Tax expert, Dr Terry Dwyer, demonstrates the inequities of a GST that taxes food and other necessities

GST/VAT: Zero-rating Theory and the Taxation of Food


Given that the arguments for taxing food rest upon theoretical economic models of optimal commodity taxation, it is necessary for the reader to understand the crucial assumptions behind these models.  At its simplest, the argument for taxing food and other necessities of life rests on the assertion that a broad tax base at a lower rate is less distorting than a narrower tax base with a higher tax rate.  This argument is correct, all other things being equal.  But other things are not equal where it is proposed to broaden the tax base by taxing costs of production.  The whole point of a GST/VAT is to tax final consumption only and to free inputs to production from tax.

What has been overlooked by the “broad base/low rate” argument for taxing food is that labour supply is an input to production and the supply of present and future labour itself involves real costs of production.  As classical economists, such as Adam Smith and Alfred Marshall would put it, food for the worker is as much a cost of production as is fuel for a machine or fodder for farm animals.  Food consumed in raising up the workers’s children to replace his worn-out body in the productive process is also as much a cost of production as replacement of worn-out machines.

Whereas income tax system recognize these subsistence costs through tax-free thresholds, zero-rating of food is required to achieve a similar exemption for human food under a GST/VAT.  Zero-rating is the equivalent for labour supply to input tax refunds on other inputs to production such as fodder for animals.  There is no reason in economic theory why input taxes should be refunded on fodder for animals but human food is taxed.  To the extent that modern optimal tax models ignore necessary inputs to labour supply such as food, they lose real world relevance and are less insightful than classical analysis.


Most countries with a VAT or GST do not tax food or give it lower rate treatment.  Yet many modern economic theorists today (in contrast to their classical predecessors) argue that any argument for not taxing the necessities of life runs counter to the conclusions of optimal tax theory on economic efficiency.

The comment of Kay and King (1983 p 240) is typical “There is a long tradition in Britain of slightly hysterical opposition to the taxation of food; the origins of these sentiments lie in controversies over the relative political roles of agriculture and commerce and the desirability and effects of  Imperial Preference in tariff policy which have no contemporary relevance.  It is now suggested that food is a necessity; but very little of the food which would not be eaten if it were 15 per cent dearer is in the least necessary and other taxed  goods such as clothing are “necessary” in exactly the same (limited) sense. The substantive argument for exempting food is that such exemption is progressive in its distributional effects; the rich spend a smaller proportion of their incomes on food than the poor.”

It is the purpose of this chapter to explain that such modern views misconceive both economic history and the history of economic thought and that this has led to serious error in contemporary theorizing.  When looking at economic phenomena, if a practice is common, economists usually ask whether there is an underlying or implicit theory or rationale for the approach taken.  Economists are – or should be – reluctant to assume irrationality on the part of legislators or voters.  Rather than assuming that the non-taxation of food represents an irrational action, economists might begin by looking more closely at the arguments presented for taxing food under a GST or VAT.

At the outset, it should be remembered that no one in Australia is advocating VAT or GST as an end in itself.  The purpose, for business at least, has been improving international competitiveness by removing taxes on business inputs.  No country, not even Australia, can expect foreigners to pay taxes on our exports.  These taxes get shifted back to Australia in the form of reduced exports, reduced incomes and reduced tax bases.  A shift to a goods and services tax with full refunding of input taxes, especially one that involves getting rid of fuel excise, would eliminate the bulk of indirect taxation of exports.  The refund feature of VAT/GST results in removal of the cascading effect suffered by exports from existing indirect taxes.[1]

But competitiveness also depends on labour costs.  It should not be forgotten that consumption taxes can feed into high labour costs just as European social security taxes feed into the basic costs of European exporters, even though they get VAT refunds and may be exempt on foreign sales income.  As we shall see, the question of how taxation of the necessities of life, such as food, affected wage costs lay at the heart of the condemnation of such taxes by  Adam Smith and his successors in the classical British school of political economy.[2]

The arguments for taxing food

It is worth setting out the arguments currently advanced for not zero-rating food in a GST/VAT.

1.         Food is already taxed by WST and fuel excise.[3]

This argument proves nothing.  First, the level of embedded indirect taxation of food is far lower than it would be under a standard rate GST/VAT.  Second, if one favours the non-taxation of food it is a strong argument for zero-rating food to eliminate the existing embedded  indirect taxes.

2.         One cannot afford to forgo the revenue from not taxing food.

This is essentially a revenue argument[4], which is to say that it is no argument.  If all that matters is the collection of tax revenue, there is no need to concern oneself with tax theory.  Any tax will do, good or bad.  If getting money in is all that matters, one could even suggest that revenue raising be farmed out to the highest bidder (as was done in Renaissance France).  But one suspects that outsourcing revenue raising to the Mafia would not commend itself to most people.

3.         It would be administratively complex to exempt food.[5]

This argument has somewhat more merit but should not be accepted uncritically.  First, the vast majority of countries do cope with the administrative difficulties of classifying food.  The zero rating of “basic groceries” was never even an issue in Canada.   Second, the tax administration has had to cope with far worse definitional problems under the wholesale sales tax.

4.         The benefit of zero rating goes to the middle income earners and the rich.

This argument applies to any general exemption in any direct or indirect tax.  It applies equally to the universal tax threshold in the income tax.  Most taxpayers who get the benefit of the tax-free threshold have incomes well above it.  Yet to take away the tax free threshold would still disadvantage lower-income earners more than higher income earners.  From a vertical equity perspective, what counts is the proportional, not absolute, benefit of an exemption to high versus low-income taxpayers.  Second, any attempt to limit a tax exemption to low income earners creates disincentive traps (ie higher equivalent marginal tax rates) over the income ranges where the exemption is phased out.  Finally, this argument really falls to be dealt with under the next argument on how to deal with the regressivity of taxing food

5.         Taxing food is regressive but that can be compensated for through means tested welfare measures.

That a tax on food is regressive (ie biased against low income earners) can hardly be doubted.[6]   As Quiggin (1993, p 65) puts it “A food tax is highly regressive.  This fact reflects the limited capacity of the human stomach.  Although richer people can substitute more expensive foods for the cheaper fare of the poor, they eat no more in total than the poor.  Although much rhetorical use has been made of the presently tax-free status of luxury foods such as caviar, these are in fact of very little importance.”[7]

This argument, unlike the previous argument, correctly recognizes the essential regressivity of taxing food.  It is also correct to say that, in a statistical sense, the regressivity of one part of an indirect tax system can be compensated for by changes in the direct tax-transfer system.  But this in turn raises fundamental problems.  If compensation is not means tested, eg there is a general increase in family tax thresholds or lump sum demogrants or family payments are made, then one incurs similar net costs to the revenue as if zero rating were given in the first place with the added administrative costs of churning the revenue and possible disincentive effects on labour supply.  To get the benefit of zero-rating or exemption, a taxpayer has to work first to produce the income to spend; he has to do nothing to get a general cash transfer.

