng-Hin Poh 2003; Reuven Avi-Yonah and Yoram Margalioth 2007). Instead, policies to reduce unequaloutcomes for women and children may be better directed from theexpenditure side of budget. While some parity is achieved through a largebudget for social services and a relatively extensive system of social welfaregrants in South Africa, many households are unable to access these, andmarked gender (and income) inequities persist. These factors suggest thatthere is room on the revenue side of the budget to effect change, and at thevery least, they demonstrate a need for constant evaluation of tax policy onboth income and gender equity grounds.

Daniela Maria CasaleSchool of Development Studies, University of KwaZulu-Natal

Durban 4041, South Africae-mail: [email protected]

ACKNOWLEDGMENTS

The author thanks Caren Grown, Sue Himmelweit, Janet Stotsky, StephenYounger, Diane Elson, Debbie Budlender, and Imraan Valodia for theircomments and input over the course of the Taxation and Gender Equityproject, as well as the members of the country teams who participated inmany fruitful discussions. Thanks also to Morne Oosthuizen and IngridWoolard, who shared their price and excise duty data, and to theanonymous reviewers of this paper who provided many valuable suggestions.The financial support of Economic Research Southern Africa is acknowl-edged.

NOTES1 This study was part of a multicountry project on ‘‘Taxation and Gender Equity’’

funded by the IDRC, Ford Foundation, and UNDP, coordinated by AmericanUniversity and the University of KwaZulu-Natal. The main objective of the project wasto investigate gender bias in tax systems in countries of varying levels of development.The project focused on the gendered impact of personal income taxes (PIT) and themain indirect taxes. The findings from the PIT analysis for South Africa are availablein Debbie Budlender (2009). A longer version of this paper is available in DanielaCasale (2009). The results of the overall project are synthesized by Caren Grown andImraan Valodia (2010).

2 I do not use the terms ‘‘female’’ and ‘‘male’’ here to imply any form of essentialism,but simply as shorthand to describe the presence of women or men in thesehouseholds.

3 I use a revised set of population weights based on the 2001 Census provided by SSA.4 The zero-rated food items are: brown bread, dried mealies, mealie rice, brown bread

flour, samp, eggs, fruit, vegetables, dried beans, lentils, maize meal, rice, pilchards intins or cans, vegetable cooking oil, milk, cultured milk, milk powder and dairy powderblend, edible legumes, and pulses of leguminous plants (peas, beans, and peanuts).


51

5 Paraffin was zero-rated only in April of 2001. Although the data are from October2000, I calculate tax incidence as if the zero-rating had applied in 2000 – that is, usingthe information on households’ spending behavior on paraffin from 2000. I do this toobtain a more realistic picture of the current incidence, especially among the poor.However, this assumption ignores any knock-on effects that an effective reduction inthe price of paraffin would have on other spending patterns.

6 An alternative method would be to estimate the likely effective VAT rate by using adetailed input–output table for South Africa. However, this would lead to a loss ofdetail (and precision), as the IES has more detailed expenditure categories than theinput–output table for South Africa, and the categories do not correspond well witheach other.

7 Another approach would be to adopt different sharing rules between men andwomen for different classes of goods. However, in the absence of sufficient casestudies (or other) research on the intrahousehold allocation of resources in SouthAfrica that could inform the choice of sharing rule, this exercise proves to be largelyarbitrary and highly subjective. In addition, for cross-country comparison, which wasone of the aims of the broader project that this study formed a part of, an analysis atthe household level is more feasible. National expenditure surveys in most countriescollect data on at least the age and gender composition of households and theirmembers’ employment status.

8 The 2000 IES based employment status on the following question: ‘‘During the pastseven days, did [name] do any work for pay, profit or family gain? Formal/informalwork, working on a farm, casual/seasonal work, etc.’’ The italicized phrase is the promptread out to respondents to provide an idea of what counts as work.

9 These countries are Argentina, Ghana, India, Mexico, Morocco, Uganda, and the UK.10 For all household categories, total indirect tax incidence tends to fall most heavily on the

middle quintiles, particularly quintiles three and four, with the poorest quintile paying asmaller share of expenditure on tax than the richest quintile. For VAT and excise duties,the incidence is predominantly on the middle quintiles, while the fuel levy is stronglyprogressive. These results are largely consistent with those in Woolard et al. (2005),which calculates the incidence of indirect taxes by decile using the same data.

11 For most of the remaining figures and tables, only the results for the householdheadship category are presented, due to space constraints and for ease of exposition.The full set of results is available in Casale (2009) and shows that, with few minorexceptions, the analysis remains unchanged when the presence of adults andemployment status definitions are used. I chose to present headship here because thecategory of headship by women encompasses two important types of households inwhich women are reported as being head: those in which the only employed arewomen, and those that have no employed members.

