A Road to More Employment? Analysing an Employment Allowance for Australia

Research Paper 6 1996-97

Phil Hanratty
Economics, Commerce and Industrial Relations Group


Major Issues


The Core of the EA Proposal

Could an EA Increase Employment and Help Reduce Unemployment on a Sustainable Basis ?

Labour Market Substitution

Abuse Through Over-Claiming

Dissipating the EA's Effectiveness Through Wage Increases

Dissipating the EA's Effectiveness Through Price Reductions

Inaccurate Targeting of an EA

How Long Should the EA Subsidy Last ?

An Illustrative Example

Financing an EA

Comparisons with the Blandy Scheme

Perverse Equity Effects of an EA ?

Perverse Pro-Cyclical Effects of an EA ?

Tax System Complexity

Does an EA Fit with Existing Active Labour Market Programs ?

Are There Better Alternatives for Increasing Sustainable Employment ?



Major Issues

This paper analyses an unusual scheme for increasing aggregate employment, and thus helping to reduce aggregate unemployment, on a sustainable basis. This 'Employment Allowance' (EA) scheme operates through the business taxation system and seeks to increase the financial incentives for businesses to take on extra workers at the margin. Schemes of this type have long existed as theoretical ideas but have not, up to now, been implemented on a nation-wide basis by any country.

Under an EA, businesses (both companies and unincorporated enterprises) would be offered a tax concession for taking on extra workers above a certain employment growth threshold. Businesses operating both in Australia and overseas would be able to claim the extra concession for extra employment only in the Australian component of their activities.

The financial costs of such employment of extra workers (wages, training, etc) would attract a deduction of more than 100 per cent in the calculation of taxation liabilities. The current taxation standard is to allow a deduction of only 100 per cent. This extra deduction would reduce the effective cost of such employment, at the margin, to employers and thus encourages it.

EA schemes can only be effective in reducing underlying employment (which abstracts from Keynesian-style unemployment which occurs in recessions) if such unemployment is at least partly caused by real (ie. inflation-adjusted) labour costs being 'too high' in relation to worker productivity at the margin of employment. Other causes of underlying unemployment are not well-suited to attack through EA schemes and require other measures to improve the quality and flexibility of labour supply.

It seems clear that underlying unemployment levels in Australia are presently generated in part by excessive real wages at the margin of employment. However, other factors relating to labour supply have also clearly been at work here.

All business tax concessions need careful design, because many professional business advisers have great expertise in using such tax provisions in unexpected and clever ways to minimise taxation burdens. An EA would require careful design of several of its features to ensure that it was cost effective and was not abused.

  • The employment growth threshold for each business would need to be carefully specified in terms of 'effective full-time jobs' rather than just in terms of numbers of workers employed in any year.
  • The magnitude of the EA subsidy would need to be equalised across all prospective workers, both to reduce overclaiming and to increase job assistance for low wage workers.
  • Careful screening of businesses applying for the EA for the first time would be necessary to ensure that employment levels in existing businesses were not disguised as new employment in new businesses.

Fears about the wasting of the effectiveness of an EA through wage increases and/or price reductions, both of which would eliminate the profits on extra employment created by the EA scheme, seem much exaggerated. Wage increases would be contained by the targeting of an EA to extra employment at the margin and the related reality of the vulnerable position of such jobs, while price reductions would be unnecessary in many sectors because of output demand conditions which are strongly sensitive to price changes.

Any remaining adverse developments in wages and prices can be further contained by strengthening tax incentives to resist wage increases and by using macroeconomic policy, especially monetary policy, to expand aggregate demand and thus bolster prices.

It would be important to contain real wage increases for EA-subsidised positions for the length of the EA. This might require extra policy measures such as further targeted tax incentives, the creation of special EA job categories in the industrial relations system and warnings to holders of EA-subsidised jobs of the need to temporarily forgo such benefits.

As with wage subsidy schemes for specific categories of workers, EA schemes involve an inherent waste factor, paying tax subsidies on jobs which would have been created without an EA. Calculations using current Australian economic data seem to indicate that this waste factor would be of reasonable, tolerable proportions in the case of an EA. As well, such a waste component could be reduced by varying the features of the EA (such as the employment growth threshold) across industries in order to 'customise' the incentives of the EA to the economic circumstances of each industry.

An illustrative example of an EA of 10 per cent (ie. a total tax deduction of 110 per cent of the costs of extra workers) is examined to give readers a more 'concrete' impression of how such a scheme would work and of how it might affect the economy. Based upon reasonable assumptions, it is calculated that such an EA, once fully operational, could increase aggregate employment by about 80, 000 jobs and probably be self-financing in terms of its effects upon the public sector's budget balance. This desirable budget outcome occurs because the extra employment and national income generated would increase direct and indirect tax revenue by enough to cover the business tax revenue forgone through the EA concession.

EA schemes seem quite attractive compared to some alternative policies for increasing sustainable employment and are thus to be preferred.

  • EA schemes appear more cost effective than payroll tax cuts because the former are targeted to impact at the margin and are less vulnerable to offsetting wage increases.
  • General wage subsidy schemes could not use the administrative capabilities of the current business tax system as EA schemes do, especially in terms of the tax system's comparatively high quality information base and its considerable anti-abuse provisions and penalties for the supply of incorrect information.