Second, if the compensation for the taxing of food is given on a means-tested basis, then far worse disincentive traps can be created, especially for larger families.  Abatement of the compensation can mean very high equivalent marginal tax rates (EMTRs) over large income ranges.   Problems of this kind have been found to have been created by the compensation schemes for the GST in New Zealand.[8]  Such problems are already a serious enough problem in Australia with current income transfer schemes such as family payments and AUSTUDY.  There seems little merit in taxing food if it involves making these problems worse.

At a fundamental level, the very fact that one has to compensate for a tax change such as taxing food must lead one to question the merit of the taxing proposal in the first place.

6.         It is not economically efficient to zero-rate food under a GST.

This argument is the core of the Treasury argument for taxing food.  It has several strands which need to be taken apart and examined separately.  It is the major intellectually respectable argument for taxing food under a GST and therefore stands in a class of its own in terms of policy importance.  It is also quite wrong.

The economic efficiency argument for taxing food

There are several levels to the economic efficiency argument for taxing food.

1.         First, it is argued that not taxing food under a GST/VAT creates economic deadweight losses though distortions of consumption.  If anything, food should be more heavily taxed, being inelastic in demand.[9]

The argument here depends on comparing the deadweight loss generated by a GST at one rate which does tax food with the deadweight loss generated by a GST at a higher rate raising an equal revenue which does not tax food.  In this scenario it is not hard to show that there is a higher deadweight loss caused by distorting consumption.

However, the assumptions behind this kind of analysis are not often understood.

X         First, it may be pointed out that the alternative state of the world specified is not one the taxpayer may wish.  Why should the revenue “lost” from zero-rating food have to be made up from a higher rare of GST?  The taxpayer may point out (as did Professor Paul Samuelson in his historic 1951 paper on optimal tax theory) that land taxes are much more non-distorting and he would prefer to see tax revenue come from land taxes rather than taxing food, or indeed, any goods or services.  Second, the taxpayer may prefer that government cut back welfare transfers instead of pushing up taxes: he may argue that instead of following the European pattern of high GST/VAT taxes to pay for ageing welfare states, Australia would be better served by following a lower welfare/lower tax route that made Australia more, rather than less internationally competitive.

X         The computation of deadweight loss depends on computed price elasticities.  These elasticities vary between types of taxpayers with different consumption patterns – families for example consume more basic food than single people.  Elasticities can also change with large tax changes, as opposed to marginal tax changes.

X         One might argue that since food demand is not elastic than there is little distortion created by zero-rating so why not zero-rate food if equity concerns incline you in that direction?[10]

X        It needs also to be noted that the question of whether food is an intermediate input to production is not considered in this sort of analysis (this is discussed later).

A more sophisticated argument for taxing food might go along the following lines.  The Ramsey rules for optimal taxation of commodities argue for tax-induced reductions of consumption in equal proportions.  To achieve this, it is necessary to have commodities taxed inversely in relation to their compensated price elasticities.  That is, the more sensitive demand for a commodity is the less it should be taxed and the less sensitive the demand for a commodity the more it should be taxed.  Hence, being inelastic in demand, food should, if anything be more heavily taxed than the standard GST/VAT rate.

It is then argued that, as a practical efficiency matter, however, since we cannot know all the price elasticities of all consumers in the economy, it is likely that a standard equal rate of GST/VAT on all goods and services will most clearly approximate efficiency.

This sort of argument for taxing food has been advanced by Freebairn (1997, p 381) but, as Creedy (1993b, p 86) has pointed out, it is not really satisfactory even on its own terms.  If you do not know enough about price elasticities to impose the varying optimal tax rates prescribed by the Ramsey optimal tax analysis, why do you know enough to know that equal taxation of all commodities is efficient?  If you are ignorant, you are ignorant and hence cannot say that not taxing food is inefficient.  You may guess that the administrative savings of a “tax all things equally” policy may cure the inefficiencies of departing from the optimal Ramsey taxing rules but you cannot demonstrate that there is overall improved efficiency.  This is particularly so, if you are trying to optimize collective economic welfare in a many-consumer world where different consumers have different tastes and endowments and you attach different weights to different consumers’ levels of well-being.

Thus the Ramsey optimal tax rules do not support uniformity – they supports taxing things in inelastic demand, provided these are final consumer goods, not inputs to production.  A uniform GST/VAT is a second best guess, even if we are sure we taxing only final consumption.

Questioning the Ramsey optimal tax rules where there are necessities of life

Surprisingly, there appears to have been little questioning of the Ramsey optimal tax models as regards the modelling treatment of the necessities of life such as food.

GST/VAT is not a consumption tax: it is a tax on non-business purchases of goods and services.  A tax on purchases by consumers is not the same thing as a tax on consumption.

What tends to be forgotten is that a consumption tax should only be on final consumption.  That is why business inputs to production are freed from GST/VAT.  But business inputs are not the only costs of production.  Under the usual form of GST/VAT labour supply is input taxed and costs of producing labour income are not zero-rated or refunded.  Yet zero rating or exemption with credit is the correct method of applying VAT to those goods or services required by labour to contribute to production.

If it is argued that GST/VAT should exempt from tax the costs of purchases made by a business why should we not also exempt from tax the cost of the food and fares necessary for a working man to keep his body and soul together and arrive at his place of work?  It is not true that the only inputs to production are what businesses buys and that salaries or wages are a pure net surplus.