12 Indeed, in some of the other countries studied in Grown and Valodia (2010), wherethe tax on food is levied at a reduced rate only or where zero-ratings apply to a lesswell-targeted selection of products, the gendered results were not as positive as in theSouth African case.

13 Zero-rating all non-confectionary food items and children’s clothing is an interestingpolicy experiment also because these items are currently zero-rated in the UK.

14 Baby food (milk and grain only) and other fuels for household use (particularly coal,firewood, and candles) were also possible candidates, but I did not consider themfurther here because, for the former, there is some concern about the implications forbreastfeeding, while for the latter, there are possible environmental consequences(Casale 2009). I also excluded white bread, white sugar, and tea because of thenutritional implications, although they do form a larger relative share of the budgetsof (poor) female-type households compared with male-type households.

ARTICLE

52

REFERENCES

Ahmad, Ehtisham and Nicholas Stern. 1991. The Theory and Practice of Tax Reform inDeveloping Countries. New York: Cambridge University Press.

Aizenman, Joshua and Yothin Jinjarak. 2006. ‘‘Globalization and Developing Countries –A Shrinking Tax Base?’’ NBER Working Paper 11933, National Bureau of EconomicResearch.

Alderman, Harold and Carlo del Ninno. 1999. ‘‘Poverty Issues for Zero Rating Value-Added Tax (VAT) in South Africa.’’ World Bank Informal Discussion Paper 19336.

Avi-Yonah, Reuven and Yoram Margalioth. 2007. ‘‘Tax and Development: SomeRecent Support and Challenges to the Conventional View.’’ Virginia Tax Review27(1): 1–21.

Barnett, Kathleen and Caren Grown. 2004. ‘‘Gender Impacts of Government RevenueCollection: The Case of Taxation.’’ Economic Paper 62, Commonwealth Secretariat.

Bird, Richard M. and Barbara D. Miller. 1989. ‘‘The Incidence of Indirect Taxes on Low-Income Households in Jamaica.’’ Economic Development and Cultural Change 37(2): 393–409.

Budlender, Debbie. 2009. ‘‘Gender and Personal Income Taxes in South Africa.’’ SouthAfrican Country Paper, American University and the University of KwaZulu-Natal.http://www.sds.ukzn.ac.za/files/South%20Africa_PIT_Budlender.pdf (accessed July2012).

Casale, Daniela. 2009. ‘‘Indirect Taxation and Gender Equity: Evidence from SouthAfrica.’’ South African Country Paper, American University and the University ofKwaZulu-Natal. http://www.sds.ukzn.ac.za/files/South%20Africa_IT_2009.02.01_1.pdf (accessed July 2012).

Casale, Daniela and Dorrit Posel. 2005. ‘‘Women and the Economy: How Far Have WeCome?’’ Agenda 64: 21–9.

Department of Finance. 2000. Budget Review: 2000. Pretoria: Department of Finance,Republic of South Africa.

Doss, Cheryl. 2006. ‘‘The Effects of Intra-Household Property Ownership onExpenditure Patterns in Ghana.’’ Journal of African Economies 15(1): 149–80.

Fourie, F. C. v. N. and A. Owen. 1993. ‘‘Value-Added Tax and Regressivity in SouthAfrica.’’ South African Journal of Economics 61(4): 281–300.

Gibson, John. 1998. ‘‘Indirect Tax Reform and the Poor in Papua New Guinea.’’ PacificEconomic Bulletin 13(2): 29–39.

Go, Delfin S., Marna Kearney, Sherman Robinson, and Karen Thierfelder. 2005. ‘‘AnAnalysis of South Africa’s Value Added Tax.’’ Policy Research Working Paper WPS3671, World Bank.

Grown, Caren and Imraan Valodia, eds. 2010. Taxation and Gender Equity. London:Routledge.

Hoddinott, John and Lawrence Haddad. 1995. ‘‘Does Female Income Share InfluenceHousehold Expenditures? Evidence from Cote D’Ivoire.’’ Oxford Bulletin of Economicsand Statistics 57(1): 77–96.

Hosegood, Victoria, Nuala McGrath, and Tom Moultrie. 2009. ‘‘Dispensing WithMarriage: Marital and Partnership Trends in Rural KwaZulu-Natal, South Africa 2000–2006.’’ Demographic Research 20(13): 279–312.

Lundberg, Shelly J., Robert A. Pollak, and Terence J. Wales. 1997. ‘‘Do Husbands andWives Pool Their Resources? Evidence from the United Kingdom Child Benefit.’’Journal of Human Resources 32(3): 463–80.