EA schemes may have a more benign (less polarising) effect on the distribution of income than extensive labour market deregulation. Deregulation could substantially reduce real wages at the bottom of the wage spectrum, unless it was combined with redistributive tax/ transfer schemes such as an Earned Income Tax Rebate.


There is a consensus that unemployment in Australia is far too high. It is generating large social and psychological costs and represents an enormous waste of the most valuable economic resource of all: human labour. While considerable thought and public policy effort has been put into measures to reduce unemployment, much of this has been hindered, perhaps even halted, by substantial technical, institutional and political barriers and by other policy priorities.

New thinking and new ways of approaching the problem appear to be called for. One interesting proposal has existed in the economics literature for some time but has been neglected in the policy arena so far (except for one recent similar proposal). The proposal involves the use of one type of tax incentive for increasing employment and reducing unemployment, and is considered in a revised form in this paper.(1)

Specifically, the paper examines the usefulness of including an Employment Allowance (EA) in the business taxation system in order to reduce effective labour costs and, in turn, to increase labour demand and employment.

Such an allowance would comprise an additional tax deduction for employers, over and above the existing 100 per cent tax deduction for labour costs, based upon the employment of new (or better, additional) workers.

The specific example of an EA of 10 per cent of the costs of employing new (extra) workers-a total tax deduction of 110 per cent of such costs-is examined, as are the problems with this type of taxation scheme. Comparisons are also made with other methods of encouraging higher employment, particularly with other fiscal policy instruments, such as payroll tax cuts and general wage subsidies, and with further and extensive labour market deregulation.

The Core of the EA Proposal

All costs (ie. financial outgoings) related to employment are already fully tax deductible for employers. Under an EA, all costs relating to the employment of new workers would attract an additional tax deduction of, say, 10 per cent of those costs. Thus, wages, recruitment expenses, formal and informal training expenses, employer superannuation contributions and fringe benefit/ payroll tax payments incurred by employers in a financial year for employees taken on within that financial year could all attract the extra deduction for that financial year (and for some number of subsequent years). The extra deduction is only available to businesses for extra employment within Australia; businesses with overseas operations will not be able to claim the extra EA deduction on those operations.

Little extra administrative and compliance expense would be incurred by the employer in simple versions of an EA. Relatively few extra resources of the Australian Tax Office (ATO) would need to be used to implement such an EA itself since computer calculations will adjust tax liabilities once the necessary information is fed in. A considerable amount of the information needed for the EA is already collected in the current business tax/ Superannuation Guarantee (SG) system.

However, more complicated and desirable versions of the EA would require more information from businesses, would increase their decision-making costs in using an EA, and would entail higher administrative costs for the ATO. As well, it is difficult to judge the resources required for the counter-measures needed to prevent abuse of an EA (which are discussed later in this paper).

Could an EA Increase Employment and Help Reduce Unemployment on a Sustainable Basis ?

The aggregate unemployment rate is usually divided into cyclical (business cycle) and non-cyclical (underlying) components. The latter is usually called the Non Accelerating Inflation Rate of Unemployment (NAIRU) and is the unemployment rate required to stop inflation from accelerating on an enduring basis. The NAIRU abstracts from 'Keynesian' unemployment, which occurs during recessions and is caused by insufficient demand for goods and services, and focuses upon the more enduring, underlying sources of unemployment.(2) A policy proposal can reduce unemployment on a sustainable and enduring basis only if it reduces underlying unemployment (the NAIRU) in some way, and so the components of that unemployment need to be considered.

First, frictional unemployment is always one component, and is generated by workers remaining unemployed to search for better jobs and by employers leaving jobs vacant to search for better workers to fill them. Workers experiencing frictional unemployment may be new entrants to the labour force or those who have moved to a new suburb, town or region.

Second, mismatch unemployment is usually a component of underlying unemployment and is generated by mismatch between the skills, qualifications and geographical location of the unemployed and those required by employers. These types of unemployment are not generally amenable to reduction by an EA. As usually conceived of, an EA is essentially about increasing the overall aggregate demand for labour, while these two types of unemployment essentially require measures to improve the effective supply of labour and the overall efficiency of the labour market. However, it would be possible to design an EA which, along with its general provisions to be described in the rest of this paper, sought to raise labour demand for particular types of workers and thus helped reduce mismatch unemployment.

Underlying unemployment can also be thought of as often containing a third component, 'classical unemployment', brought about because real (ie. inflation-adjusted) wages are too high in relation to the productivity of workers at the margin of employment. An EA holds out good prospects of reducing this type of unemployment because it can help to generate the extra labour demand to match the existing and available labour supply.(3) Currently, Australian unemployment seems to contain a large element of classical unemployment, and thus it could be amenable to reduction through policies such as an EA. For example, as perceived by employers, the problem of long term unemployment is very largely one of the real labour costs of these unemployed being substantially above their productivity contributions to the business, at least in the short to medium term period after they are initially hired.

Thus, there are strong and plausible grounds for arguing that the EA could reduce underlying unemployment. It could do this by reducing the level of effective Real Unit Labour Costs (RULC) for employers in the economy.(4) Many econometric studies have shown that RULC does have strong effects upon aggregate employment and unemployment.(5) The EA could directly reduce RULC at the margin since, for any given levels of payments relating to new workers made by employers, less tax would be payable because of the higher deduction entailed in the EA. Thus, other things being equal, an EA reduces the after-tax RULC for employers of employing new workers and this encourages the employment of new workers.