As the French philosopher Bertrand De Jouvenal observed, “the profit-seeking enterprise has a treble advantage over the family, which is taxed at progressive rates and is not allowed to provide for depreciation of its assets or to deduct operating expenses.  And yet the family performs in society no less important a function than the firm.  The firm produces the goods, the family produces the people.  It is puzzling that the needs of the former should be so well understood by the law-makers and the needs of the latter so disregarded.  It seems that law-makers can picture only the firm as an institution with a purpose and therefore respectable.  The income recipient, on the other hand, his day’s work done, is seen as going round the booths of a fair, blueing his rights to consumer-satisfaction.  It is not realised that he is an entrepreneur in his own right.  He marries, sets up a house, raises children, and, it is to be assumed, struggles to bring himself and his family to the greatest possible degree of accomplishment.  His achievement is to be recognised as useful to society in that he fits himself and his descendants for their roles as producers; in this respect, it is an indirect contribution to national income … It is quite incomprehensible that a breeder of dogs for the race-track should be allowed his cost, depreciation etc, while the father of a family is not.  It is as if the law-makers sympathised more with the purpose of the former, which is to sell dogs for the track, than with the purpose of the latter, which gives men to society – incidentally for soldiering and taxpaying… Admittedly, it is impossible to disentangle from family accounts something which might be called the net income of the family.  A net income can easily be arrived at in the case of firms because net income is precisely what they are out to obtain. But if something of the thoughtfulness which has gone into the appraisal of corporate needs was brought to bear on family needs, the cost of maintaining a home, of developing talents and so forth might certainly be taken into consideration .”[11]

Curiously the point becomes clear if one thinks of a slave economy.  No one could argue that the costs of subsistence of slave labour was not an input to production and GST/VAT would be refundable on food purchases by the slave owner just as the GST/VAT on food for cattle is refunded.  Alfred Marshall (1920 I, p 691) observed long ago that “as a rule the shrewd slave owner goes to some trouble and expense to promote rough musical and other entertainments, on the same principle that he provides medicines: for experience shows that melancholy in a slave is as wasteful as disease, or as cinders that clog the furnace of a boiler.  Now if the standard of comfort of the slaves were to rise in such a way that neither punishment nor the fear of death would make them work unless provided with expensive comforts or even luxuries, they would get those comforts and luxuries; or else they would disappear, in the same way as would a breed of horses that did not earn their keep.”[12]

It would be a curious result if taxation theory were allowed to proceed on the basis that the  costs of subsistence of slave labour are to be treated a business input not taxable under a GST/VAT but the subsistence costs of free labour are to be ignored.

The parallel with income tax treatment of production costs

There is a close parallel between income and consumption taxes.  Since income equals consumption plus savings, whatever is a cost of earning income and deductible for income tax purposes can hardly be taxable as final consumption under a properly conceived consumption tax.

Zero rating or exemption with credit is to GST/VAT what deductions or exemptions are to income tax: if an expenditure on a good or a service is a cost of doing business, GST/VAT should be refunded or not charged on the input.

It follows that where income tax deductions are allowed for work related expenses, such as tools of trade, then GST/VAT should be refunded or not charged on such purchases.  Otherwise, we end up with a situation of input taxing a cost of production.  In the income tax deductions for work related expenses recognize that some costs of labour are directly relevant to producing income and should be subtracted in arriving at net income.

There are also personal exemptions in income tax systems which recognize the subsistence costs of the worker and his or her dependants.  In addition, income tax systems often recognize that medical expenses, educational or superannuation or insurance expenses should be deducted because they represent indirect labour costs – before a worker can be productive he has to be fed, clothed and educated and it is necessary that he not be ill.

In the long run, for an economic system to continue it is also necessary that workers reproduce themselves.  The raising of children is to the labour force what the replacement of machines is to the capital stock.  Hence income tax systems have often allowed exemptions or deductions in respect of dependant costs and governments have even often subsidized such costs, this being analogous to government subsidies or grants to investment such as bounties or investment allowances over and above reasonable depreciation.

Once it is recognized that some costs of labour are either directly relevant to earning an income or are costs of maintaining, improving or renewing human capital, then the case for zero-rating such costs under a GST/VAT becomes immediately apparent.  For GST/VAT to be truly taxes on final consumption only and not taxes on gross rather than net wages, GST/VAT must zero rate or refund taxes on these direct and indirect labour inputs to production so that only net surplus from wage income spent on consumption is taxed.

Income tax and GST/VAT equivalences lead to the following conclusions.  If an expenses is properly wholly deductible or exempt under an income tax, then it should be zero-rated or exempt with credit under GST/VAT.  If an expense is partially deductible under an income tax it should receive the benefit of a reduced rate of GST/VAT.

Both income tax and GST/VAT have to deal with what are costs of production.  The problem of a GST/VAT inappropriately taxing costs of production is analogous to the distortions caused where an income tax does not recognize costs as deductible.  For either tax base, if the base includes costs there is economic distortion and this is equally true for labour or capital or business income.

Although it is commonly stated that a “broad base, low rate” tax system is more efficient than a “narrow base, high rate” tax system, this is only true if all other things are equal.  The “broad base, low rate” dogma is in fact quite wrong if it is mistakenly taken to mean it is efficient to lower the rates by broadening the base to extend to taxing costs of production.  Indeed quite dangerously wrong –  would any business group accept a company tax rate of 10% paid for by abolishing all ordinary business (the former s 51) deductions?

There may appear to be two apparently conflicting tax principles:

X         there should be no tax on costs of production[13]

X         a broad base with a low rate is superior to a narrow base, with a high rate.

The conflict is resolved once it is realized that costs are not part of the tax base and one should never broaden a tax base by including costs of production as part of the base.

Just as an income tax should only tax net income, a consumption tax should only tax final consumption.  But, in practice, a GST/VAT is not limited to taxing final consumption.

