GENDER AND INDIRECT TAX INCIDENCE IN GHANA

Transcription

1 GENDER AND INDIRECT TAX INCIDENCE IN GHANA Isaac Osei-Akoto, Robert Darko Osei and Ernest Aryeetey Institute of Statistical, Social and Economic Research (ISSER) University of Ghana Abstract Broad-based consumption taxation is increasingly becoming a key strategy to finance development programmes in Ghana. Despite this importance, attention drawn to gender biases in taxation and how governments can use tax policies to reduce inequalities in welfare by minimising existing gender differences (both explicit and implicit) in tax liabilities has been minimal. The paper seeks to analyze how tax policies and tax reforms impact differentially on women and men, and in particular on poor women. The results show that on average households in Ghana use about 7.3% of total expenses on indirect taxes, and the degree of progressiveness or otherwise differ across localities and across different gender types of households. On equity grounds, the paper suggests that there should be policy considerations to lessen the burden of poor households that have more children. The analysis on policy options points to the fact that there are varying gender and poverty effects from variations in various tax rates. Examples from the analysis, which have been considered feasible in this paper, on the grounds of reductions in economic burden of taxation and to a lesser extent on revenue grounds is complete removal of taxes on children clothes and footwear, and 50 percent reduction in the tax for kerosene. Possible sources of funding within the indirect tax system include doubling of tax rates for tobacco and alcoholic beverages and the continuous use of the newly introduced communication service tax. January 2009

2 Table of Contents 1. Introduction Description of Ghana s gendered structure Household composition Employment and income profile Trends in Tax Revenue and Reforms in Indirect Taxes Trends in Tax Revenue Political Economy of Indirect Taxes in Ghana, Incidence analysis of indirect taxes Source of data and key definitions Details of indirect taxes Findings: - incidence of indirect taxes Conclusions References Appendix A: Notes on expenditure groups (GLSS 2005/06 classification) Appendix B: Appendix Tables ii

3 1. Introduction Over the years the efforts to improve the efficiency of the tax system and to raise additional revenue, have led many developing country governments to attempt to broaden their tax base. However, concerns have been raised on the regressive nature of broad based consumption taxes (Musgrave et. al., 1974; Pechman, 1985; Brashares et. al., 1988). Other schools of thought argue that in developing countries such broad based taxes are rather progressive (Poterba 1993; Medcalf, 1995; Ebrill et. al., 2001) and that if tax exemptions and differential tax rates determine lower effective tax rates, redistributive policy measures will impact positively on the poor (Jenkins et. al., 2006). The debate in developing countries is inconclusive with respect to these issues (Bird, 2008). Indeed, little attention is drawn to gender biases in taxation and how governments can use tax policies to reduce poverty for different socio-economic groups by minimising existing gender differences (both explicit and implicit) in tax liabilities. This paper attempts to answer two questions on these issues: who bears the burden of indirect taxes in Ghana, and what can be learnt about gender dimensions of tax burden in Ghana? More specifically, the paper seeks to analyze how tax policies and tax reforms are impacting differentially on women and men, and in particular on poor women.the paper goes beyond the more traditional gendered household typology, which uses headship of households to understand differences in male and female welfare. The study introduces new dimensions 1 that combine employed and potential employable members characteristics to group households for gender analysis on tax incidence. 2. Description of Ghana s gendered structure 2.1 Household composition The recent Ghana Living Standards survey conducted in 2005/06 shows that the average household size in Ghana is about 6, comprising 3 children and 3 adults (Table 2.1). Male headed households contain more people than female headed households, and the male headed group of households constitutes nearly 77 percent (76.7%) of all households in the country (Table 2.2). Other classifications show that households with only male earners also contain more people than households with only female earners (5.8 persons versus 4.6 respectively), and there are more of such households in Ghana (30.5%) than households, which have only female adult earners (20.5%). With regards to the classification by number of adults in households, adult male-majority households are rather fewer (19.8%) than adult female-majority households (36.4%). The adult female-majority households also contain more people than adult male-majority households but the latter type of households is predominantly adults with significant number of single member households (Table 2.1). 1 Just as other five papers do for a comparative analysis for countries of different economic development. 1

4 Table 2.1: Household composition by household type Children Adults All Adult Population (%) Household type Mean (Numbers) Share of Population Single HHs Presence of adult earners: Male earners Female earners Dual earners No earners Adult sex composition: Male-majority Female-majority Equal number Headship: Male Female All Source: Calculations are based on data from GLSS 5, 2005/06 Table 2.2: Distribution of adults and children by household type Household type Children Adults All Percentage in total Presence of adult earners: Male earners Female earners Dual earners No earners Adult sex composition: Male-majority Female-majority Equal number Headship: Male Female Source: Calculations are based on data from GLSS 5, 2005/06 Sex ratio of the active population in Ghana shows that there are more female adults than male adults with the exception of few age groups, which are mainly in their fifties. This implies that if employment opportunities were fairly distributed one should observe more job placements for adult females across various employable groups. 2