Labour Market Substitution

However, other things may not be equal after the implementation of an EA. One problem immediately springs to mind which could undermine the effectiveness of an EA in reducing underlying unemployment. This relates to the well-known problem of 'labour market substitution' in labour market programs. Employers substitute new employees for old ones, thus qualifying for the EA on the new workers and probably increasing profits without actually increasing their total workforce. In the extreme case, aggregate unemployment and its underlying component are unchanged but after-tax profits are higher and total tax revenue lower because of the use of the EA to reduce business tax burdens.

The degree of labour market substitution in this situation under an EA would be determined by whether the EA covered the additional costs of replacing old workers with new ones. Such costs relate to differences in productivity between old and new staff as well as the costs of recruitment and training for the new workers. Replacement is only profitable if it more than covers these costs. It is possible to imagine businesses engaging in collusion to abuse EA provisions. Businesses could exchange workers of similar skills and experience (thus minimising the costs of taking on such 'new' workers) and qualify for the EA on such 'new' workers, without changing their employment patterns to any substantial degree.

If substitution were a substantial problem (as it probably would be) then it would be possible to design an EA which avoided it. Thus, the EA might only be available for net aggregate increases in employment by the employer in the given financial year rather than being available for the employment of all new workers in the financial year. For example, if an employer hired 100 new employees during the year but also dismissed 40 existing employees then the EA could only be claimed for the net increase in employment of 60 in that year. This converts the EA into a 'marginal wage subsidy scheme'.(6)

Even in the extreme and unlikely situation that aggregate hours worked for the employer are unchanged (so that average hours worked per employee falls) this EA would have the beneficial effect of redistributing work towards those previously unemployed.

Alternatively, only net employment increases above some benchmark rate of growth for employment might attract the EA. If average employment by businesses is expected to grow by X per cent in the coming year or over the next few years because of the normal growth path of the economy, then the EA might be designed so that it only applies to employment increases over and above that X per cent. Here, new workers above the X per cent threshold attract the EA for the first and some following years and are then absorbed into the general normal employment expansion path of the industry and economy, with the EA generating employment for a new set of workers in the next year. In this way an EA holds out the prospect of an outward and permanent jump in the aggregate labour demand and employment path of the economy.

Such an EA would have the very useful advantage of encouraging structural change and thus, in turn, encouraging higher productivity growth in the economy. Businesses in industries expanding rapidly would easily meet the X per cent employment growth requirement to qualify for the EA, and the extra concessions of the EA would help them to grow even more quickly. Labour resources would be encouraged to move into expanding industries. Businesses in declining industries would generally not be able to meet the X per cent employment growth requirement for the EA and would thus not qualify for it.

Alternatively, the X per cent rule could be varied for each industry so that all industries were effectively qualified for it. In this case, the EA could encourage structural change within each industry by further encouraging the most rapidly growing firms.

One major advantage of this restriction is that, by reducing the possibility of abuse/ overclaiming through labour market substitution, it better targets the tax concession on increasing aggregate employment and thus allows the rate of the concession to be higher than it would otherwise be for any given level of revenue forgone by the concession.

The direct tax system would seem to be the best administrative vehicle for implementing an EA structured to have an aggregate employment qualifying rule. This is because much information on employment is already supplied by businesses to the ATO with their tax returns, especially in relation to the requirements for satisfying the provisions of the Superannuation Guarantee (SG) system, and it would thus be relatively easy for the Government to implement the employment qualifying rule of the EA through the tax system. It could also use the existing substantial penalty provisions of the tax system in regard to the supply of incorrect information.

Abuse Through Over-Claiming

Abuse by 'New Businesses'

Any EA with aggregate employment qualifying rules would need to take special care to prevent abuse by businesses seeking to use the EA for the first time. In general, new (or young) businesses should be given access to an EA because they will be major vehicles for employment growth and will thus enhance the effectiveness of an EA. However, abuses could thus occur.

For example, existing businesses could seek to abuse an EA by continually changing their registered names and other paraphernalia to give the appearance that they are 'new' businesses.(7) In principle, this would allow them to claim the EA for their entire workforce since they will give the appearance of having an entirely new workforce (ie. going from 0 to X number of workers and thus qualifying for the EA on all of them) whereas in reality it may be the case that no new workers have been employed. Such firms, with continually changing names and identities (and Tax File Numbers), will clearly abuse the EA scheme and waste the tax concession provided.(8)

EA guidelines would need to be put in place to eliminate such abuse. They could, for example, only allow businesses to access the EA for the first time if they could prove that they were genuinely new businesses rather than existing businesses masquerading as new ones. The proprietors of businesses seeking to access the EA for the first time could be required to show that their businesses were different from those owned and operated by proprietors who had already accessed the EA by having, say, a different location of operations from previous businesses and with substantially different workforces such as, say, a 50 per cent turnover in workers. Such qualifying rules would seem to be both administratively feasible and effective in preventing abuse, but would entail higher administrative and compliance costs.