Taxation of the necessities of life

The traditional argument against taxing the “necessities of life” is an economic efficiency argument, not just an equity argument.   Once it realized that labour supply is an input taxed activity under GST/VAT, one sees that labour supply costs are being taxed.  Yet food, clothing and shelter may be seen as part of the costs of labour inputs.  They are inputs to the business of supplying what Adam Smith described as “useful labour”.[14]

Taxation of food

In particular, food should be zero-rated because subsistence requirements are meant to be untaxed – they are clearly a cost of the supply of labour.[15]  If it be argued that labour only works 8 hours out of 24 and therefore food costs are only partially cost of production, it should be noted that depreciation is allowed on a machine for the whole 365 days a year even if the machine is worked only one 8 hour shift a day.  Down time or rest time is still part of the cost of an input to production.

If it be argued that food consumption may be in excess of what is required for subsistence, it may be noted that basic food consumption is limited by human stomach size: one can assume virtually all basic foods are inputs to production, unlike many other consumer purchases.  It might also be noted that business costs generally are deductible under income tax as costs of earning income even though they may be to some extent a matter of choice – the Taxation Office does not deny deductibility of rent as excessive merely because an office is on the 35th floor with a view of Sydney Harbour.

It may also be observed that if health and education are to be zero-rated because they are seen  as indirect inputs to production (and they are subsidized by governments) there is even more reason to zero-rate food (which is unsubsidized), especially since good food is a form of preventative medicine.

Components of labour income

The case for zero-rating the necessities of life may be seen by considering short-run labour costs and long-run labour costs.[16]

Conceptually, wages may be broken into several components, reflecting different aspects of the supply cost of labour and final personal consumption.  Thus we may break down wages, W, as follows:

W = W1 + W2 + W3 + C, where C is true final consumption


W1       represents immediate short-run labour costs: s 51 deductions (these clearly should be untaxed for the same reasons they are deductible eg work clothes, tools of trade, subscriptions to associations etc.)

W2         represents less proximate costs of working – journey to work, subsistence costs, food, clothing, shelter, health and education etc.  These are short-run and medium-term costs.[17]  By definition, the necessities of life means you have to live before you can work.[18]

W3         represents long-run costs of human capital formation and workforce replacement.  These are essentially the costs of raising and educating children to replace their parents in the workforce.[19]

It should be noted that food is an element in all these components of labour supply costs.  In  the first component, we may note that meal allowances and rations of personnel serving on ships etc are exempt from income or fringe benefits tax, being accepted as costs of production.  (A curious inconsistency arises if food is not zero-rated: if an employer, such as the merchant navy, supplies rations to its sailors GST/VAT will be refunded but if the employer is a trucking company supplying meal allowances, the food purchased by the truck driver on the road will not be refunded as input tax.)

In the second component, subsistence food requirements have been met historically by a tax-free threshold.  Taxing food is analogous to taxing the tax-free threshold.

In the third component, food figures largely as a cost associated with raising children and this has been recognized historically by the provision of child tax allowances, child endowment and family payments.  It is also worth noting that families with children benefit most from zero-rating food.[20]  The fundamental problem of an indirect consumption tax is that it does totally ignore this necessary investment in human capital if an economic system is to continue.  Yet it may be more relevant than it was in the nineteenth century.   European birth rates have declined as VAT has raised the costs of child rearing and the European demographic collapse  has as a consequence the long term insolvency of their unfunded welfare states.[21]

Marshall (1920, I p 67-68) asks “It is common to distinguish necessaries, comforts, and luxuries; the first class including all things required to meet wants which must be satisfied, while the latter consist of things that meet wants of a less urgent character. … When we say that a want must be satisfied, what are the consequences which we have in view if it is not satisfied?  Do they include death?  Or do they extend only to the loss of strength and vigour?  In other words, are necessaries the things which are necessary for life, or those which are necessary for efficiency?”  He goes on to argue for the latter sense, stating that “the income of any class in the ranks of industry is below its necessary level, when any increase in their income would in the course of time produce a more than proportionate increase in their efficiency.  Consumption may be economized by a change in habits, but any stinting of necessaries is wasteful.”[22]  He then enumerates “the necessaries for the efficiency of an ordinary agricultural or unskilled town labourer and his family, in England, in this generation.  They may be said to consist of a well-drained dwelling with several rooms, warm clothing, with some changes of underclothing, pure water, a plentiful supply of cereal food, with a moderate allowance of meat and milk, and a little tea, etc., some education and some recreation, and, lastly, sufficient freedom for his wife from other work to enable her to perform properly her maternal and her household duties.  If in any district unskilled labour is deprived of any of these things, its efficiency will suffer in the same way as that of a horse that is not properly tended, or a steam-engine that has an inadequate supply of coals.  All consumption up to this limit is strictly productive consumption: any stinting of this consumption is not economical, but wasteful.”[23]  Marshall’s language may now seem old-fashioned, but his insistence that much consumption is an input to production remains of crucial significance for any theory of efficiency in taxation.

History of tax theory on taxing necessities of life

Although the effect of a GST/VAT on labour inputs to production has been largely overlooked in the literature, it figured prominently in both the theory and practice of economic policy in Britain from the seventeenth to the twentieth century.  It was particularly important in the writings of Adam Smith, David Ricardo and John Stuart Mill in the classical era of British political economy.

Essentially, the classical theory of subsistence wages, the “iron law” of wages, held that labour supply would in the long-run adjust if wages were above or below subsistence level.  It is important to note that “subsistence” meant something more than a physiological minimum and meant a conventional minimum level of decent living standards.  From this it followed that a tax on wages or the necessities of life would be passed on the employer or, if not, result in a shrinkage of labour supply and the economy.

The classical economics argument for not taxing (subsistence) wages or necessities was thus that taxes on the poor (working people) become taxes on profits, as these costs are passed on to employers.[24]  Whereas the Physiocrats in France and David Ricardo in England had tended to assume that all wages were consumed by subsistence requirements, Adam Smith was more subtle.  He drew a distinction between necessaries and luxuries, recognizing that there was a part of labour income which could represent a true surplus above the real costs of labour supply.  Nonetheless, as a general proposition, Adam Smith, too, pointed to a crucial limitation in the modern economic models which are generally used to support a case for a consumption tax.  He made the very simple and obvious point that an increase in the cost of living caused by a consumption tax will necessarily make it more difficult to bring up families.  The long term labour supply of a country may therefore respond adversely to the introduction of a consumption tax.  For an economic system to continue, workers have to eat, to be sheltered from the elements and raise up their successors.