5 Figure 2.1 Sex Ratio for Persons Aged years by Age, 2005/ Males per 100 females (18-64 years), 2005/ Source: Calculations are based on data from GLSS 5, 2005/ Employment and income profile Despite the disproportionate number of females in the active population group in Ghana less than a tenth (8.6%) of women (as compared to over a quarter of men, 26.9%) is in regular paid employment in their main job. The incidence of unpaid family labour amongst females is also high and accounted for almost 30% of the employment of females in their main job in 2005/06. This, in addition to over-concentration of females in low paying economic sectors, clearly account for the marked inequalities in income profiles of male and female types of households in Ghana. Females are mostly found in the informal sector, where they engage in trade and other service activities. The 2005/06 datasets also show that workers in the informal sector are usually not protected by labour laws and conventions enjoyed by association with trade union membership, and have no basic rights such as pension benefits or health care. Distribution of household expenditures The distribution of the Ghanaian population in the middle to upper income groups does not show significant variations between male and female headed households. However it is observed from Figure 2.2 that the population in the female headed households are densely concentrated in the lower income groups 2. The above observation is consistently held even if different methods of characterising households are used. This suggests, quite clearly, that poverty rates in Ghana differ significantly among male and female types of households at the lower tails of the income distribution. The implication for income redistribution is that the burden of taxes which fall more on lower income groups will disproportionately affect disposable incomes of female types of households (households with more adult females, female earner households as well as female headed households) 3, all things been equal. On the other hand, policy interventions, which positively 2 3 Expenditure is used as a proxy for permanent income in this paper. Detail definitions are provided for these types of households in section 4. 3

6 0 Density Density Density address taxes on consumption baskets for lower income groups will disproportionately benefit female types of households. Figure 2.2 Per Capita Household Expenditure by Type of Household Sex of HH Head Per Capita HH expenditure(c'0,000) GLSS5: Accra 1999 Prices Female Headed Male Headed Adult Sex Composition Per Capita HH expenditure (c'0,000) GLSS5: Accra 1999 Prices Male-majority Equal Number Female-majority Presence of Adult Earner Per Capita HH expenditure (c'0,000) GLSS5: Accra 1999 Prices Male earner Dual eaner Female earner Non Employed 4

7 % of GDP 3. Trends in Tax Revenue and Reforms in Indirect Taxes 3.1 Trends in Tax Revenue Government revenue in Ghana has seen a significant increase from the 1980s level. As at 1982 total government revenue as a percent of GDP was about 6%, but by 1985 it had reached over 10% of GDP. Most of this increase in revenue is due to indirect taxes. Generally taxes constitute the bulk of the government revenue averaging over 80% of the total government revenue 4. By 2007 government tax revenue was about 23% of GDP compared to the mid-1980s to mid-1990s average of about 13%. Over the period , tax revenue was dominated by trade taxes, which reached its peak of about 5% of GDP in However, by 1988 the proportion of trade taxes in GDP dropped and became equal to that of indirect taxes at less than 5% of GDP (about 3.8% of GDP). From about 1990, indirect taxes have dominated tax revenue, reaching its peak in 2004, at a little over 10% of GDP. Figure 3.1: Tax revenue in Ghana, % 20.00% 15.00% 10.00% 5.00% 0.00% Tax revenue Direct Taxes Indirect Taxes Trade Taxes Other Taxes Of the indirect taxes, petroleum and sales taxes contributes the most. In recent times sales taxes have been the most important component of the indirect taxes and by implication, total taxes. However this is not the case for the entire period. From about 1991 to the mid-1990s, petroleum taxes were the most important source of indirect taxes. However this changed from about the late 1990s to date. It is not difficult to relate the change in the importance of indirect taxes to the change from the sales tax to the VAT. We discuss in the next section the political economy of taxation in Ghana. 4 Government revenue here is without grants. 5

8 % of GDP Figure 3.2: Trends in Indirect Taxes, % 10.00% 8.00% 6.00% 4.00% 2.00% 0.00% Total Sales Tax Petroleum Excise 3.2 Political Economy of Indirect Taxes in Ghana, Osei and Quartey (2005) discuss tax reforms in Ghana by distinguishing three stages namely; the restoration of the tax base, the provision of tax incentives and enhancing efficiency of the tax administration system. We discuss the political economy of indirect taxes, largely following the stages discussed in Osei and Quartey (2005). Restoring the Tax Base, This period was dominated by institutional reforms. These reforms created the National Revenue Secretariat (NRS), the Internal Revenue Service (IRS) and the Customs, Excise and Preventive Services (CEPS), all with an expanded degree of independence from the Ministry of Finance. These reforms set the stage for long-term developments in the tax system, as the PNDC, which was incapacitated by the lack of public revenue when it came to power, dramatically expanded tax collection. The adoption of the Economic Recovery Programme (ERP) with its associated depreciation of the exchange rate resulted in a significant expansion of both imports and exports. This, in turn, led to an increase in excise taxes and sales taxes over this period. It also provided the appropriate platform for an expansion in tax revenue in the period that followed. Consolidation and Improving Efficiency, This period coincided with the replacement of the military regime that was associated with government suppression of the opposition and with electoral competition at the end of Over this period tax policies became relatively free of political conflict. Fiscal trouble began for the government in as it reduced rising end-of-service benefits to public officials at just the same time that the push for elections presented labour unions with new political leverage and new political urgency. Thus by 1992 Ghana was beset with major labour unrest, which 6