It should also be noted here that many new businesses will not be able to have direct access to the EA in their first few years of operation because they will be incurring operating losses rather than profits. However, those losses will be carried forward as deductions to be used in the years when the businesses begin to make profits, so that the extra EA deductions relating to higher employment at the margin should also be able to be carried forward to those latter years. In this case there may also be a need to prevent the transfer of losses between related businesses which include an EA amount so that the benefit remains with the employing firm.(9)

Alternatively, the EA might be made ' tax refundable' and thus in effect be converted to marginal wage cash-payments for new businesses which need some more immediate assistance to encourage their establishment and development. That is, payments would be made to firms whose tax liability is already zero. This action would be necessary to maximise the employment effects of the scheme and to dampen complaints that an EA 'discriminated' against new businesses. Making the EA refundable would be essential if the scheme was to be extended to public sector agencies since the bulk of these organisations are not a part of the direct tax system.

Loading the EA with High-Wage Workers

Another likely avenue for abuse and overclaiming would be for businesses to seek to concentrate their high-wage workers in the positions subsidised through the EA in order to reduce their tax burdens and their average after-tax labour costs. Inflows and outflows of workers in the business could be manipulated to put the highest waged workers into the EA-assisted positions.

The obvious method for dealing with this problem is to reduce the EA available for high waged workers compared to that for low waged workers, and thus to remove the incentive for this type of substitution. As well, a dollar limit might be placed on the value of the EA available for each extra job created at the margin within the business. Indeed, the EA might be so structured that it offers equal tax benefits for all extra workers, regardless of the wage paid, so that businesses would receive no financial gain from changing the workers whose wages receive the tax benefit.(10)

In terms of general anti-abuse provisions, an aggregate limit on the dollar value of the EA tax relief available to each business might also be useful/ necessary in containing overclaiming.

Breaking up Full-time Jobs into Part-time Jobs

Because the number of unemployed persons in Australia is far too large, aggregate increases in employment by businesses would be rewarded by an EA scheme. However, businesses are likely to abuse the EA by breaking up full-time jobs into part-time ones, thereby increasing the total number of persons employed while not necessarily increasing the aggregate hours worked in the business at all. The constant amount of aggregate work time would merely be spread over a larger number of people.

While such redistributions of work would be welcome to some extent for equity reasons, if allowed to go unchecked this manipulation of the scheme is very likely to make an EA scheme financially and politically unworkable.

One obvious way of preventing this manipulation is to specify a base for the EA in terms of 'equivalent full-time jobs'. From the one or several years preceding application for the EA, a base would be formed in which part-time jobs would be converted into fractions of equivalent full-time jobs. This will produce a number for equivalent full-time employment and the EA would be only available on employment over and above this employment base threshold. This method for calculating EA benefits takes away the incentive to break up full-time jobs into part-time ones.

Dissipating the EA's Effectiveness Through Wage Increases

Probably the most serious set of problems which might undermine the effectiveness of an EA in reducing underlying unemployment relates to wages. The lower initial unemployment generated by the EA might provoke generalised stronger wage pressures throughout the economy, and workers and unions might take advantage of the tighter labour market to press for higher wage growth. Indeed, after the introduction of an EA they might push for higher wages without waiting for any improvement in the labour market, so that the EA would be negated from its very inception. In either case, the benefits of the EA would be passed backward to the existing workers rather than retained by businesses to increase employment.

Such stronger wage pressures would either increase RULC towards its pre-EA levels, thus negating the EA, and/or substantially increase the growth rate of nominal wages (and thus inflation) so that restrictive macroeconomic policy measures had to be implemented, again negating the effects of an EA.

Pessimists might argue that such renewed wage pressures would be so strong as to fully negate the beneficial effects of an EA on employment. If so, underlying unemployment would not be reduced at all, in the long run, by an EA. These arguments are also often made in relation to cutting or abolishing payroll tax.(11)

The general point should be admitted here that schemes such as an EA are most applicable to situations where employment/ unemployment is very responsive to RULC but where RULC is in turn not very responsive to employment/ unemployment. An EA would not be very suitable to situations where the reverse conditions hold. However, it can be argued that a well-designed EA will be much more in the former category than the latter one.

For example, it can be persuasively argued that wage pressures would only be a small partial offset so that underlying unemployment could be sustainably reduced by an EA. Remember that in the case of classical unemployment (where RULC at the margin is too high) an EA creates extra labour demand for a labour supply that effectively already exists-ie. some of the unemployed want to work at going market wages but are unable to find jobs. This balancing of demand and supply does not seem to generate a situation of substantial excess demand as assumed by the pessimistic analysts who seem to have the cases of frictional and mismatch unemployment in their minds.(12)

The case of an EA of 10 per cent, analysed later in this paper, would seem to generate moderate enough effects on employment so that wage pressures were not a major problem.

It might also be noted here that cutting or abolishing payroll tax, also a way of reducing RULC, would seem to be more vulnerable to these problems of 'offsetting wage push' than an EA. The former measure applies directly to a broader group of workers than the EA. For example, payroll tax will apply to the entire workforce of many businesses. This broader coverage would thus seem to make it more directly vulnerable to problems of organised union wage push negating the tax change. Since an EA with an employment qualifying rule by its very nature applies only to workers at the margin of employment in most businesses, it is a much less visible and more difficult target for organised wage push than are broad-based policies such as changes to the payroll tax.(13)

In any event, it might be wise to incorporate some extra elements of tax-based income policies (TBIP) in an EA scheme.(14) Under TBIP schemes, businesses face higher tax rates when they allow nominal wage (or labour cost) growth to exceed some threshold rate chosen to achieve some inflation target. This acts as an disincentive to such growth in costs above the threshold. It would thus be useful to an EA scheme in discouraging the offsetting higher wage/ labour costs which could be argued to be generated by the higher employment generated by the EA.

