Some unresolved questions in practical project appraisal: indirect taxes and the numeraire

Abstract

Aid projects are still assessed by a methodology for project economic analysis developed largely in the 1970’s. Many such projects are in infrastructure sectors producing goods that are internationally non-traded and whose benefits are valued on the basis of approximate estimates of consumer willingness to pay. An omission in one influential strand of this literature was the distinction between analyses with and without the inclusion of indirect taxes. This has led to the potential overestimation of benefits from such projects in applied development work. This note explains this complication, discusses how it can be resolved and some of its practical consequences.

Keywords:

• Projects
• economics
• impact

Introduction

Since public sector projects involve scarce resources their economic efficiency must be tested. The standard test for project economic efficiency, project economic appraisal, is a version of discounted cash flow analysis which requires a with and without project comparison of benefits and costs in each year of a project’s life. It differs from financial analysis in the scope of its definition of benefits and costs, which in principle should capture the impact of a project on the whole economy, not just on its financial stakeholders. In the context of low-income countries Aid donors, whether multilateral, regional or bilateral, have used this methodology to assess the effectiveness of the projects they support. Project economic analysis is typically used alongside other assessments of project impact, like environmental or poverty impact assessments, so that it is not the only criteria by which a project is judged. However, in many organizations passing the economic test is a necessary, but not sufficient condition, for project acceptability.

Assessing the economic impact of projects is a potentially complex task. The practice of Aid donors has been guided by a substantial technical literature on how to assess costs and benefits in a development context. The development cost–benefit literature stems from the seminal works of Little and Mirrlees (1969, 1974) and UNIDO (1972) and subsequent refinements in Squire and van der Tak (1975), UNIDO (1978, 1980) and Ray (1982) with some differences of emphasis in Harberger (1972). We term this tradition ‘development cost benefit analysis’ in contrast to analyses in the context of higher income economies. ¹ The full set of possible adjustments set out in this literature was never applied on practical grounds. ² Nonetheless with some, often quite significant, simplifications this literature has formed the basis the practical appraisal of projects financed by organizations like the World Bank, regional development banks like the Asian Development Bank, the African Development Bank and the Inter-American Development Bank, as well as bilateral donors like the UK Department for International Development.

In Europe the approach has also been applied in lending by the European investment Bank and the European Commission. This paper focuses on a specific aspect of the methodology – the treatment of indirect taxation – that is a potential source of confusion and where practice in the appraisal of Aid-funded projects has not always kept up with developments in the wider literature. This is evidenced by recent manuals on the economic appraisal of projects from three of the organizations mentioned above, which omit any reference to the issue covered here (EIB 2013; European Commission 2014; ADB 2017).

Some theory

Before setting out the detailed argument it is necessary to give a simplified statement of the underlying theory. ³

Because the original development literature tried to quantify the effects of projects on several different objectives – short-term allocative efficiency, longer term-growth and income distribution – a numeraire or unit in which different project effects could be weighted and combined was necessary. In practice applications have largely focused on allocative efficiency, so that the unit or numeraire is consumption or income at either domestic or world prices. ⁴ However, the discussion of the alternative numeraire largely neglected the issue of the level of prices used to value project effects, that is whether these should be those prices received by producers or paid by consumers.

Where project outputs and inputs are sold on a market the starting point for valuation is their price on that market. However, the presence of taxation, lack of information or various regulatory controls means that in practice markets are ‘imperfect,’ so that market prices need not reflect the true worth of goods and services to the economy. Economic appraisal under these circumstances requires the adjustment of actual market prices to what are termed ‘shadow prices’ or alternatively ‘economic prices’, which are used to value outputs and inputs in an economic appraisal. Where no market exists for the goods or services used by a project, economic appraisal requires a monetary value be assigned to these non-marketed transactions.