9 centred on demands for increased wages and the reinstatement of end-of-service benefits. The exacerbated fiscal crises faced by the government around this period set the stage for a series of political conflicts over taxation. Over this period, the first major conflict over tax was in 1993 when the government sought to increase petroleum taxes which were by far the most administratively simple means to expand revenue, and which had been successfully increased in The first increase took effect in January 1993, a few days before the inauguration of the new President and constitution, and thus it avoided parliamentary scrutiny. The budget that announced this increase (which became known as the killer budget ) was met with a public outcry. Facing a severe fiscal crisis, the government presented the second increase to parliament for ratification a few months after coming into office. However parliament rejected it even though it was dominated by the NDC (who were in power). Not only was the tax rejected, but the level of petroleum taxes subsequently declined progressively from this period until the NDC was voted out of office in This accounts in part, for the decrease in the petroleum tax revenue over this period. The second major tax event of parliament (which was dominated by NDC) came with an attempt to introduce a VAT to replace the existing sales tax in The introduction of the new tax, with a rate of 17.5%, quickly resulted in massive public protests, which are widely remembered as arguably the most dramatic public protests of the era. Both within government and outside of it, the massive protests, which eventually resulted in the repeal of the tax, are regarded as having been largely unexpected. Plans for the reintroduction of the VAT were drawn up in In order to avoid another failure two major technical changes were made before the reintroduction of the VAT. First, the rate was lowered to 10%, which was below the 15% of the existing sales tax. Second, large sums were spent in conducting public education about the new tax, to avoid problems with pricing and implementation. And, as was hoped, the new VAT was implemented with dramatically reduced public opposition. With the successful introduction of the VAT, the government returned to parliament the following year with proposals to increase the VAT by 2.5%. The government announced that the new funds would be earmarked for a Ghana Education Trust (GET) Fund, to be used primarily for educational infrastructure and scholarships, with a focus on tertiary education. Improving Efficiency and Widening Tax net, The NPP government came into power in 2001, promising greater political openness, including the immediate repeal of the Criminal Libel Law, and therefore enjoyed enormous popular goodwill. The early years of this new regime were overwhelmingly characterized by efforts to restore fiscal balance in the face of what the government claimed was near fiscal ruin brought about by the policies of the previous government. The NPP quickly set about expanding public revenue generation, with three immediate measures standing out: Accession to the Highly Indebted Poor 7

10 Countries Initiative (HIPC), a sharp increase in petroleum prices and petroleum taxes, and the introduction of the National Reconstruction Levy (NRL). Closely linked to the effort to align petroleum prices, were significant increases in petroleum taxes in 2001, which met relatively little resistance. The government was forced to increase petroleum prices again in 2005, having worn out some of the goodwill that it enjoyed in 2001, it was forced to reduce petroleum taxes as a concession to public opposition. The National Reconstruction Levy (NRL) was a surcharge tax to be paid by financial institutions, and again aroused surprisingly little resistance. While an initial promise that the tax would only last for three years helped in this regard, the reality is that even as the tax was extended until 2007 pressure for its removal mounted but was never overwhelming. An additional major revenue measures emerged later in the NPP period. The first was the National Health Insurance Levy (NHIL). For all practical purposes the NHIL was a 2.5% increase in the VAT rate, which was to be used to fund a new National Health Insurance Scheme (NHIS) to replace the existing out-of-pocket payment system for medical care. A Communications Service Tax (Talk Tax) was introduced in 2008 to raise additional revenue to finance social development programmes 5. The law charges 6% of all expenses on the use of communication and related services. To reduce the burden of taxation on consumers, other laws have been enacted to abolish the imposition of some import taxes on telephone sets, which include mobile, cellular and satellite phones in the country. 4. Incidence analysis of indirect taxes 4.1 Source of data and key definitions The analysis on incidence of indirect taxes relates three components of taxes (VAT, excise taxes and fuel levies) to total household expenditures. We used expenditure data from the Ghana Statistical Service (GSS), information on tax rates from the Ministry of Finance and Economic Planning and obtained other details on taxes from three taxation agencies in Ghana; Valued- Added Tax Service (VAT), Internal Revenue Service (IRS) and the Customs, Excise and Preventive Service (CEPS) Expenditure data As explained above, the household expenditure data sets used for this study are derived from the fifth round of the Ghana Living Standards Survey (GLSS 5). The GLSS is a multi-purpose survey of households collected on a countrywide basis, which has information on many different dimensions of living conditions. Five rounds of data have been collected, starting in but the indirect tax incidence is based on the last version conducted during the period of September 2005 to August An example is the use of at least 20% of the revenue generated from the tax to finance the National Youth Employment Programme (NYEP) introduced two years ago. 8