For example, the EA could be withdrawn from businesses which increase their nominal unit wage/ labour costs (ie. money costs per unit of output) by more than the specified threshold. However, if it proved necessary to include them, such TBIP features would add somewhat to the administrative complexity and information requirements of an EA, especially in regard to the calculation of unit costs.

Dissipating the EA's Effectiveness Through Price Reductions

In a similar vein, it might be argued that an EA could be substantially or wholly negated by the output price falls which may be required to sell the extra output produced by higher employment levels. The benefits of the EA would be passed forward to buyers of output rather than retained by businesses to increase employment. Such price falls would tend to increase RULC at the margin (since output prices are part of the denominator in the calculation of RULC) and thus tend to bring them back to where they were before the introduction of an EA.

In response, it should be noted that many prices in the Australian economy are set on world markets, so extra output in such circumstances can be sold without any fall in price received. That is, in these industries output demand is very sensitive to price. In those industries and markets where output demand is not very sensitive to price there will be no requirement for price falls if demand can be increased in other ways. Thus, for example, macroeconomic policy might be used to expand aggregate demand by a moderate amount so that price falls are not necessary to sell the extra output in these sectors.

Introducing an EA can be thought of as increasing the aggregate productive capacity of the economy, so it is fitting that aggregate demand be expanded through macroeconomic policy relaxation to match the increased capacity to produce.

Inaccurate Targeting of an EA

Any EA which is specified to have one simple aggregate employment qualifying rule applying across the whole economy will necessarily subsidise employment growth in some businesses in which growth would have taken place anyway. In this sense, such an EA always has a 'deadweight waste' component, in that it will not be as precisely targeted as would be desirable. It will also mean that the EA is not available to encourage higher employment growth in businesses which would otherwise have little (or even negative) employment growth.

For example, for an EA which subsidises employment growth above say 2 per cent in all businesses, there will be a substantial number of prosperous firms which would have increased their employment levels by more than 2 per cent without an EA. As well, the restriction would prevent access to the EA by many firms in declining or static industries who could use the EA to raise their employment.

If this problem of sloppy targeting looms too large, one obvious solution is to vary the X per cent employment growth rule across industries in order to reduce this waste component. Thus, industries with a strong existing impetus for employment growth (eg. computer hardware and software) would have their employment growth qualifying rule set at, say, X=4 per cent, while industries in employment decline (eg. textiles and clothing) would have their employment growth qualifying rule set at, say, X= -4 per cent.(15)

Such variations reduce the waste component of the EA by reducing its coverage for employment growth which would have occurred anyway, and increase the number of firms which can access it to take on workers which they would otherwise not do. Businesses do this because they can utilise smaller variations in employment growth within industries compared to sectoral and national variations.

Indeed, the X per cent rule could be varied for each industry each year on the basis of forecasts for employment growth for that industry. Employment forecasts at the industry level are now regularly made by a range of private and public sector organisations. The most accurate of these could be utilised for setting the EA employment growth rule at the start of each financial year for each industry. Thus, if the computer software industry is forecast to increase aggregate employment by 5 per cent per business this coming year then its EA rule could also be set at 5 per cent. Of course, it would be important to ensure that the formulation of such forecasts was not open to manipulation by the industries themselves.

New businesses pose additional problems of targeting for an EA. By their nature, new businesses would be increasing their employment by more than their industry's average growth. Allowing access to the EA for all employees of new businesses is clearly very wasteful since the great bulk of these would have been taken on without an EA. However, some sort of access to the EA for new businesses is desirable to encourage even higher employment by such firms.

Perhaps the most reasonable solution is to allow access to the EA only when firms have reached some minimum employment size, with this minimum to be chosen in relation to each industry's level of average employment per firm.

How Long Should the EA Subsidy Last?

The optimum length of time of EA subsidy is just long enough so that the EA-assisted jobs continue to exist after the subsidy ends. Ongoing human and physical capital accumulation will boost labour productivity so that eventually these subsidised jobs will become profitable without the benefit of the subsidy, so long as wage increases do not eat up all of these improvements. Subsidised jobs will eventually 'stand on their own feet' and at this point the EA subsidy can be withdrawn.

In order to minimise the time period for the EA it would be very useful to freeze real wages for EA-subsidised jobs during the period that the EA applies. This would seem fair since these jobs would not exist without the EA in the first place. A reasonable quid pro quo for the EA would be for the holders of such subsidised jobs to accept constant real wages over the period in which the EA applies.

In terms of policy mechanisms to ensure that the real wages of such jobs do not increase over the period in which the EA applies, the elements of TBIP policies mentioned earlier could be used to help achieve this. Businesses which allowed real wages to increase for EA-subsidised positions could face withdrawal of the EA. More generally, they could face higher overall tax rates on profits in this situation. As well, it would be very useful for policy makers to communicate to the holders of such jobs that the EA will only last for a certain period of time, after which they will have to stand on their own feet and demonstrate to businesses that their jobs are profitable to retain in their own right. This would strongly encourage wage moderation by such job-holders.