An improvement in welfare will require an increase in consumption. In theoretical terms the economic price of a project output can be thought of as its contribution to additional consumption and the economic price of a project input as its impact in reducing consumption, as it is diverted to the project under consideration. Where non-marketed effects are involved in theory they can be valued in terms of ‘consumption equivalents.’ Hence, for example, environmental damage will require resources to mitigate the effect and their use in this way will create a consumption cost. Project activities which generate or use foreign currency through the production or use of internationally traded goods in turn contribute to consumption, as the availability of foreign resources will increase domestic consumption. ⁵

This paper argues that the Little and Mirrlees (1969, 1974) and UNIDO (1972) approaches and analyses based on them fail to incorporate adequately the effect of indirect taxation. The issue is that once indirect taxes are applied to a project’s output or the inputs it uses, it is necessary to be consistent in the choice of price level at which to do project calculations. The alternatives are either to use prices received by producers (‘factor cost prices’) or prices paid by consumers (‘market prices’) with the difference between these determined by indirect taxes. Indirect taxes are excluded from producer prices but included in consumer prices. The normal procedure for moving from prices at one level to the other is to apply an economy – wide ‘indirect tax correction factor’ estimated as the average rate of indirect taxation in the economy (Sugden 1999). An incorrect result will be obtained if project benefits include indirect taxes, whilst project costs exclude them.

Development cost–benefit analysis usually uses factor cost prices, but under some circumstances uses prices paid by consumers, thus, mixing the two price levels and adding together different price units. This generally applies where benefits are valued inclusive of indirect taxes, whilst some costs exclude them, so that net benefits are overstated. The main development texts do not explicitly recommend this incorrect treatment but because of a lack of clarity in the discussion of indirect taxes this potential error can arise.

Indirect tax correction factor

The indirect tax correction factor is the average rate of indirect taxation in an economy and is the adjustment used to move between the two price units. The logic is that if ‘t’ is this average rate of tax then goods valued at 1 net of tax are valued at 1 + t, inclusive of tax. Hence for a project that costs $1 million at factor cost, benefits to consumers (valued through a willingness to pay survey, for example) must be at least $1 + t million since surveys are at market prices. With costs net of indirect taxes to make benefits and cost comparable requires that benefits are divided by 1 + t. Alternatively, if benefits are kept unchanged then costs must be multiplied by 1 + t.

In the main development cost–benefit analysis texts there is little discussion of the treatment of indirect taxation. This becomes an issue in relation to the valuation of non-traded goods (that is goods sold in the domestic market with no export or import replacement effect). The most obvious examples are various infrastructure projects in the road, rail, power and water sectors. Regarding such goods there is a distinction, either explicit or implicit, between incremental and non-incremental effects created by a project. In UNIDO (1972) the output of non-traded goods, which are incremental in the sense that project output adds to consumption is determined by what consumers are willing to pay for them, which will be a value inclusive of any indirect tax. Non-incremental project output arises where supply from the project replaces supplies from another source (such as a renewable energy project allowing an older thermal plant to be phased out). In this case, benefits will be the saving in cost from this displaced production and these will be valued net of any indirect tax. In practice for some projects producing non-traded outputs, benefits may be a mixture of both types of effect (so a new power plant can both add to supply to the grid and allow an older plant to be closed). Similarly, non-traded inputs used by a project can be either incremental, in the sense that project demand leads to additional production of the input, or non-incremental so its use by a project diverts the input from other users. In the incremental case such goods are valued at their cost of supply with indirect taxes deducted, whilst in the non-incremental case they will be valued at what other users are willing to pay for goods, which will be inclusive of any indirect tax. ⁶

Little and Mirrlees (1974) adopt a similar framework, but with an important difference. They have a brief discussion of indirect taxation on consumer goods and suggest that these should be omitted from their valuation of incremental project output, but only the condition that any indirect tax reflects a charge for a development ‘bad’, such as use of a polluting material or luxury consumption. Thus, they make the strong assumption of a rational government imposing indirect taxes to cover negative externalities, rather than for general revenue purposes. As they also recommend removing indirect taxes from the supply cost of incremental inputs they do not acknowledge the potential inconsistency in situations where indirect taxation is used for revenue reasons, so by their argument should not be omitted from the valuation of output. ⁷

In practice in a number of Aid appraisals the approach for non-traded goods is that indirect taxes are included in an estimate of benefits that reflects willingness to pay of consumers and in the valuation of non-incremental inputs taken from other users by a project, as the tax-inclusive price is taken to reflect willingness to pay. This approach is fully consistent with use of a market price unit. On the other hand, where project outputs are non-incremental, so competing supplies are displaced, or where inputs are incremental, as their supply is expanded to meet project demand, valuation is on the basis of the resource costs involved in production and the standard procedure is to omit indirect taxes on the grounds that these are transfer payments. Hence, there is the potential for the mixing of valuation at factor cost and market prices.