11 4.1.2 Measuring total household consumption expenditure The aggregated household expenditures analysed in this paper are entirely different household expenditures analysed by other studies, which discuss income and expenditures profiles in Ghana. To conform to definitions used for cross-country comparisons this study re-constructed total household expenditures by removing imputed data from the raw datasets provided by the Ghana Statistical Service. Due to the large size of Ghana s informal economy, the resultant welfare measure derived from this analysis could be quite different from the patterns derived by the GSS analysis, which is based on almost all items (both actual and imputed) in the studied households consumption basket. Among other things, the GLSS surveys collect sufficient information to estimate total consumption of selected households (GSS, 2007). This covers consumption of both food and non-food items (including housing). Documentation on the GLSS 5 explains that the expenditure measure includes imputations for consumption from sources other than market purchases. These imputations include consumption from the output of own production (mostly agriculture, but also from non-farm enterprises), and imputed rent from owner-occupied dwellings. This paper omitted home production, maintenance and remittances, gifts (including cash gifts), donations, direct taxes, investments, pension contributions, savings, repayments, dowries, net losses of selfemployment. Gifts received from other households were however included in the computation of the household total expenditure. An imputation of implicit rent derived by owner-occupied houses were also excluded from the calculations but an imputation was made for consumption of services derived from durable consumer goods owned by the household, rather than including expenditure on the acquisition of such goods (these are lumpy expenditures, e.g. purchasing a car, more like investment rather than consumption). Consumption flows (use values) for these durable goods were estimated based on assumed depreciation rates, applied by GSS (i.e. 25% of replacement value). Total consumption expenditure is estimated for a twelve-month period based on information collected with two different modules of the questionnaire; frequently purchased items and less frequently purchased items. In the case of frequent purchases (e.g. food purchases, and other non-food items such as soap and tobacco) this is estimated by grossing up responses relating to a shorter recall period. Households received multiple visits at regular intervals of a few days in the course of the survey (eleven visits at 3-day interval). In each case, all but the first two visits, they were asked about their purchases of each item since the last visit, and the answers to these bounded recall questions (recall relative to a fixed reference point) was used as the basis for estimating annual expenditure or consumption. Similar principles were used to estimate annual expenditure on frequently purchased non-food items. A longer recall period, generally twelve months, was used in collecting information on less frequently purchased consumption items (e.g. clothing and footwear); this again is grossed up as necessary. 9

12 Allowing for cost of living variations To get an index, which reflects comparable standard of living across all the households a cost of living index constructed to capture the different dimensions of variation has been added to the datasets. Two main sources of variation were controlled in the survey are: (i) (ii) Differences in the cost of living between different localities at a point in time; and Variations in prices within the time periods covered by the surveys 6, which can occur due to inflation, seasonality and other reasons. We have used this time and spatial index to deflate the estimate of total household consumption expenditure, so all the calculations reported in this paper (unless otherwise sated) are expressed in constant prices of a reference locality and time period (i.e. Accra, January 1999 constant prices). Reclassification of expenditure items The analysis is based on tax/expenditure ratio for household quintiles ordered by per capita household expenditure. The GLSS 5 follows the United Nations Statistical Commission s classification approach (Classification of Individual Consumption by Purpose - COICOP), to organise the expenditure items into groups. The categories have been re-arranged to allow for general comparison of findings on tax incidence in Ghana to findings from other countries, covered in this comparative analysis. Eighteen main categories are re-defined for this purpose (see Appendix A for the broad classifications used by GLSS 2005/06 survey). Reclassified Expenditure Groups Food o Basic Food Unprocessed o Basic Food Processed o Sugar/Confectionary o Other food items Meals Out Non-Alcoholic Beverages Alcoholic Beverages o Alcoholic-Beer o Alcoholic-Spirits o Alcoholic-Wine Tobacco Clothing and Footwear o Clothes and footwear- Adults o Clothes and footwear- Children Housing, Water & Electricity o Housing utilities o Housing general Fuel for House Use House Furnishing, Equipment & Routine Maintenance Domestic Services & Household Services 6 As explained above the survey lasted for 12 months between September 2005 and August

13 Medical Care Transport o Transport- collective o Transport- private Fuel for Transport Communication Recreation & Culture Education Personal Care Miscellaneous Definition of variables Household For the GLSS data sets, a household is defined to consist of a person or group of related or unrelated persons, who live together in the same housing unit. This group should acknowledge one adult male or female as the head of the household, they should share the same housekeeping and cooking arrangements, and are considered as one unit in a dwelling. Head of household The head of a household is the person acknowledged as such by members of the household and who is usually responsible for the upkeep and maintenance of the household. The head of household is identified by the household members themselves. He or she is the person who is named in reply to the question Who is the head of this household? Most often, but not always, it will be the person who is the main provider and who is familiar with all the activities and occupations of household members. The head of household can be male or female. Dwelling The dwelling is the structure or group of structures (rooms or buildings), separate or contiguous, occupied by the members of the household. It can be a single-family house/hut, a flat/apartment (self-contained), rooms (in a compound house), several huts/buildings (same compound), or several huts/buildings (different compounds). Household types The incidence analysis is done for different groups of households as explained below. Each of the categories has been further disaggregated into households which contain children and compared to those which do not have children: Household headship o Male headed households o Female headed households 11