It would also be wise to create special categories of EA-subsidised workers for the purposes of enterprise bargaining processes and compliance with the awards system. This would separate them from other workers, but only for the period in which the EA applies, in order to insulate them from general wage bargaining and real wage increases. Such separation would only be temporary for each cohort of subsidised jobs. However, this would require some major reforms to current and envisaged industrial relations systems in Australia.

These policy measures to contain real wage increases for EA-subsidised jobs might help to reduce the waste component entailed in the EA subsidising jobs which would have been created in any event. For example, businesses may wish to reduce the extent of their claims for the EA to minimise their exposure to the chances of higher tax rates entailed by TBIP policies, or the chances of losing the EA completely, due to too rapid wage increases for EA-subsidised jobs. This would help to confine the EA to those jobs actually dependent on the EA for their creation and ongoing existence.

An Illustrative Example

An EA of 10 per cent: The Effects on Employment and the Budget

For an EA with an aggregate employment qualifying rule, the net cost to the budget would depend on factors such as the rate of the concession, the effective tax rate on business income (for both companies and unincorporated businesses), the effectiveness of an EA in increasing aggregate employment, and in the coverage and waste component of the EA. Offsetting factors such as higher personal and business tax revenue and lower unemployment benefit payments arising from the higher employment generated should also be taken into account in assessing the total effect on the public sector's budget balance.(16)

Many of these parameters are subject to considerable uncertainty, so the calculation that follows must be viewed as only a very rough and approximate depiction of the economic and budgetary effects of an EA.

Let us examine an average 10 per cent EA available to businesses and public sector agencies on the costs of increasing their total employment over employment levels in the previous year, measured in 'full-time equivalent units'. There is an extra 10 per cent deduction, on average, for the costs of employing extra workers, with the percentage varying across the range of wage rates to give an average 10 per cent deduction. Here, the X per cent qualifying rule is set at X=0 across the economy since average numbers of employees per business has been roughly static in the last few years. If this average had been increasing by, say, 1 per cent before the EA then it would be reasonable to set the EA qualifying rule at 1 per cent.

As noted earlier, this is an economy-wide average; many businesses will have employment growth below this benchmark of 0 per cent (ie. they will be reducing their employment levels) and some new businesses will not have yet reached the minimum size levels set for access to the EA. Such firms thus do not qualify for the EA.

We have already noted that it is possible to design more complicated EA schemes in which the benchmark employment growth qualification varies across industries. Let us also assume a marginal rate of tax on business income of 30 per cent (this is an average of the company and unincorporated business sectors).

One crucial issue to be considered is the number of years for which the EA concession on a specific batch of extra employees is to apply. This needs to be long enough so that, by the end of the period, enough physical and human capital accumulation has taken place to ensure that those extra employees taken on due to the EA become absorbed into the natural expansion path of employment of the industry and economy and will not be retrenched once the EA for them ends. The EA concession for extra employees probably needs to last at least two to three financial years to ensure this. However, the length of this required period needs much more empirical examination.

For prospective employees above the benchmark growth for employment (in this EA, all extra employees) the 10 per cent EA with an average business tax rate of 30 per cent would reduce RULC for such workers by about 3 per cent (.10 x .30 = .03). Now, if we assume a reasonable elasticity (percentage magnitude) of response of say -0.70(17) for effects on aggregate employment, then we can calculate that employment in those firms qualifying for the EA will increase by about 2.1 per cent after this EA is introduced. This responsiveness will probably be larger in the medium to long term as there will be a stronger inducement to invest because such additional productive capital will now be more profitable than before.

Since such firms will be only a portion of the total number of firms the growth in national employment will be somewhat less than 2.1 per cent. If such EA-eligible firms are 50 per cent of all firms then national employment might increase by about 1 per cent (on top of the normal employment growth in the economy) after the introduction of an EA.(18)

Clearly, any EA should be designed so that the percentage of firms eligible for the concession should be as large as possible to maximise the employment effects. This effect of EA eligibility rules on this percentage is of vital importance and would require much closer empirical scrutiny. Our assumption of 50 per cent access for this simple aggregate EA seems reasonable, since it implies an approximately normal distribution about the known average for employment growth.

For the purposes of further calculation let us assume that aggregate employment increases by 1 per cent. In terms of current employment levels, this 1 per cent increase resulting from the EA would amount to about 80, 000 extra jobs, spread across both the private and public sectors. This is still quite an impressive result. Of course, a higher rate of EA would, on these calculations, generate a proportionately higher gain in national employment.

If we assume average labour costs of about $35, 000 per worker (the approximate current private sector average across both full-time and part-time workers), then this implies an extra labour cost bill of about $2.8 billion (80,000 x 35,000). For a 10 per cent EA and an average business tax rate of 30 per cent, this implies, in itself, a direct fall in business tax revenue of about $85 million (2.8 billion x .10 x .30). If the EA applies for three years this implies an annual revenue loss of about $250 million.

It was noted earlier that simple aggregate EAs such as this scheme have an inherent component of 'waste' in the sense that it is inevitable that the EA will be given for some employment growth which would have taken place in any event. Many rapidly growing businesses will be increasing their employment levels even without an EA. We need to add in this cost to revenue in our calculations. This is a difficult area since we do not have comprehensive data, for the economy as a whole, on firms changing their employment in any year.