Thus Adam Smith opposed both taxes on the necessities of life and the wages of labour.  “A tax upon the necessaries of life operates exactly in the same manner as a direct tax upon the wages of labour”[25]“Any rise in the average price of necessaries, unless it is compensated by a proportionable rise in the wages of labour, must necessarily diminish more or less the ability of the poor to bring up numerous families, and consequently to supply the demand for useful labour…”[26]“If direct taxes upon the wages of labour have not always occasioned a proportionable rise in those wages, it is because they have generally occasioned a considerable fall in the demand for labour.  The declension of industry, the decrease of employment for the poor, the diminution of the annual produce of the land and labour of the country, have generally been the effects of such taxes.”[27]

Because taxes on wages or the necessities of life had adverse effects on wage costs, consumer demand and long term labour supply, Adam Smith urged that “the middling and superior ranks of people, if they understood their own interest, ought always to oppose all taxes upon the necessaries of life, as well as all direct taxes upon the wages of labour.”[28]  In denouncing taxes on the necessities of life, Adam Smith was particularly scathing.  “There is nothing so absurd, says Cicero, which has not sometimes been …asserted by some philosophers.”[29]

The fallacy that all labour income is surplus income, capable of being taxed, is present in our existing income tax system.  But, at least in theory, our income tax system through the exemption of a subsistence threshold and through some residual recognition of dependent rebates does recognise that not all labour income is surplus income capable of being taxed.  The great error of an indirect consumption tax of the value added kind is that it totally overlooks this fundamental economic truth.

Ramsey optimal tax rules reconsidered

The Ramsey optimal tax rules are usually said to imply the need for higher taxes on goods in inelastic demand such as food.  But this is a static or short run model approach – in effect, it assumes consumer existence is fixed and he has only to allocate time between leisure and other commodities.  Such an approach abstracts from the point that in the long-run the consumer is dead if he does not eat and both his utility and his labour supply are zero.

The phrase “necessities of life” connotes the idea that some goods are necessary to life and to work – they are inputs to labour supply and therefore intermediate goods in the process of production.  Once food is seen as an intermediate good then it makes no more sense to tax food because it is inelastic demand than it would be to tax fuel because demand is inelastic.

Ramsey pricing does not necessarily make sense when one is talking about intermediate inputs to production.  It is generally accepted that taxes on inputs to production (and that is what Ramsey pricing above marginal cost amounts to in the case of business customers of utilities) are inferior to general factor taxes or broad based taxes on final consumption.  As Docwra and Kolsen (1993 p 114) point out, Ramsey pricing should only be applied to final goods and services, not producers’ inputs.[30]  Just because the demand for a product is inelastic does not mean a high price should be charged for it: the elasticities of demand for all the factors or products which ultimately bear the tax imposed on a producer’s input may be higher.  The Ramsey rule is irrelevant to intermediate goods, eg. you would not tax the coal used by a power station just because the power station had an inelastic demand – that demand is a derived demand from the demand for electricity which in turn is derived from demand for other goods and services.

In the longer run, food and other necessities enter into wage costs.  Most economic models of excise taxes showing the optimality of Ramsey rules or arguing for equal taxation of all goods and services do not address questions of the existence of labour supply at all.  The consumer’s income is taken as a given endowment, instead of the result of short-run and long-run labour supply decisions.  Such decisions would naturally reflect the costs of working or of raising children.  If, for example, a GST/VAT raises the cost of children, all other things being equal, one would expect a reduced supply of long-term useful labour.

Problems of optimal tax models

The Ramsey optimal tax models assume that labour or time endowments are fixed and ignore both current costs of labour supply and future long run labour supply (raising children).  Yet as Marshall (1920, I p 529) recognized, some consumption is really “productive” consumption rather than final consumption – “there is a certain consumption which is strictly necessary … if any of it is curtailed the work cannot be done efficiently: the adults might indeed take good care of themselves at the expense of their children, but that would only defer the decay of efficiency for one generation.”  The Ramsey analysis is really about incremental commodity taxation of extra purchases, once the consumer exists, and not about taxing everything from the first unit.   In addition, the Ramsey rules change with elasticities which in turn are affected by demographic variables such as the presence of children, see Ray (1988a).

It is worth noting that the common claim of some tax economists that household production is a source of untaxed production and therefore represents a tax distortion misses a major point.  Much household production is indeed untaxed.  But so it should be, because it represents an input to current and future labour supply and the current and future productivity of the economy.  Alfred Marshall saw this in his discussion of productive consumption and efficiency wages but did not recognize it in his consumer surplus discussions which were looking at partial equilibrium marginal adjustments.  Optimal tax theory has grown out of consumer surplus models but has forgotten the subsistence or efficiency wages issues discussed in Marshall’s Principles of Economics in taking labour endowments as fixed.

Back to the classical tradition of not taxing food as a necessity

If the Ramsey optimal tax model were modified in this way, one could see that the classical tradition of British political economy can be linked to modern mathematical economics.  Essentially, it would restore a recognition of the difference between gross and net wages and that taxing necessities really involves input taxing labour as a production input, since necessities are intermediate goods to the supply of useful labour.

Cheap labour costs as a source of international competitiveness

Business enthusiasm for a GST/VAT largely reflects a desire to remove taxes on business inputs and exports.  The European VAT tax regime for exporters involves allowing them to sell free of indirect taxes on world markets.

A shift to a GST/VAT in Australia with full refunding of input taxes, especially one that involves getting rid of fuel excise, would eliminate the bulk of indirect taxation of exports.  No country, not even Australia, can expect foreigners to pay taxes on our exports.  These taxes get shifted back to Australia in the form of reduced exports, reduced incomes and reduced tax bases. [31]  A GST without full refunds could, of course, worsen the competitive position of some industries eg. the tourism or financial sectors could suffer if they were not zero-rated.  Of course, one could always solve the problem by eliminating indirect taxes altogether!