Phipps, Shelley A. and Peter S. Burton. 1998. ‘‘What’s Mine Is Yours? The Influence ofMale and Female Incomes on Patterns of Household Expenditure.’’ Economica65(260): 599–613.


53

http://www.sds.ukzn.ac.za/files/South%20Africa_PIT_Budlender.pdf

http://www.sds.ukzn.ac.za/files/South%20Africa_IT_2009.02.01_1.pdf

http://www.sds.ukzn.ac.za/files/South%20Africa_IT_2009.02.01_1.pdf

Poh, Eng-Hin. 2003. ‘‘Broad-Based Consumption Tax Reform: The Economicsand Politics of the Equity Implications.’’ The International Tax Journal 29(1): 41–65.

Posel, Dorrit and Daniela Casale. 2003. ‘‘What has been Happening to InternalLabour Migration in South Africa, 1993–1999?’’ South African Journal of Economics71(3): 455–79.

———. 2009. ‘‘Sex Ratios and Racial Differences in Marriage Rates in South Africa.’’Working Paper 153, Economic Research Southern Africa (ERSA).

Posel, Dorrit and Michael Rogan. 2009. ‘‘Women, Income and Poverty: Gendered Accessto Resources in Post-Apartheid South Africa.’’ Agenda 81: 25–34.

Rajemison, Harivelo, Steven Haggblade, and Stephen D. Younger. 2003. ‘‘Indirect TaxIncidence in Madagascar: Updated Estimates Using the Input–Output Table.’’ CFNPPWorking Paper 147, Cornell Food and Nutrition Policy Program.

Sahn, David E. and Stephen D. Younger. 1998. ‘‘Fiscal Incidence in Africa:Microeconomic Evidence.’’ CFNPP Working Paper 147, Cornell Food and NutritionPolicy Program.

———. 2003. ‘‘Estimating the Incidence of Indirect Taxes in Developing Countries,’’ inFrancois Bourguignon and Luiz Pereira da Silva, eds. The Impact of Economic Policies onPoverty and Income Distribution: Evaluation Techniques and Tools, pp. 27–40. Washington,DC and New York: World Bank and Oxford University Press.

Simkins, Charles. 2004. ‘‘What Happened to the Distribution of Income in South AfricaBetween 1995 and 2001?’’ University of the Witwatersrand. http://www.sarpn.org/documents/d0001062/P1175-simkins_Nov2004.pdf (accessed July 2012).

South African Revenue Services. 2007. Value-Added Tax Guide for Vendors – VAT 404.Pretoria: South African Revenue Services.

Southern Africa Labour and Development Research Unit (SALDRU). 2008. NationalIncome Dynamics Study 2008, Wave 1. Version 4.1. Cape Town: Southern Africa Labourand Development Research Unit.

Statistics South Africa (SSA). 2000a. Income and Expenditure of Households, 2000. Pretoria:Statistics South Africa.

———. 2000b. Survey on Retail Prices, 2000. Pretoria: Statistics South Africa.———. 2002a. Income and Expenditure of Households, 2000. Statistical Release P0111.

Pretoria: Statistics South Africa.———. 2002b. Labour Force Survey, September 2002. Pretoria: Statistics South Africa.Stotsky, Janet. 1997a. ‘‘Gender Bias in Tax Systems.’’ Tax Notes International June 9,

1913–23.———. 1997b. ‘‘How Tax Systems Treat Men and Women Differently.’’ Finance and

Development March, 30–3.Woolard, Ingrid, Charles Simkins, Morne Oosthuizen, and Christopher Woolard. 2005.

‘‘Tax Incidence Analysis.’’ Final report for the Fiscal Incidence Study BeingConducted for National Treasury, South Africa.

Younger, Stephen D. 1993. ‘‘Estimating Tax Incidence in Ghana: An Exercise UsingHousehold Data.’’ Working Paper 48, Cornell Food and Nutrition Policy Program.

Younger, Stephen D., David E. Sahn, Steven Haggblade, and Paul A. Dorosh. 1999. ‘‘TaxIncidence in Madagascar: An Analysis Using Household Data.’’ World Bank EconomicReview 13(2): 303–31.

ARTICLE

54

http://www.sarpn.org/documents/d0001062/P1175-simkins_Nov2004.pdf

http://www.sarpn.org/documents/d0001062/P1175-simkins_Nov2004.pdf

from South Africa Indirect Taxation and Gender Equity ...iks.ukzn.ac.za/sites/default/files/Indirect Taxation and Gender Equity... · introduced (F. C. v. N. Fourie and A. Owen 1993;

Documents