Let us make the reasonable assumption that such 'pre-existing' employment growth in eligible businesses and public sector agencies-which would have occurred anyway-will be about 5 per cent of total employment each year.(19) This means that the direct cost to revenue of this illustrative EA scheme would increase to about $1.5 billion per year ($250 million x 6).

In a full costing of this proposal it is also important to add in the indirect gains to the Budget arising from an EA.

  • First, total business profits will be higher as a result of the introduction of an EA, thus tending to boost business net tax revenue in itself. Let us assume that the ratio of wages to business profits (for companies and unincorporated enterprises) in the national accounts remains constant at about 2:1. Let us also assume that the $2.8 billion in labour costs translates into about $2 billion in extra wages and salaries. This implies that business profits would increase by about $1 billion and would generate extra business tax revenue of about $ 300 million (1.0 x .30). Here, we are including unincorporated enterprises with companies even though the former actually pay their tax through the personal tax system.
  • Second, personal income tax revenue will be higher. As above, let us assume that about $2 billion in extra personal taxable income is generated (for wage and salary earners). If we then assume an average marginal personal tax rate of 25 per cent then an extra $500 million in personal tax revenue is generated.
  • Third, aggregate unemployment benefit payments will be lower. If we assume that these 80, 000 extra jobs reduce the number of unemployment benefit recipients by 50, 000 (the other 30, 000 jobs being taken by non-recipients) and that the average annual unemployment benefit payment is $8, 000, then the reduction in spending on unemployment benefits is $400 million (50, 000 x 8, 000).
  • As well, the rise in aggregate household income generated by the higher employment will in turn generate some increase in consumption spending. If allowed to go on, this will generate further rounds of income increases and higher consumption and investment spending. All of this extra activity will raise extra tax revenue through the indirect and direct tax systems. Let us add in one round of tax raised from household consumption spending. Assuming a marginal propensity to consume of 0.9 and an average indirect tax rate of say 11 per cent, this will generate about $200 million in extra revenue ($2 billion x .9 x .11).

However, this multiplier process assumes that there is some cyclical slack (idle capacity) in the economy which can be whittled back. To the extent that this is limited at the time of implementation of an EA then these Keynesian processes will be restricted by counter-cyclical macroeconomic policy designed to prevent the emergence of excess demand and accelerating inflation. Although it was argued earlier that some expansion in aggregate demand should accompany the introduction of an EA, aggregate demand should not be expanded so much that excess demand and inflationary pressures are also generated.

Overall, these calculations produce the somewhat surprising result that the net impact on the Commonwealth Budget from the introduction of an EA of 10 per cent could be close to zero, unless the waste factor is larger than expected. Here, the EA would be very largely self-financing. If the required length of time for the subsidy was two years rather than three, then this EA would actually generate net revenue for the Commonwealth Budget.(20)

However, if the 'waste element' of pre-existing employment growth in eligible firms is substantially higher than expected, and the required length of time for the subsidy was three years, then such an EA will very likely generate a net move towards deficit in the Budget and extra financing options will then need to be considered. On the other hand, for a required subsidy time length of two years, a substantial waste element of say 10 per cent of total employment will generate a relatively small net revenue loss of about $400 million.

Of course, it is also important to point out that these calculations assume the most optimistic case of no offsetting factors which might tend to push RULC back to its previous levels. Factors such as renewed real wage push in response to lower unemployment or the wage and price effects of indirect tax increases used to fund the EA, are discussed below. Reasonable rates of business eligibility for the EA and a reasonable timelength to convert EA-subsidised employees/ jobs into profitable ones are also assumed.

On the other hand, the use of eligibility rules that vary across industries and employ good forecasting techniques for industry employment hold out the prospect of substantially increasing the percentage of firms eligible for the EA, and of considerably reducing the waste component of the EA. This would increase the employment impact and reduce the budgetary costs of an EA.

Financing an EA

If the above example is at all plausible and realistic then an EA could well be self-financing (and possibly generate net revenue for the budget) and little further explicit consideration need be given to how it might be financed. However, for the sake of completeness and in case the waste component is large and some portion of an EA does need external financing, let us consider how this financing issue affects our evaluation of an EA.

In such a case, the prospects of an EA reducing unemployment will depend upon the way in which an EA is financed. The various financing options involve expenditure reductions, increases in direct taxes or indirect taxes or some combination of these. There is also the option of larger budget deficits and partial financing through higher public debt levels.

In regard to the options involving no permanent increase in the budget deficit, it would seem that the effects of an EA in reducing unemployment would be largest through financing by expenditure reductions and personal tax increases, and somewhat less in the case of increases in indirect taxes and company taxes.(21)

In the case of indirect taxes, there would be substantial offsetting factors which would tend to push RULC back towards its original setting. Nominal wages would tend to rise in consequence of the higher after-tax prices of goods and services while prices in the hands of employers, their after-tax receipts, would tend to fall somewhat as they bore some of the burden of the higher indirect taxes.(22)

Both economic effects would tend to push RULC back towards its previous levels. However, even in this case some substantial fall in unemployment could still result if RULC was reduced in a permanent and sustainable way.