14 Presence of adult earners (gender employment classification) o Male earner: -Households with at least one employed 7 adult (18 years and over) male and no employed females o Female earner: - Households with at least one employed adult (18 years and over) female and no employed males o Dual earner: - Households with at least one employed adult (18 years and over) male and one employed adult females o Non-employed: - Households with no employed adults Gender dominance (household adult sex composition): o Male-majority households (more adult males than adult females) o Female-majority households (more adult females than adult males) o Equal number (equal number of adult males and females) We have further disaggregated the main incidence analysis into locality of residence (i.e. urban/rural) and also into income levels of households ranked by per capita household expenditure quintiles. 4.2 Details of indirect taxes Three main types of taxes on households have been considered for the analysis of indirect tax incidence. These are the value added tax, excise taxes and fuel taxes Value added tax Ghana started the implementation of Value Added Tax (VAT) to replace sales tax in 1998 after an initial attempt was abandoned in 1995 because of inadequate investment in institutional capacity and other implementation difficulties (Addison and Osei, 2001). As it is in many countries, VAT in Ghana is consumption based which uses invoice credit method and collected on a destination basis. The starting tax rate in 1998 was 10% but reforms in June 1999 raised it to 12.5% with the additional 2.5% saved in a separate account (Ghana Education Trust Fund or GETFund) for the use of capital investment in the educational sector. In 2004, an additional 2.5% was introduced to fund the National Health Insurance Scheme which was introduced in the previous year. Recently, an attempt to widen the tax base and also improve the efficiency of tax administration led to the introduction of a new system (presumptive tax) which collects taxes using a flat rate scheme. This tax scheme was introduced in 2007 mainly for small retailers and the marginal rate for the tax is 3%. The rates used for VAT in the computation of tax burden in this paper exclude the adjustment made in 2007 because the expenditure data sourced for this study relate only to the survey year of 2005/2006. Different set of goods and services under VAT are taxed differently; some are taxed at almost full rates while other items avoid taxation at the retail level. Items that attract zero tax rates under 7 Employed person in this case refers to an individual doing work which brings the household money. That is, this employment category excludes subsistence work and unpaid family work. 12

15 VAT are export of taxable goods and services, goods shipped as store on vessels and aircrafts leaving the country, locally produced textbooks and exercise books and locally manufactured agricultural machinery and implements or tools. As the classification shows the list of items in this group attract no payments from individuals. Certain goods and services are exempted from the payment of VAT. This is different from the zero-rated VAT because they are exempted only at the final stage in the value chain. These items include live animals and edible animal products like meat and offal, goods for the disabled, educational items and services, medical supplies and financial services like the provision of insurance, issues and transfers. Others are water which excludes the distilled and packaged, electricity supply up to a specified consumption level prescribed in regulations by the Minister responsible for energy, printed matter such as books and newspapers and postal services. It is worthy of note that all items exempted from VAT are also exempted from the National Health Insurance Levy (NHIL). As a consequence of the exemption, the purchase prices of these items are expected to have lower proportion of VAT content than goods and services which are subject to VAT at all stages of their production and distribution chains. Estimation issues on exemptions, and the effects of tax compliance, and tax administration on effective VAT rates are well documented (Jenkins et. al., 2006; Tuan Minh Le, 2007) but we could not isolate the latter in the incidence analysis because of the nature of expenditure data the team inherited. This is likely to affect taxable items poor households purchase. Until 2007 small retail outlets were not registrable under VAT 8. Since the poor tend to purchase a larger proportion of goods and services from the retail shops and other unregistered sale outlets, the goods they buy are either not taxed at all, or are more lightly taxed. The expenditure data from the GLSS series do not indicate outlets from which consumer items are purchased. This leads the team to no other option than to suggest that the consumption of the same goods and services by different income groups are taxed at the same effective rate of VAT. Since the application of the VAT system uses the destination principle, imports are taxed in the same way as domestically produced goods. We therefore applied full VAT rate to food items which are labelled as imports in the data and also to domestic goods which have significant competing imports in the domestic market. Other zero-rated and exempt items attracted zero VAT rate in the analysis Excise and fuel taxes There is an excise tax system for certain commodities, including tobacco products, beer, table waters, malt drinks and spirits. Since 2007, calculation of excise duty rates has been based on the units or quantity and branding. Other criteria include factors such as percentage of alcoholic content, volumetric measure in a litre, numeric measure by number of sticks for cigarettes (tobacco products). However, at the time of the survey in 2005/06, excise taxes were mainly 8 The law classified a registrable retailer of goods in 2005/06 to have a turnover exceeding: cedi 100,000,000 over a 12-month period; or cedi 75,000,000 over a 9-month period; or cedi 50,000,000 over a 6-month period; or cedi 25,000,000 over a 3-month period, whichever is achieved the earliest. 13