Where internationally traded outputs and inputs are involved, world prices will be converted to national currency at an exchange rate termed a shadow exchange rate (SER). The most commonly applied formula for the SER adjusts the market rate for the average rate of taxes and subsidies on foreign trade. This can include both import tariffs and any domestic indirect taxes also levied on traded goods. Normally, the net average tax on trade will be at a different rate to the average rate of indirect taxation as applied to domestically produced goods. Where the tax rate on traded and non-traded domestic goods is the same, the use of the SER in economic appraisal is equivalent to the application of the indirect tax correction factor to traded goods.

Important exception

The indirect tax correction factor has rarely figured in discussion on cost–benefit analysis in a development context. The important exception is a separate strand of the development literature that stems from the work of Arnold Harberger and Glenn Jenkins, who have approached the problem of economic appraisal from a different perspective (Jenkins and Harberger 1992; Harberger and Jenkins 2002). ⁸ A simple way of thinking of the distinction between the different approaches is that Little and Mirrlees (1969, 1974) and UNIDO (1972) can be interpreted as seeking direct estimates of economic prices. Hence, for example, for a project output like power or water an estimate is made of what consumers are willing to pay for this output as a measure of what this is worth to them in terms of additional consumption. Similarly, the economic cost of employing a worker, technically termed the ‘shadow wage’, can be based on an estimate of their productivity in their alternative employment, as this approximates the lost consumption as a result of their employment on a project.

On the other hand, the approach following Jenkins and Harberger (1992) is to create an estimate of an economic price starting from the actual market price and adjusting for the effect of taxes or subsidies, which are a negative tax. Thus, in the two previous examples willingness to pay can be approximated by adding indirect taxes paid by consumers on to the ex-factory price. The economic cost of labour can be taken as the wage paid, net of income tax payments as a measure of the consumption compensation workers require to take on a new job on a project. ⁹ Hence their approach is to use taxes and subsidies directly in the estimation of economic prices. Thus, indirect taxes figure prominently in their analysis and their work avoids the inconsistency, which arises in practical applications of the other literature.

Some examples

To illustrate the potential error of ignoring the factor cost/market price distinction we turn to some simple examples, with numerical values given in Tables 1–3. First, we take a simple case of a project (such as a power or a water supply plant) with incremental non-traded output as a benefit (so either power or water consumption is increased by the project) that uses two non-traded inputs one incremental and the other non-incremental and labour. ¹⁰ By definition, the incremental input has its supply expanded to meet the demand of the project and the non-incremental input is diverted from other users, so it is this latter case that is relevant for the analysis of indirect taxation.

Table 1. Example 1.

t = 0.10	Factor cost	Market prices	Adjusted development cost–benefit approach
Pj = 100	Pj = 100/1.1 = 90.9	Pj = 100	Pj = 100 – 100*(1 – 1/1.1) = 90.9
P n = 50	P n = 50/1/1.1 = 45	P n = 50	P n = 50 – 50*(1 – 1/1.1) = 45
C y = 30	C y =30	C y = 30*1.1= 33	C y = 30
L = 10	L = 10	L = 10*1.1 = 11	L = 10
	NB1 = 5.45	NB2 = 6.0	NB3^1 = 5.45

Table 2. Example 2.

t = 0.10, t f = 0.20, OER = 1.0, SER = OER*(1 + t f ) = 1.2	Adjusted development cost–benefit approach
Pj = 100	Pj = 100*1.2 = 120
P n = 50	P n = 50/1/1.1 = 45
C y = 30	C y =30
L = 10	L = 10
	NB4 = 35

Table 3 Example 3.

t = 0.10, t f = 0.20, OER = 1.0, SER = OER*(1 + t f ) = 1.2	Adjusted development cost–benefit approach
Pj = 100	Pj = 100*1.2 = 120
P n = 50	P n = 50*1.2 = 60
C y = 30	C y =30
L = 10	L = 10
	NB5 = 20

In the second and third examples, we introduce internationally traded output and inputs.