This arguments in favour of a GST/VAT does not conclude the question of whether it will improve international competitiveness.   It should not be forgotten that consumption taxes can feed into high labour costs just as European social security taxes feed into the basic costs of European exporters, even though they get VAT refunds.

In this context, it is worth recalling that a major factor behind the pressure in early nineteenth century Britain to repeal the Corn Laws was the recognition by Lancashire manufacturers that  cheap food could be a source of competitive advantage for Britain in international trade.  Cheap bread meant labour could be freed for industry and wages would not have to rise to feed the people.[32]

The comparative advantage of nations depends in no small measure on things which allow moderate wages to remunerate labour adequately in terms of its necessities.  Cheap food, no less than a warm climate, is a cost advantage to a trading nation.  It may not be very advantageous if indirect taxes on exports are removed by a GST/VAT but wages are driven up because the GST/VAT drives up the cost of living through heavier taxation of necessities such as food.

The insights of classical economic theory and British trade policy in the nineteenth century may be more relevant to the current debate on taxing food and other necessities of life than  most modern economic theorists may realize.  True it is that a rigid subsistence or “iron law” of wages theory is no longer with us but that does not mean it is correct to assume there is no relationship between taxes, living costs, wages and international competitiveness.  Classical economics may not have got the answers entirely right but at least it asked important policy questions.  Is it not better to ask the relevant question and have a simplified answer that may be roughly right than not ask the question at all but assume it away by not integrating labour supply into Ramsey models of optimal indirect taxation?

Summary on efficiency argument for taxing food

The efficiency arguments in favour of zero-rating food may be summarized as follows.

1.         Ramsey (optimal) tax theory does not support uniform taxation and argues for non-taxation of inputs to production.

2.         Food is an input to production.  To the extent that models treat labour endowments as fixed and independent of the “productive” consumption of the necessities of life they are mis-specified.  The efficiency argument really points to zero-rating food and other necessities of life, since labour endowments are not given independently of the economic system.

3.         Taxing food may undermine any benefit to international competitiveness from taking indirect taxes off exports as wage costs are driven up by increased living costs.  Cheap food is a source of competitive advantage for a trading nation.


Modifying the Ramsey model to take account of necessities

The Ramsey optimal tax models depend on maximizing the utility of an individual or individuals subject to production possibility and taxation and budgetary constraints.   But, as Marshall (1920, I, p 133 n 1) noted “There is however a special difficulty in estimating the whole of the utility of commodities some supply of which is necessary for life.  If any attempt is made to do it, the best plan is perhaps to take that necessary supply for granted, and estimate the total utility only of that part of the commodity which is in excess of this amount.”  In other words, the optimal tax problem presupposes the existence of the consumer and that in turn presupposes he has enough food to live.

There seem three possible analytic approaches to modifying Ramsey optimal tax models to take account of the necessities of life.

First approach

The first possible approach would be to take Samuelson’s exposition of the Ramsey model and in addition to subtracting exogenously fixed government consumption which has to be financed, subtract as well the required bundle of subsistence goods necessary for the consumer’s existence and ability to supply labour in perpetuity.  This would give effect to Marshall’s suggestion (see also Marshall 1920, I, p 841, Mathematical Appendix, Note VI).

Thus there should be an additional constraint to ensure labour supply, L, exists, ie

qi > 0 for i = 1,…N

qi > Qi for i = 1,… K, the necessities of life, without which the labour endowment, L, does not exist.

The Ramsey rule of equi-proportionate reduction in consumption then is seen to really mean  an equally proportionate reduction in demand for consumer goods over subsistence levels.  We assume enough is being consumed to keep the consumer alive with a labour endowment.

Second approach

A second, more general, would be to specify a functional interdependence between the consumption vector in the Ramsey model and the supply of labour.  This would recognize that  labour supply depends on some mix of necessities of life

L = F(qi) for i=1,… K the necessities of life and L > L*, some physiological minimum level of labour supply.

Third approach

A third possible approach would be to recognize that households are producers as well as consumers.  One could then treat the production of labour as part of intermediate goods production, just as Becker in his theory on the allocation of time recognized households produce goods and services at home.

The household in effect should be split into a producer and consumer, and only the true final consumption should be taxed.



Aaron, Henry J (ed) (1981) The Value-Added Tax: Lessons from Europe Brookings Institution, Washington D C

Auerbach, Alan J (1985) The theory of excess burden and optimal taxation in Auerbach, Alan J and Feldstein, Martin (eds) (1985) Volume I pp 61-127

Auerbach, Alan J and Feldstein, Martin (eds) (1985) Handbook of Public Economics Elsevier, New York

Australian Chamber of Commerce and Industry (ACCI) (1993) Australia’s Tax System and International Competitiveness, Canberra, March 1993

Becker, G S (1965) A theory of the allocation of time Economic Journal v 75 pp 493-517

Blaug, Mark (1968) Economic Theory in Retrospect second edition,  Heinemann, London

Bladen, V W (1974) From Adam Smith to Maynard Keynes: the heritage of political economy University of Toronto Press

Brennan, Geoffrey and Buchanan, James M (1977) Towards a tax constitution for Leviathan Journal of Public Economics v 8 pp 255-273

Brooks, Neil (1993) Sales Tax Reform and Tax Mix Change: A Canadian Perspective in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 243-307

Cannan, Edwin (1917) A History of the Theories of Production and Distribution in English Political Economy from 1776 to 1848 reprinted 1967 A M Kelley New York

Chisholm, Andrew (1993) Indirect taxation, effective tax rates and consumption efficiency in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 309-349

Coffield, James (1970) A Popular History of Taxation: From Ancient to Modern Times Longman, London

Court, W H B (1958) A Concise Economic History of Britain from 1750 to Recent Times Cambridge University Press

Creedy, John (1993a) The role of selectivity in consumption taxation: should consumption taxes be uniform? Economic Analysis and Policy v 23 March pp 1-13

Creedy,  John (1993b) The role of selectivity in consumption taxation in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 83-95

Deaton A S (1977) Equity, efficiency and the structure of indirect taxation Journal of Public Economics v 8 pp 299-312

Deaton A S (1979) Optimally uniform commodity taxes Economics Letters v 2 pp 357-361