16 collected on specific ad-valorem basis, as explained below. All alcoholic beverages and tobacco products are also subjected to VAT rate of 12.5% and NHIL rate of 2.5%. Item Rate Beer 50% Spirits and wine 25% Akpeteshie (local distilled gin) 20% Mineral or Aerated water 20% Malt 5% Tobacco 140% Even though petroleum products are VAT exempt, Ghana levies a number of taxes on petroleum products in addition to 15% ad valorem excise duties on ex-refinery current prices. These specific excise duties were reintroduced in 2006 and added to a number of earmarked levies. The fuel taxes are collected from Oil Marketing Companies, which pass on the tax to the final consumer. Expressed as a proportion of ex-pump prices, the total indirect levies involved translate to about 9.1% and 18.4% respectively for excise tax and other fuel levies for diesel. The corresponding rates for other products are presented in Table 4.1. Table 4.1: Pricing formula for petroleum products and implied fuel taxes, January 2006 Variable name Description Premium Kerosene Gasoil Premix RFO LPG Ex-refinery full-cost recovery price (cedis/liter) Ex-refinery current price (cedis per liter) % ad-valorem excise tax Specific charge Debt recovery levy (up to amount indicated) Road fund Energy fund Exploration levy Stock fund Social Impact Mitigation levy Cross subsidy levy Prim. Distr. margin Set Jan Boost margin Set Jan Ex-depot prices (cedis per liter) 13. Sum of UPPF margin Set Jan Dealers' margin Marketer's margin Filling plant Costs (LPG) Distribution compensation margin Ex-pump current maximum price (cedis per liter) Sum of tax1 (Excise) excise/ex-pump tax2 (Fuel levies) (3-10)/ex-pump tax3 (Total fuel tax) (2-10)/ex-pump Source: Calculations are based on National Petroleum Authority data for January 2006 Transportation by bus, train, boat and air 9 are VAT exempt. However, we provide analysis of tax incidence for the use of public transport because of the levels of fuel levies that are passed on 9 US$50/trip are levied as airport taxes on international travelers and a little over the equivalent of US$1/trip are also levied on domestic travelers to improve investments in the aviation sector in Ghana. Twenty-two expenses 14

17 from the commercial transport sector to households. A study by ISSER for the Ministry of Energy on social impact of petroleum deregulation showed that fuel costs constitute 62.6% of operating costs for public transport firms in Ghana (ISSER, 2004). We applied this proportion to the rates for excise duties and other levies on fuel for households which reported expenses on the use of public transport services during the survey period. This estimation leads to an approximate excise tax rate of 5.1% and other fuel levies at a rate of 16.5% if the average of diesel and premium tax rates is applied to this component Estimation of indirect tax incidence Total indirect taxes paid are indirectly estimated from the GLSS 5 household expenditure data. As noted in the comments on tax compliance, we are not able to determine whether a particular scheduled tax was honored or evaded because we do not know from where it was purchased. We used reported expenses, applied respective scheduled or estimated tax rates as described in section and estimated the tax paid on each expenditure item as follows: 10 where rate j is the j th tax rate for VAT, estimated excise tax rate or the estimated tax rate for fuel levies. taxpaid i rate i *(expend i /(1 rate expend i is the reported expenditure on expenditure item i. j j )) Tax incidence for the i th household (HH) on the k th commodity group ( ) is defined as the ratio of the per capita yearly tax outlay of the i th household on the k th commodity group ( T capita yearly total expenditure of the i th household ( E ). i.e. i k i k i ) and per Where, k i T k i /( Ei ) k T i T i and k k T i = j T, j k. ij We aggregated the taxes on individual expenditure items using the agreed categories described in section 4.2.1, cleaned up 32 outliers and calculated the ratio of tax to total household consumption for each category. The three types of taxes (VAT, Excise duties and fuels levies) were analysed separately in addition to the analysis of all indirect taxes put together. 10 were made on air transport in the expenditure data but we did not use them in the incidence analysis due to the difficulty to identify which ones are related to domestic travels and which ones belong to the other type of travel. The three tax rates are applied to the cost of production before each tax is applied. 15

18 0 Density Density Findings: - incidence of indirect taxes Distribution of total indirect taxes in different types of households The distribution of total indirect taxes by type of household suggests that relatively more of female types of households pay lower indirect taxes than male types of households. The lower parts of the indirect tax distribution in Figure 4.1, show that the proportion of male types of households (male headed households or households that have more male adults than females or households that have only male adult who earn income) are relatively smaller than their female counterparts. Since Ghana does not have differential indirect tax rates for women and does not also have different exemptions for women specific expenditure items, it could be inferred that the social dimensions of indirect tax incidence really depend on the mix of consumer items families choose. The subsequent analyses further try to isolate commodities, which potentially lead to such differences in the burden of indirect taxes in the latter section of the paper. Figure 4.1: Distribution Total Indirect Taxes by Type of Household Sex of HH Head Per Capita Total Indirect Tax (c'0,000) GLSS5: Accra 1999 Prices Female Headed Male Headed Adult Sex Composition Per Capita Total Indirect Tax (c'0,000) GLSS5: Accra 1999 Prices Male-majority Equal Number Female-majority 16