In example 1, algebraically for output j the willingness to pay price inclusive of indirect tax is Pj. The willingness to pay price for non-incremental input n is P _n which is inclusive of indirect tax, whilst the cost of production at factor cost (exclusive of any indirect tax) for incremental input y is C _y . Labour cost L is labour’s opportunity cost at factor cost prices. The average indirect tax correction factor is 1 + t where t is the average rate of indirect taxation. Hence in the two price alternatives net benefits (NB) are given as follows:(1) (2)

In this adjustment in a factor cost analysis all values initially at a willingness to pay price are divided by 1 + t and correspondingly in a market price analysis all values at factor cost are multiplied by the economy-wide average rate of indirect taxation. The results will be equivalent but in different numeraire so that NB2 = NB1*(1 + t).

Where individual items like output j and inputs n and y pay specific tax rates t _j , t _n and t _y the equivalence will breakdown. Now,(3) (4)

and there is no simple equivalence between NB1¹ and NB2¹.

Table 1 example 1 uses the prices P _j  = 100, P _n  = 50, C _y  = 30 and L = 10, while the average tax rate t = 0.10. Putting these values into the formulae gives NB1 = 5.45 and NB2 = 6.0.

The potential error following what we term the development cost–benefit approach arises where output j is valued at P _j , as a willingness to pay price and input n is valued at P _n , as a measure of the willingness to pay for the input. However, y at is valued at C _y , which is the economic cost of supply net of taxes and labour is valued at L, as output foregone, both at factor cost prices. Hence net benefits are,(5)

which is a mixture of market price and factor cost values. In factor cost terms relative to NB1 this implies benefits are overstated by P _j *(1 – 1/1 + t) and costs are overstated by P _n *(1 – 1/1 + t).

Using the values from Table 1 this gives NB3 = 100 – (50 + 30 + 10) = 10. The overestimation of net benefits relative to NB1 is (100 – 100/1.1) – (50 – 50/1.1) = 4.55.

This inconsistency can be resolved by arguing that with total expenditure in an economy unaffected by an individual project incremental expenditure on a non-traded output reduces expenditure on other goods by an equal amount, in this case by P _j . This reduction in expenditure elsewhere means a reduction in government indirect tax revenue to offset the taxes paid on j (Jenkins et al. 2011c). The fall in government revenue elsewhere is treated as a cost which reduces the benefit from the production of j. An equivalent argument but in the opposite direction is used in the case of non-traded non-incremental inputs diverted from other users. In this case other users have P _n per unit to spend on other goods and this diverted expenditure will create tax revenue that is a benefit of the project.

This treatment of changes in taxes as an additional benefit or cost may seem inconsistent with the canonical principle of cost–benefit analysis that taxes are transfers that merely redistribute income. Its rationalization is that in a competitive market a tax (treating a subsidy as a negative tax) is the only difference between a demand price (reflecting willingness to pay) and a supply price (representing unit resource costs). Hence, a tax measures net economic benefit and this will accrue as income to the government. Therefore, when demand shifts away from a substitute good as a result of the production of incremental output the resulting change in indirect tax revenue represents not a transfer but a change in economic benefits as the difference between the demand price and the supply price (interpreted as a consumer surplus) is lost. Similarly, when demand shifts towards a substitute input as a result of the use of a non-incremental input on a project the difference between the demand price and the supply price (or consumer surplus) is gained. ¹¹

In calculating the loss in tax revenue as demand for j diverts expenditure away from substitutes the average indirect tax rate t is used on the assumption that the diverted expenditure could go on any domestically produced good. The reduction in expenditure P _j triggers a loss in government tax revenue of P _j *(1 – (1/1 + t)). ¹² Using this assumption where j is also taxed at the average rate there will be no net change in the government’s tax position and the adjustment will be equivalent to valuing P _j at its willingness to pay price minus the indirect tax that is included in this price.

Similarly for non-traded input n which is diverted away from other users by a project, other users will have funds equal to P _n to spend on other goods. P _n includes the tax component and if it is assumed again that this available expenditure could be spent on any good there will be an additional tax take of P _n * (1 – (1/1 + t)), which reduces the cost of using n.

By incorporating the loss in government tax revenue due to production of j and the gain of tax revenue by the use of n, NB3 can be rewritten as:(6)

which reduces to,(7)

or P _j /(1 + t) – P _n /(1 + t) - C _y - L

so that NB3¹ equals NB1.

In Table 1 example 1 the adjusted development cost–benefit calculation is NB3¹ = (100 – (100 – 100/1.1)) – (50 – (50 – 50/1.1)) – 30 –10 = 5.45.