Deaton A S and Stern N H (1986) Optimally uniform commodity taxes, taste differences and lump-sum grants Economics Letters v 20 pp 263-266

de Jouvenal, Bertrand (1952), The Ethics of Redistribution, Cambridge, 1952 (reprinted Indianapolis, 1990)

Diamond, Peter A and Mirrlees, James A (1971a) Optimal taxation and public production I: production efficiency American Economic Review v 91, March pp 8-27

Diamond, Peter A and Mirrlees, James A (1971b) Optimal taxation and public production II: tax rules American Economic Review v 91, June pp 261-278

Docwra, G E and Kolsen, H M (1993)  Fuel Taxes and Road User Pricing in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 113-121

Dowell, Stephen (1888) A History of Taxation and Taxes in England: from the earliest times to the present day 2nd edition, reprinted by A M Kelley, New York, 1965

Fay, C R (1937) Great Britain from Adam Smith to the Present Day: an economic and social survey Longmans, London

Freebairn, John (1997) Options and Prospects for Taxation Reform Economic Record v 73 pp 373-386

Fukushima T and T Hatta (1989) Why not tax uniformly rather than optimally?  Economic Studies Quarterly v 40, September pp 220-238

Groves, Harold M (1974) Tax Philosophers: Two Hundred Years of Thought in Great Britain and the United States edited by D J Curran, University of Wisconsin Press

Head, J G and Krever, R (eds) (1997) Taxation towards 2000 ATRF Conference Series No 19, Australian Tax Research Foundation, Sydney, 1997

Hollander, Samuel (1973) The economics of Adam Smith Heinemann London

Kay, J A and King, M A (1983) The British Tax System, 3rd Edition, Oxford University Press, 1983

Kennedy, William (1913) English Taxation 1640-1799: an essay on policy and opinion reprinted Frank Cass, London, 1964

Marshall, Alfred (1920) Principles of Economics Ninth (variorum) edition edited by C W Guillebaud, Macmillan 1961

Mathias, Peter (1969) The First Industrial Nation: An Economic History of Britain 1700-1914 Methuen,  London

Musgrave, Richard (1985) A brief history of fiscal doctrine in Auerbach and Feldstein (1985) Volume I pp 1-59

Newbery, David M (1986) On the desirability of input taxes Economics Letters v 20 pp 267-270

Poddar, Satya and English, Morley (1993) The Canadian Goods and Services Tax – Some Lessons for Australia? in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 47-64

Quiggin, John (1993)  Food and the GST in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 65-81

Ramsey F P (1927) A contribution to the theory of taxation Economic Journal v 37, March  pp 47-61

Ray R (1986) Optimally uniform commodity taxes and non-linear Engel curves: a generalization of the linear case Economics Letters v 22 pp 365-368

Ray R (1988a) Optimal commodity taxes in the presence of children Scandinavian Journal of Economics v 90, pp 75-92

Ray R (1988b) Household composition and optimal commodity taxes: do demographic variables matter?  Journal of Population Economics v 1 pp 213-224

Ray R (1989) Impact of demographic variables on optimal commodity taxes: evidence from UK family expenditure surveys 1967-85 Public Finance v 44 pp 437-452

Robbins, Lionel (1952) The Theory of Economic Policy in English Classical Political Economy Macmillan

Sabine, B E V (1980) A Short History of Taxation Butterworths, London

Samuelson, Paul A (1986) Theory of optimal taxation Journal of Public Economics v 30 pp 137-143 (original unpublished memorandum for US Treasury 1951)

Schumpeter, J A (1954) History of Economic Analysis Oxford University Press, New York

Seltzer, Lawrence H (1968) The Personal Exemptions in the Income Tax National Bureau of Economic Research, New York 1968

Shoup, Carl S (1960) Ricardo on Taxation Columbia University Press

Simons, Henry (1938) Personal Income Taxation:  The Definition of Income as a Problem of Fiscal Policy, University of Chicago Press, 1938

Sinn, Hans-Werner (1997) The Value of Children and Immigrants in a Pay-as-you-go Pension System: A Proposal for a Partial Transition to a Funded System US National Bureau of Economic Research Working Paper No 6229, October, Cambridge, Massachusetts

Smith, Adam (1776) An Inquiry into the Nature and Causes of the Wealth of Nations Glasgow edition, Clarendon Press, Oxford 1976

Sowell, Thomas (1974) Classical Economics Reconsidered Princeton University Press

Stephens, R J (1993) New Zealand’s Tax Reform: 1984-1992 in Head, John G, editor, Fightback! An Economic Assessment, ATRF Conference Series No 12, ATRF, Sydney, 1993, pp 193-242

Stigler, George J (1952) The Ricardian theory of Value and Distribution Journal of Political Economy, June 1952 reprinted in Stigler Essays in the History of Economics University of Chicago Press, 1965

Organization for Economic Co-operation and Development (OECD) (1988) Taxing Consumption Paris 1988

Tait, Alan A (1988) Value-Added Tax: International Practice and Problems International Monetary Fund, Washington DC

Tait, Alan A (ed) (1991) Value-Added Tax: Administrative and Policy Issues International Monetary Fund, Washington DC, October 1991

Thomas, M W (ed) (1960) A Survey of English Economic History Blackie & Son, London

Wolfe, Martin (1972) The Fiscal System of Renaissance France Yale University Press


[1].         There can be little doubt that the removal of taxes on business inputs has been the driving force for reform of indirect taxes, see  (ACCI, 1993, pp  2, 5, 7, 8, 11).  As Dwyer and Larkin pointed out in their report for the Mortimer review of business programmes, the refund feature of a goods and services tax is not the only way of eliminating much of the indirect tax burden on exports – a tax rebate scheme for existing indirect taxes is also possible, see Dwyer and Larkin (1997).  But if a goods and services tax refunds completely all indirect taxes, including fuel excise, it can do the job properly.  Of course, one could always solve the problem by cutting expenditure and eliminating indirect taxes altogether.

[2].         Recent studies also interpret international competitiveness in terms of unit labour costs.

[3].         Chisholm (1993, p 332) estimates the tax on food at 3.4% for meat and milk and 5.1% for food nec, but this estimate includes not only sales tax but part of petrol excise and payroll tax.

