A Tool for Assessing the Distributional Impact of VAT Tax Expenditures

VAT works by adding/applying a tax at every step of a product's journey. Each time something is made or sold, VAT is applied to the selling price. But the VAT paid in earlier stages (input VAT) is subtracted from the VAT charged in the next stages (output VAT). Finally, when the product is sold to the customer, VAT is added to the selling price, and the retailer subtracts their input VAT. This way, VAT taxes the increase in value at each stage, thus the name ‘value added tax’.

In VAT systems, there are two ways goods get special treatment: exemptions and zero-ratings. Both mean that no VAT is added to a product's price.  The difference is that with an exemption, retailers can't reclaim/get back the VAT they paid for the inputs. However, with zero-rating, retailers can reclaim the input VAT they paid, which is better for taxpayers.  VAT exemptions and zero-rating undermine the VAT system, leading to reduced revenue and complexity for taxpayers. However, they are often used to achieve/meet certain socio-economic objectives, such as poverty reduction. For instance, basic foods, which are vital for the poor, are often zero-rated.

While VAT exemptions and zero-ratings are often used to pursue noble objectives, they might not be the best way to achieve socio-economic goals. An obvious issue is that these exemptions apply to all product sales, regardless of the buyers wealth. For example, even the rich buy staple foods, leading to significant revenue loss benefiting wealthier people rather than those intended to be helped.  

There is a danger that governments are foregoing significant revenue by offering VAT exemptions or zero-rates on products when it would be more effective to tax these products at the standard rate and use the resulting revenue on specific anti-poverty initiatives.

Likewise, education and healthcare often receive VAT exemptions and zero-ratings, such as VAT zero-rating for private education. This is done partly to support the consumption (and production) of “merit goods.” However, one could argue that an alternative approach could involve applying VAT and using targeted expenditure to subsidize these goods.

An important factor in assessing the wisdom of VAT tax expenditures is the distributional impact of these expenditures, in other words, who benefits most from the government foregoing this revenue.

As part of the Asian Development Bank’s (ADB) support to the Ministry of Finance in Armenia on VAT tax expenditures, a tool was developed to assess the distributional impact of these expenditures. This tool demonstrates the financial implications of a specific VAT tax expenditure for each income percentile. A demonstration was also developed based on private education consumption in Armenia, which currently benefits from a VAT exemption. This tool can be used to analyze the effects of both introducing or removing such tax expenditures.

Additionally, the tool can be used to investigate the impact on consumption (i.e., how much VAT tax expenditures incentivize additional consumption), producers and retailers, and government revenues.

Analyzing income distribution

The model for this tool is based around 100 rows, each representing a percentile of a country’s income distribution. For example, percentile 1 represents the average income of the poorest 1% of the population, while percentile 100 represents the average income of the richest 1% of the population.

The availability and accuracy of income distribution data may vary from one country to another, but it is normally the responsibility of the national statistics agency to produce such data. In the case of Armenia, the data from a World Bank Living Conditions Survey in 2018 was used.

Figure 1: Tool for Assessing the Distributional Impact of VAT Tax Expenditures (Armenia Demo)

Source: Author.

Due to space constraints (Figure 1), only the bottom ten and top ten income percentiles are displayed. For each row representing a percentile income distribution, the tool displays the following mini-tables:

Income and consumption data per percentile: This shows the pre-tax average income per year, the estimated product consumption, in absolute terms and as a percentage of pre-tax income.

Impact on consumption from tax expenditure removal: This shows the price increase resulting from removing the VAT exemption or zero-rating (applied uniformly across all percentiles), the price elasticity of demand for the product per income percentile, the resulting percentage change in consumption, and the new consumption expenditure on the product after the exemption/zero-rating is removed.

Impact on individual welfare: The impact on individual welfare consists of both the change in price paid for post-reform consumption and the value of the change in consumption. If a VAT exemption or zero-rating is removed, the price for the (lower) consumption increases, and there is also a reduction in consumption. This welfare loss resulting from the reduced consumption is calculated by the percentage reduction from the initial consumption value. The impact on individual welfare is shown in absolute terms and as a percentage of income. This is because a fixed welfare loss has a more significant impact on individuals with lower earnings.

Additional impacts: This shows the impact on retailers/producers from the change in consumption resulting from the removal of the VAT exemption or zero-rating, in addition to the additional revenue generated. This information is also displayed per percentile, so it needs to be aggregated across all percentiles to get the total impact on retailers/producers and on government revenue.

The accuracy of the tool depends on just four assumptions:1) how consumption of the product varies with income (the consumption profile), 2) the amount of input VAT reclaimed, 3) the extent of price pass-through (i.e., how much VAT changes lead to price changes), 4) price elasticity of demand and how it varies over income.

Creating a consumption profile

The first step is to create a consumption profile of the product in question, which means understanding how consumption changes across the income distribution and how it is distributed among each income percentile. Developing a highly accurate consumption profile can be challenging and often requires survey data or other rigorous methods. Alternatively, for a more practical approach, if you have information on the total amount spent on a product (which can be obtained using VAT base data for an exemption/zero-rating), you can make assumptions based on the nature of the product.

In the case of our demo, we had figures for the total expenditure on private education in Armenia and only needed to distribute this across the income percentiles. Clearly, private education is a relatively luxury good, with expenditures likely to increase as wealth increases. For the purposes of the demo, we created an equation to achieve the following two outcomes: 1) the percentage of consumption accounted for by each percentile increases with income, and 2) the total consumption of private education equals what is found in the VAT records. 