Compatibility with a factor cost analysis, thus, requires allowing for the indirect tax impact of the use of non-traded goods on government income. This requires subtracting tax foregone – estimated as actual expenditure diverted from another good multiplied by the average indirect tax rate – from a willingness to pay estimate of value.

The adjustment through the use of an economy-wide indirect tax rate was not incorporated in the original development texts and is only an approximation since the actual distribution of marginal expenditure diverted away by a project cannot be known. Where the exact composition of diverted expenditure can be assessed with reasonable accuracy the economy-wide rate of tax should be replaced by a product-specific rate for the goods affected by the diversion of expenditure. Where a specific item such as output j or input n is taxed at the average rate the adjustment is equivalent to valuing j or n at their net of tax or factor cost price. However, the simple rule of valuing an incremental non-traded output (or a non-incremental non-traded input) net of their actual tax rate will normally be misleading since due to the existence of untaxed sectors the average economy-wide tax rate will be below the rate for sectors subject to a general tax like Value-Added Tax. Deducting a VAT rate of 15%, for example, when the economy-wide average is 5%, will understate benefits by roughly 10%.

Traded goods

In the case of traded goods, application of the economy – wide indirect tax correction factor will not be relevant, as they will not normally be subject to the same tax rate as domestic goods. As noted earlier should the tax rate on traded goods be the same as for domestically produced goods the SER adjustment will be equivalent to revaluation by the indirect tax correction factor. In the second example, we assume output j is traded as an export. Output j will be valued at its world price (in this case fob) before any trade tax or subsidy, but then converted at an exchange rate the SER that incorporates the average effect of net taxes on foreign trade. Keeping inputs and labour as before and valued at factor cost prices, net benefits will be,(8)

where P _j is the fob price, OER is the market exchange rate, t _f is net average tax on foreign trade and SER = OER(1 + t _f ).

Here all costs are at factor cost prices, whilst average indirect trade taxes are part of benefits (P _j *SER) reflecting willingness to pay for foreign exchange. This is the standard ‘partial equilibrium’ approach for traded goods. ¹³ The benefits of the project can be interpreted as P _j *OER, which is a factor cost value since P _j is before tax, plus the tariff revenue raised for the government through the generation of foreign exchange by the project exports. The export revenue is spent on other goods leading to a tax revenue for the government of Pj*OER*t _f . Here, it is the tax rate for traded goods t _f that is relevant not t. This tax revenue is a net benefit, not a transfer, as it represents the average difference between willingness to pay for traded goods and the price actually charged. However, any specific tax or subsidy on export j that affects its domestic price is treated as a transfer between the project and the government and hence export j is valued at its fob price. The foreign exchange that the sale of j creates is then revalued by the SER. As in the case of non-traded goods it is assumed that the actual pattern of expenditure is not known and hence the average can be used to approximate the way in which marginal or additional expenditure on traded goods is allocated. Hence, the average rate of tax on traded goods t _f can be applied to estimate tax revenue.

Table 2 example 2 illustrates the approach. In this case the average rate of tax on trade t _f is 0.20, whilst the average indirect tax rate t is 0.10. For simplicity the market exchange rate between foreign and local currency is taken as 1.0, so the SER = 1.2. The export price P _j is 100 and all other prices and costs are as before, so P _n is 50, C _y is 30 and L is 10. Now the export good j is valued at 100 and the foreign exchange it generates is adjusted by the SER of 1.2 to give an output value of 120, so net benefits NB4 are 35. The only difference from a standard development cost–benefit calculation is that P _n , which is a non-incremental input, is revalued by multiplication by (1/1 + t) so that it valued at a price of 45, which is net of the foregone taxes due to the diverted expenditure. As before non-traded input y is treated as incremental in this and the following example and thus is valued at its economic cost of production, which is net of taxes.

Where a project uses imported inputs a similar argument holds, but in reverse. In the third example we assume that whilst output j is an export good input n is now imported or importable, so P _n is a cif import price. The net benefit can now be rewritten as,(9)

As before P _j and P _n , as world prices before tax, can be interpreted as factor cost prices, whilst the change in the net revenue position of the government is (P _j *OER*t _f )– (P _n *OER*t _f ). Export j creates import tariff revenue for the government as the export revenue it generates is spent on imports, whilst import n reduces tariff revenue as its use of foreign exchange diverts expenditure away from other goods. Again, the assumption is that the average rate of tax on traded goods is relevant as the actual marginal allocation of expenditure is unknown. The project-specific import tariff on n, which is omitted from the calculation, is a transfer between the importer and the government. As in the case of the export good the average rate of tax on foreign trade t _f is treated as a net economic benefit arising from the difference between a demand price, reflecting willingness to pay and a supply price, reflecting economic cost. It is a net benefit that accrues to the government.