[4].         As Seltzer (1968, p 90) remarks “excise and sales taxes do not yield large revenues unless applied to goods and services in wide use.  In fact, excise taxes on necessities such as salt, and general turnover and ‘value-added” taxes have been chosen deliberately in various countries because of their universal application and their great revenue yields”.

[5].         See Kay and King (1983, p 122).  Poddar and English (1993, p 55) give some amusing examples from Canada but it must be noted that part of the complexity arose from provincial exemptions on top and Canada still decided early on to zero-rate “basic groceries.”

[6].         Some have tried to suggest zero-rating of necessities has little effect on regressivity, see Head (1993, p 229, Kay and King 240), but this is not a mainstream view.  Seltzer (1968, p 89) points out that both in the UK in 1872 and the USA in 1913 higher personal exemptions from income tax were seen as necessary to offset the heavier weight of indirect taxes on small incomes.

[7].         Quiggin at p 69 also suggests that most countries with a GST zero rate, exempt or tax food at low rates on equity grounds.  Kay and King (1983, p 121) point out that if regressivity is not a concern one would not even bother with income or commodity taxes: one would use poll taxes.

[8].         Stephens (1993, p 218) suggests that the NZ Government “in attempting to provide greater income redistribution by its policies of targeting the welfare state, has effectively undone most of the simplification of the tax system, whilst increasing adverse incentive and efficiency effects from very high EMTRs.”

[9].         The argument for taxing items in inelastic demand to minimize deadweight loss of consumer surplus is found in Marshall (1920, I p 467) but Marshall recognized that this is a partial equilibrium analysis at the margin and pointed out at p 841 that the necessities of life had to be treated more carefully in the analysis.  Kay and King (1983, pp 120-122) recognize the theoretical weakness of uniform commodity tax rates but argue that information gaps may make uniformity a third-best approach.

[10].       Quiggin (1993, p 70-71) estimates the efficiency loss from zero rating food at less than $10 per person per year on the standard Harberger triangle calculation.  However, supporters of taxing food may argue that the increased tax rate on other goods may create a greater loss there.  But both arguments rest on models which abstract from the role of necessities in ensuring the existence of the consumer – and the existence of the optimization problem.

[11].       De Jouvenal, Bertrand (1952), The Ethics of Redistribution, Cambridge, 1952 (reprinted Indianapolis, 1990), pp 61-63

[12].       Marshall (1920, I, p 691) goes on to observe that “if it were true that the real wages of labour were forced down chiefly by the difficulty of obtaining food, as was in fact the case in England a hundred years ago; then indeed the working classes might relieve themselves from the pressure of Diminishing Return by reducing their numbers.”  He then points to the abolition of the Corn Laws as instrumental in keeping subsistence wages down through the free import of wheat.  Given the negative reproduction rates in most Western countries, a classical economist might surmise that direct and indirect taxation of the necessities of life had reduced wages below long-run subsistence levels.

[13].       See eg Kay and King (1983, pp 120,125) The argument for not taxing inputs but only final consumption is found in Diamond and Mirrlees (1971).

[14].       See Robbins (1952, p 74)

[15].       The classical argument of Ricardo on necessities and wages is discussed in Robbins (1952, p 86) and Stigler (1952, p 171).   Seltzer (1968, p 6) remarks that personal deductions cover the essential living expenses of taxpayers and that “The exclusion of the poor from a global or comprehensive income tax is nowadays generally accepted as necessary to avoid impairing their health, their economic efficiency and the welfare of their children.”  At p 86, he remarks “Apart from any equitable or humanitarian considerations, the … health and education of the poor and their children are important for the nation’s economic efficiency …”

[16].       Seltzer (1968, pp 96-97, 99-100) notes the arguments for a personal exemption from income tax on the grounds that much consumption expenditure is really required to earn income and only “clear”, “surplus” or “net” income should be taxed.


[17].       Brooks (1993, p 262) notes that “Persons who earn income from labour incur numerous expenses which are not recognized for tax purposes, such as the unavoidable costs of commuting, work clothes and additional household expenses.”

[18].       Henry Simons (1938, pp 54, 73-74) deals with the problem of “productive” consumption by ignoring it.

[19].       Marshall (1920, I pp 175-192) discusses the relationship between food, population and labour supply.  At pp 506-509 he discusses the “iron law” of subsistence wages and at pp 529-531 how “subsistence” may be defined by conventional concepts.  Robbins (1952, p 77) points out how Malthus’ changed ideas meant the concept of the subsistence wage supply price of labour changed to become a psychological rather than a physiological variable.

[20].       See Stephens (1993, p 230)

[21].       See Sinn (1997)

[22].       Marshall (1920, I, p 69)

[23].       Marshall (1920, I, pp 69-70)

[24].       Quiggin (1993, p 79) also argues workers will seek wage rises to recoup losses from indirect tax changes.

[25].       Adam Smith (1776, Vol 2, p 871, #5)

[26].       Adam Smith (1776, Vol 2, p 872, #8)

[27].       Adam Smith (1776, Vol 2, p 865, #3)

[28].       Adam Smith (1776, Vol 2, p 873, #9)

[29].       Adam Smith (1776, Vol 2, p 876)

[30].       In the final analysis, when excises have operated as taxes on factor incomes, what Ramsey pricing really points to is the imposition of taxes on factors of production in least elastic supply.  In that sense, it is an argument for imposing taxes on an immobile factor of production such as land, rather than mobile factors such as labour and capital.

[31].       A tax rebate scheme for existing indirect taxes is also possible.  See Dwyer and Larkin (1997)

[32].       Robbins (1952, pp 70) notes the response of Malthus to “the cry of the master manufacturers and merchants for low wages to enable them to find a market for their exports.  If a country can only be rich by running a successful race for low wages, I should be disposed to say at once, perish such riches.”   Of course, Smith, Malthus and others believed that cheap food and high wages were consistent with general prosperity.  On the interest of manufacturers in opposing the Corn Laws and securing cheap food for their workers, see Court (1954, p 235-236, 310, 315); Thomas (1960, p 278, 336, 451)