The equation was as follows: % of consumption undertaken by each percentile = n2/X, where n = number of percentile (e.g., 1 for the first percentile), and X is a denominator identified using the Excel “goal-seek” function (e.g., for the 50th percentile, the equation is 502/X).

The value of "X" was then determined using the Excel “goal-seek” function to find a value that ensures that the sum of percentages equals 100%. This shows the percentage of total consumption which is accounted for by each percentile.

The results are presented in Figure 1 (columns C and D) and in graph form in Figure 2 below, where both consumption expenditure and consumption as a proportion of income are displayed. Consumption increases in absolute terms but accelerates before gradually declining as a proportion of income.

Figure 2: Consumption in Absolute Terms and as a Proportion of Income (Armenia Demo)

Source: Author.

Estimating price impact of VAT

Once the tool has modelled the existing consumption across income percentiles, it can be used to model the impact of removing a VAT exemption or zero-rating on real income. The process consists of three stages: 1) estimating the impact of VAT on prices, 2) estimating the impact of price changes on consumption, and 3) estimating the impact of consumption changes (both quantity and value) on real income.

To estimate the impact of VAT on prices, there are two steps:

1. Effective VAT impact: This is the VAT rate minus input VAT, which would be subtracted if VAT was applied. It represents the input VAT which cannot be reclaimed from exempt items. Users must estimate the input VAT that would be collected if exemptions were removed. For zero-rated suppliers, this would be 0%, but for exempt supplies, it could be any value up to the VAT rate.
2. Price pass-through: This reflects the extent to which suppliers and retailers pass on taxes or tax cuts to consumers instead of absorbing them as additional costs or benefits. A value of 1 indicates that the tax is entirely passed on to consumers, while a value of 0 means that the VAT is entirely absorbed by retailers.

In our Armenia demo, we assume the input VAT reclaimed is 2%, resulting in an effective VAT impact of 18%. Additionally, we assume a price pass-through of 0.5, meaning that 50% of the VAT exemption is passed on to consumers, while the other 50% is absorbed by suppliers. Consequently, removing this VAT exemption will lead to a 9% price increase (18% multiplied by 0.5).    

The impact of price changes on consumption depends on the price elasticity of demand (PED). This measures how consumption responds to price, with a value of 1 implying that a 10% price increase/decrease leads to a 10% decrease/increase in demand, while a value of 0 implies no change. PED typically tends to be higher for lower-income groups who are generally more price-sensitive. Several other factors also influence PED, including the availability of close substitutes and the essential nature of the product.

As PED is likely to differ across the income distribution, the tool allows users to adjust the PED based on income level. Two inputs are required: the PED median (representing the PED for the 50th income percentile) and the PED range (indicating the maximum extent by which the PED deviates from the median).

In our demo, we use a median PED of 0.5, signifying that for the medium percentile, a 10% increase in price results in a 5% decline in demand. Additionally, we consider a range of 20%. This means that for the highest percentile, the PED is 0.3, and for the lowest percentile, it is 0.7. This accounts for the likelihood that as income increases, PED tends to decrease, and removing a VAT exemption will have less impact on consumption.

To calculate the impact on real income due to changes in consumption, we combine 1) the change in the cost of current consumption (post-reform consumption multiplied by the price increase, in this case 9%); and 2) the change in consumption (dependent on the price change and PED).

In this demo, the net impact of removing the VAT exemption closely aligns with the consumption profile discussed earlier, meaning that the most significant effects, in terms of income proportion, were experienced by those between the 50th and 70th percentiles.

This tool can also be used to examine the welfare impact on producers or retailers, which results from changes in consumption and the VAT applied on post-reform consumption. In our demo, removing the exemption for private education and applying a 20% VAT leads to both reduced sales and increased taxes paid by suppliers on those reduced sales. The additional tax paid by suppliers depends on the extent of “pass-through,” which in our demo, was assumed to be 0.5.

Finally, the tool can also be used as an additional revenue forecasting tool, for both estimating the revenue foregone from introducing VAT zero-rating or exemptions, or the expected revenue from ending such a tax expenditure.

This tool is valuable for countries conducting analysis on the distributional impact of VAT tax expenditures. It can be used by officials and policy makers to understand the distributional impact of VAT policy changes, shedding light on how the removal of VAT exemptions or zero-rating can primarily benefit wealthier groups, both in absolute and, at times, proportionate terms.

This could have significant policy implications, providing a quantitative justification for VAT reforms aimed at broadening the VAT base. This has the potential to generate substantial revenues for the government and may lead to the implementation of more effective measures to alleviate poverty. It could entail placing VAT on essential goods and services such as food, education, and health, resulting in increased revenues. Consequently, taxes paid would generally rise with increasing wealth, allowing for the funding of more impactful poverty relief initiatives, including generous welfare systems, education, and health expenditures.

However, it is crucial to note that, like any tool, the quality of the analysis heavily relies on the quality of the data. The consumption profiles for distinct products and the required assumptions for price pass-through and price elasticity of demand are not readily available for most products in many countries. Furthermore, the suggestion that exemptions and zero-ratings should be replaced by direct government expenditures hinges on the effectiveness of public financial management systems.

Asian Development Bank (ADB). Armenia: Tax Policy Review (Tax Incentive). Domestic Resource Mobilization Fund.