Table 3 example 3 illustrates this case. As before the SER is 1.2, P _j is 100 and now P _n of 50 is treated as an import price, which must be revalued by the SER, to give a value of 60. Net benefits NB5 = 20. In this case there is no diversion of expenditure as a result of the project’s diversion of expenditure on non-traded goods and hence the indirect tax correction factor has no role. Hence in this example the development cost–benefit approach does not require any modification.

Conclusions

This paper aims to clarify the adjustments needed to make what we have termed the development cost–benefit approach to project economic appraisal and the factor cost/market price distinction compatible. The key adjustment is in the valuation of non-traded goods. Where a project produces additional output of such goods willingness to pay benefits need to be adjusted for the loss in tax revenue that would have been spent on substitute goods, otherwise there is inconsistency and an over-estimation of benefits. This means that the tax foregone per unit of output as a result of expenditure diversion due to the project must be deducted from the willingness to pay unit value. The adjustment is not relevant where the project output is non-incremental, so that it replaces other supply sources, as willingness to pay is not the basis of output value. Although in practice it is less important a similar argument applies where non-traded inputs are diverted from other users as a result of the demand by a project. Now there will be a higher expenditure on substitutes for such inputs and the tax collected from such expenditure should be deducted from the tax-inclusive price of such goods. Similarly, this adjustment is not relevant in the case of non-traded inputs which are incremental, where valuation is on the basis of costs of production. Use of these adjustments will convert estimated economic benefits and costs to factor cost prices and remove any inconsistency.

As with other aspects of applied economic appraisals major simplifications are involved. This procedure is based on the assumption that total expenditure in an economy is given, that the allocation of diverted expenditure can be approximated by an economy-wide average, that taxes reflect the only difference between demand and supply prices, that any further indirect effects on taxes further back in the supply chain can be ignored and that indirect taxes do not reflect a deliberate charge to cover an external cost. If applied, this adjustment will have the effect of lowering benefits from projects in non-traded sectors producing incremental output, although there is a partial offset in reducing the cost of using non-traded non-incremental inputs.

In sectors like transport or energy where indirect tax rates may be well above the economy average and where it may be possible to predict the destination of diverted expenditure (for example, from road to rail, or from different energy sources) the tax implications of such diversion could be significant. In such cases where the destination of diverted expenditure can be estimated the ‘indirect tax correction factor’ representing the economy-wide average will need to be replaced by the product or sector-specific tax rates. ¹⁴

The potential inconsistency highlighted here is relevant for projects in the non-traded sectors where currently development aid tends to be focussed, principally in infrastructure activities such as power, transport, water supply and urban development. Its practical implications could be important under two scenarios. First, where the public investment portfolio is dominated by non-traded (typically infrastructure projects) which primarily add to the supply of goods and services, rather than replace or phase out existing supplies. The faster the growth of an economy the more likely it will be that the incremental effect of adding to supply dominates the non-incremental replacement effect of infrastructure projects. Second, where indirect taxation rates in the economy are high and the tax net is spread widely so that there are few untaxed sectors.

Where infrastructure projects have mostly non-incremental effects these will be valued at cost savings and the treatment of willingness to pay will be unimportant in the overall appraisal. Similarly, if average indirect tax rates are low at an economy-wide level, due to the existence of many untaxed sectors, the practical implications of the inconsistency discussed here will be small. For example, the indirect taxation to GDP ratio for India was recently around 6%, which would imply an over-estimation of gross benefits of a little under 6%. ¹⁵ Deducting indirect tax no longer collected due to diverted expenditure by a project under these circumstances may thus have only a limited impact on benefits, particularly as there may be some offset on the cost side. In addition, if applied consistently this adjustment could lead to a small lowering of the discount rate if this is based on an opportunity cost rate, although any adjustment would be very minor.