19 0 Density Presence of Adult Earner Per Capita Total Indirect Tax (c'0,000) GLSS5: Accra 1999 Prices Male earner Dual eaner Female earner Non Employed Distribution of households, which report expenditure items by locality of residence and gender of household head A total of 8,637 households - 5,048 rural and 3,589 urban - out of the surveyed number of 8,687 spent on at least one of the items listed in Appendix A. For a number of the items, urban households reported higher proportion of expenses except on alcohol, tobacco, fuel for house use and medical care expenses. For example, the proportion of households reporting tobacco expenses in the rural areas was 13.0% compared to 3.2% of households in urban areas (Table 4.2). This clearly has implications for the incidence of indirect taxes for rural households vis-avis the urban households. Table 4.2: Distribution of items reported by households by locality and sex of household head Locality of Residence Sex of Household Head Consumption Category Urban Rural Female Male Food Meals Out Non-Alcoholic Beverages Alcoholic Beverages Tobacco Clothing and Footwear Housing, Water & Electricity Fuel for House Use House Furnishing, Equipment & Routine Maintenance Domestic Services & Household Services Medical Care Transport Fuel for Transport Communication Recreation & Culture Education Personal Care Miscellation Goods & Services Sample observations (HHs) 3,589 5,048 2,412 6,225 8,637 Source: Calculations are based on data from GLSS 5, 2005/06 Total 17

20 The proportions of households reporting on expenditures on clothing and footwear; housing, water and electricity; and house furnishing, equipment and routine maintenance are equally high for both male and female headed households 11. However, Table 4.2 illustrates that, more households headed by males reported expenditures on tobacco (11.7%) than female headed households (1.6%) as well as on alcoholic beverages; 42.9% and 18.7% respectively Incidence of indirect taxes by types of households and locality of residence The results show that on average households in Ghana use about 7.34% of total expenses on indirect taxes. Incidence of VAT is higher (5.60%) as compared to excise duties (0.90%) and taxes on fuel (0.86%). The overall incidence is higher for male types of households as compared to female types of households (Table 4.3). The story is similar for all the components of taxes in the table, stressing the fact that in general Ghanaian households with over concentration of males (as main income earners, as heads of families or those with more adult males than adult females) bear higher burden of taxes. Table 4.3: Overall incidence by household types Tax as a percentage of tax expenditure (Standard Errors in brackets; not expressed as percentage) Total Tax VAT Excise Tax Fuel Tax Number of Households Headship Male headed 7.57 a 5.66 a 0.98 a 0.94 a 6224 [0.0010] [0.0007] [0.0003] [0.0006] Female headed 6.68 b 5.40 b 0.66 b 0.62 b 2412 [0.0009] [0.0007] [0.0002] [0.0002] Employment Categories Male breadwinner 7.85 a 5.80 a 1.10 a 0.95 a 2902 [0.0015] [0.0010] [0.0004] [0.0007] Female breadwinner 6.79 c 5.45 b 0.70 b 0.65 b 1935 [0.0009] [0.0008] [0.0003] [0.0002] Dual earner 7.37 b 5.57 b 0.85 c 0.97 a 2257 [0.0014] [0.0010] [0.0004] [0.0010] None employed 6.95 c 5.42 b 0.91 c 0.64 b 1542 [0.0013] [0.0009] [0.0006] [0.0004] Household Sex Composition Male-majority 7.81 a 5.83 a 1.09 a 0.90 a 2242 [0.0013] [0.0009] [0.0004] [0.0006] Female-majority 6.88 c 5.45 b 0.72 c 0.72 b 3033 [0.0011] [0.0010] [0.0002] [0.0003] Equal # females & males 7.47 b 5.59 b 0.94 b 0.95 a 3361 [0.0013] [0.0008] [0.0003] [0.0009] Note: All groups [0.0008] [0.0006] [0.0002] [0.0004] Tax incidence rates for different categories within groups are statistically 12 compared column-wise. Cells with different letters are statistically different at 5% significance level Patterns revealed by re-classification using other typologies of households do not differ much from what is reported in Table 4.2, so the respective tables for those patterns are not reported. The test procedure is a Wald test of equality of two ratios obtained by using post-estimation statistics for survey data. The standard errors are equivalent to those derived from the use of variance estimators, which apply the delta method for estimating variance of population ratios (StataCorp. 2007). 18