However, the adjustment for tax represents a form of economic surplus foregone and for initially marginal projects with a high proportion of benefits in the form of incremental outputs or for economies where the tax base for commodities is spread fairly widely, this omission could be important in over-estimating project benefits. An indirect tax rate of 20%, for example, if applied widely might give an average rate of 15%, which if ignored would overstate gross benefits from incremental output by approximately 13%. Even where the tax net is spread more widely an average rate of 10% would imply an overestimate of 9%. ¹⁶ These figures are clearly large enough to affect the ranking of some projects and could alter an accept/reject decision.

Disclosure statement

No potential conflict of interest was reported by the author.

Acknowledgement

I am grateful to Glen Jenkins and James Laird for comments which stimulated this paper and for Armin Reiss for comments on the paper. Any errors are my responsibility.

Notes

1. The well-known textbook on cost–benefit analysis Boardman et al. (2014) has a separate chapter on ‘applications to developing countries’ where it summarises this literature.

2. Reviews of the practice of project economic analysis at the World Bank are given by Deverajan et al. (1995) and Jenkins (1997). World Bank (2010) is a detailed internal review by the Independent Evaluation Group.

3. This summary is subject to numerous simplifications and the original works should be referred to for a fuller analysis.

4. Apparent differences between the alternative approaches arose from the basic choice of numeraire government income or investment measured at world prices in one (Little and Mirrlees 1974) and domestic consumption at domestic prices in the other (UNIDO 1972). Subsequent work showed how the alternatives could be reconciled relatively easily (Curry and Weiss 1993, 2000; Potts 2002). In practice a distinction is rarely drawn between income that is saved and income that is consumed, or between government and private income. There are also practical reasons why the simplified version of the world price approach applied in operational work can cause an additional level of misunderstanding (ADB 2017).

5. As is discussed further below foreign currency is valued at an economic price, the shadow exchange rate (SER), which should in principle reflect consumer willingness to pay, as a result of the additional goods that foreign currency makes available.

6. See UNIDO (1972, 53–57).

7. See section 12.5 in Little and Mirrlees (1974, 223–228).

8. Harberger (1971) provides the theoretical basis for their approach based on valuation through consumer and producer surplus and Harberger (1972) gives practical applications.

9. In updated versions of their work Jenkins et al. (2011b) elaborate on their approach to valuation of labour and Jenkins et al. (2011c) discuss their approach to the valuation of internationally non-traded goods like power and water. It should be noted that the use of taxes and subsidies to adjust market prices is only valid theoretically where there are no other controls or rigidities in the markets concerned. In practice use of this approach usually ignores any other ‘distortions.’

10. As noted above, in practice it is possible for a project’s output to create a mixture of incremental and non-incremental benefits. We ignore this possibility to allow a focus on the key conceptual point regarding the role of indirect taxation.

11. This discussion ignores any additional indirect taxes further down the supply chain. An exception to this argument is where the indirect tax on a substitute good genuinely reflects an adjustment for a negative externality, such as a higher than average tax on a pollutant. This is the assumption made by Little and Mirrlees (1974). In this case the difference between the demand and supply price due to the tax will reflect an external cost not a consumer surplus. In such cases the change in indirect tax revenuedue to diverted expenditure cannot be treated as a cost (in relation to project output) or a benefit (in relation to a project input). Therefore, the tax-inclusive price must be used to value an incremental project input and the tax foregone from diverted expenditure on the incremental output side should not be deducted from benefits.

12. As P _j and P _n are inclusive of tax at rate t the non-tax component is (1/1 + t) in per cent.

13. More complex versions of the SER formula depend upon what is assumed about the uses or sources of funds for the foreign exchange generated or used by a project (Jenkins et al. 2011a). The simple approach assumes there are no non-tariff barriers to trade.

14. However, as noted above, where the indirect taxation on the alternative to the project has an environmental dimension the indirect tax involved should not be deducted from project benefits. Thus, for example, an urban subway project may be intended to both divert existing traffic off roads and allow the future growth of traffic by rail. This latter incremental effect will mean lower expenditure on road travel. Where the indirect taxes involved cover charges for emissions or congestion the tax component of the diverted expenditure is a cost avoided not a loss of consumer surplus and should not be deducted from project benefits.

15. Data from the Government of India, Ministry of Finance, Department of Revenue; downloaded from www.dor.in/direct_indirect.

16. Following Equation (1(1) ) benefits are overstated by the expression 1 – (1/(1 + t)), where t is the average tax rate.