21 With respect to locality differences, tax burden in rural areas, in general tend to be higher than the tax burden in urban areas (Table 4.4). This locality difference is not surprising because the definition of total expenditures does not take account of the subsistence nature of the rural economy in Ghana. There are interesting differences for the burden of different taxes for male and female types of households. For male types of households, incidence rates for VAT and excise tax are lower in urban areas than in rural areas but incidence rates for fuel tax are higher in urban areas. This pattern is reversed for female types of households in the two different types of localities. Table 4.4: Overall incidence by household types and locality of residence Tax as a percentage of tax expenditure (Standard Errors in brackets; not expressed as percentage) Urban Rural Total Tax VAT Excise Tax Fuel Tax # of HHs Total Tax VAT Excise Tax Fuel Tax # of HHs Headship Male headed 7.38 a 5.54 a 0.87 a 0.98 a a 5.83 a 1.13 a 0.88 a 3810 [0.0015] [0.0010] [0.0003] [0.0009] [0.0014] [0.0010] [0.0004] [0.0006] Female headed 6.69 b 5.42 a 0.67 b 0.61 b b 5.37 b 0.65 b 0.65 b 1237 [0.0012] [0.0010] [0.0002] [0.0003] [0.0011] [0.0010] [0.0003] [0.0003] Employment Categories Male breadwinner 7.73 a 5.82 a 0.99 a 0.92 a a 5.78 a 1.22 a 0.97 a 1819 [0.0020] [0.0015] [0.0005] [0.0009] [0.0022] [0.0013] [0.0006] [0.0011] Female breadwinner 6.76 b 5.44 b 0.68 b 0.64 b b 5.47 b 0.72 c 0.65 b 978 [0.0013] [0.0011] [0.0003] [0.0003] [0.0014] [0.0011] [0.0005] [0.0003] Dual earner 7.16 b 5.34 b 0.77 b 1.06 a a 5.96 a 0.99 b 0.81 a 1222 [0.0020] [0.0014] [0.0004] [0.0015] [0.0019] [0.0015] [0.0006] [0.0006] None employed 6.95 b 5.55 ab 0.82 ab 0.59 b b 5.27 b 1.00 b 0.69 b 1028 [0.0019] [0.0014] [0.0008] [0.0005] [0.0017] [0.0012] [0.0008] [0.0005] Household Sex Composition Male-majority 7.76 a 5.78 a 1.01 a 0.97 a a 5.91 a 1.23 a 0.78 b 1161 [0.0019] [0.0013] [0.0005] [0.0010] [0.0017] [0.0011] [0.0006] [0.0005] Female-majority 6.78 c 5.39 b 0.67 c 0.73 b b 5.54 b 0.79 c 0.71 b 1751 [0.0015] [0.0012] [0.0002] [0.0005] [0.0017] [0.0016] [0.0004] [0.0004] Equal # females & males 7.21 b 5.44 b 0.82 b 0.97 a a 5.78 ab 1.09 a 0.93 a 2135 [0.0018] [0.0012] [0.0004] [0.0015] [0.0016] [0.0009] [0.0005] [0.0009] All roups [0.0011] [0.0008] [0.0003] [0.0007] [0.0012] [0.0008] [0.0004] [0.0005] Note: Tax incidence rates for different categories within groups are statistically compared column-wise. Cells with different letters are statistically different at 5% significance level Incidence of indirect taxes by types of households, presence of children and expenditure quintiles Table 4.5a illustrates the incidence of indirect tax by gender employment and expenditure quintiles for all households with or without children. The incidence of total indirect tax is higher for households without children than households with children for all types of taxes considered (Table 4.5a and Table 4.5b). The incidence of total indirect tax is lowest among households with no member employed. This is consistent with expectations given that such households will in all likelihood spend more on essentials which are likely to be exempt or zero-rated. There are marked differences in the incidence rates between families with children and families without children for all male types of households. The same pattern of differences is also observed for households which have both employed males and females or have equal number of adult males and females. The female types of households also have higher incidence rates for higher income 19

22 households without children but incidence rates for poorer female types of households that do not have children are lower. Incidence of indirect taxes by gender of employment status Table 4.5a shows that, excise taxes are generally regressive for male breadwinner households, irrespective of whether they have children or not. This is not surprising as we note from Table 4.2 that male headed households are likely to spend more on goods such as alcoholic beverages and tobacco. Incidence rates for fuel taxes for this group are also consistent for different expenditure quintiles but they are generally progressive meaning that richer male breadwinner households have higher incidence rates than the poorer ones. The pattern for female breadwinner households does not show consistency with respect to different levels of expenditures. The statistical test of equality of ratios reveals that excise taxes for female breadwinner households with children are generally regressive but the taxes for female breadwinner households without children are generally proportional (with significant lower rates for the 3 rd and 4 th quintiles). Fuel taxes for female breadwinner households are also generally progressive (with or without children), with some degree of proportionality for the 1 st and 3 rd quintiles. VAT is progressive for all gender employment groups of households, except for male breadwinner households, which were found to be u-shaped, meaning that the middle expenditures quintiles have lower VAT incidence rates than the others. Putting all these differences together, the results show that total indirect taxes are generally proportional for male breadwinner (earner) households, with the exception for the 3 rd quintile for male breadwinner households with children, which tend to have lower total indirect tax incidence than the others. On the other hand, total indirect taxes for female breadwinner households (with or without children) tend be generally progressive. Incidence of indirect taxes by sex composition of adults With the exception of incidence rate for VAT, Table 4.5b shows that incidence of indirect tax of all kinds for adult male-majority households are higher than adult female-majority households. Another general observation, which is consistent for all groups averages, is the fact that average tax incidence rates for households without children are higher than the households, of all adult dominance types that have children. There are significant variations with respect to rates for lower income groups among the female-majority households. All tax incidence rates for adult female-majority households with children in the 1 st, 2 nd and 3 rd expenditure quintiles are higher than those without children in the same expenditure groups. Excise taxes are regressive for all family types in this household typology, except for femalemajority households without children, for which excise duties are proportional for all the expenditure groups. Fuel taxes, on the other hand were found to be generally progressive for all the groups except female-majority households with children, for which the levies are proportional for all the expenditure groups. The test of equality of ratios revealed that VAT rates are generally progressive for all male and female-majority groups without children and generally proportional for the groups with children. 20