In the cases of personal tax increases and spending cuts these offsetting factors will be much smaller in magnitude, especially where classical (real-wage based) unemployment is substantial. For example, wages do not seem to be very responsive to changes in personal income tax rates (at least in comparison to changes in indirect taxes) in the current Australian situation where substantial unemployment exists.

Reliance upon personal tax increases to fund an EA would be much more effective than reliance upon company tax increases because of the latter's substantial impact upon employment and investment by companies.

Any short term effects on demand and output of any of these fiscal policy packages could be accompanied by a monetary policy adjustment (eg. lower official interest rates if the package dampens aggregate demand and economic activity) to ensure robust aggregate demand, macroeconomic stability and a smooth transition to the higher employment outcomes.

Comparisons with the Blandy Scheme

In 1993, in a submission to the Committee on Employment Opportunities, Professor Richard Blandy put forward a plan for a marginal wage subsidy scheme which has some similarities with the EA scheme examined here.(23) As with the EA, the Blandy scheme proposed that the employment of additional workers be subsidised so as to reduce the real unit costs of doing so, in order to increase aggregate employment and thus reduce unemployment. However, in contrast to the EA, Blandy proposed that

  • such subsidies be explicitly financed through a special tax levy on the earnings of existing workers,
  • workers be given the choice of paying such taxes to the Federal Government or to the business in which they work (Blandy predicted that the great bulk of workers would choose to contribute such funds to their own firm),
  • the entire costs of additional workers be subsidised so that such workers become 'free resources' to businesses, and
  • these subsidies be only temporarily utilised until aggregate unemployment is reduced to some designated 'full employment' level, after which time the subsidies will be scaled back and then abolished.

The Committee's Discussion Paper referred to the Blandy plan in only one very brief endnote.(24) It argued that such a plan would only generate a massive form of jobsharing in which businesses would substitute reduced work hours of the existing employed for increased numbers of workers so that aggregate work hours and output would change very little. This would occur because the wage subsidy would not cover the costs of additional materials and overheads; and these costs constraints would prevent any substantial increase in aggregate production and work hours. The Committee also pointed to problems of policing the requirement that these funds be only spent by businesses on employing additional workers.

The Committee's critique of the scheme on dissipation through jobsharing effects does seem true in one respect. As noted above with the EA, businesses would resort to such substitutions unless restrained in some way. This paper proposes that base levels of 'full-time equivalent employment' be established for each business and that only employment above and beyond this base be subsidised. Such a rule would also be useful in the case of the Blandy scheme and would orient businesses towards increasing their aggregate labour inputs into production.

However, the Committee's argument that costs of materials and overheads might prevent increases in output and in aggregate work hours seems very weak, especially under current economic circumstances. Businesses would expand output and work hours if they expected it to be profitable, and it is the purpose of marginal wage subsidy schemes to make it so. Once extra profits are expected, businesses would finance 'working capital' requirements of materials and overheads through the use of trade credit and existing financial assets; thus, these costs do not seem to constitute any substantial barrier to expansions in output and employment.

It is also very interesting to note that the Blandy plan's feature of giving workers the choice of paying the levy to the Federal Government or to their business of current employment would almost certainly mean that at least part of the levy would be paid by employers and that this would mean that something less than 100 per cent of the costs of employing extra workers would be actually be financed by the jobs levy. Blandy argues, quite persuasively, that this would come about because employers would seek to entice their workers to lodge their jobs levy funds with them by offering to pay part of the levy if they opted to keep such funds within the business. This feature of his scheme makes it considerably closer in nature to the operation of an EA scheme.

However, there are several remaining points of difference which indicate that EA schemes are to be preferred to the Blandy scheme. First, under the EA scheme policy makers have much more direct influence on the quantity of new jobs created through their setting of the rate of the EA which is to apply to each extra worker employed (eg. in the worked example analysed above an EA of 10 per cent, on average across the wage spectrum). Under the Blandy scheme, policy makers directly influence the quantity of funds raised through their setting of the rate of the jobs levy, but they do not seem to have any equivalent lever on the quantity of job creation except for the rule that such funds must be used for the employment of extra workers.

Blandy does include a proviso that each participating business is to be compelled to increase its workforce by at least the increase which would occur if all extra workers were taken on at average wage levels for the business. However, he does not specify a mechanism for bringing this about and seems to imply the use of very heavy-handed direct controls and sanctions by the ATO to achieve the desired pattern of job creation. In contrast, the EA provides incentives for the employment of substantial numbers of extra workers, with the strongest incentives for employing low-wage workers, without the need for direct controls.

Second, the Blandy scheme is founded upon a quite fixed, inflexible funding source of proportional taxes on workers (and businesses). The EA's reliance upon the general taxation system for funding allows scope for much more diversified and progressive funding patterns. Of course, for both schemes it needs to be mentioned that considerable financing will be raised through the higher employment and national income which can be generated by the schemes, so that any residual funding sources might be a quite secondary issue of importance.

Third, Blandy himself argues that his scheme would do little to keep the economy at full employment once it had arrived there since, in his view, it would not generate ongoing increases in employment. The logic of this argument seems rather obscure since he explicitly frames his scheme as one which will come into use where necessary to stabilise the economy at whatever rate of unemployment is regarded as full employment.

In any event, it could be argued that EA schemes will automatically help to bolster employment, without the need for conscious adjustment to do so as in the case of the Blandy scheme, because they will be always present and acting at the margin to encourage businesses to increase their employment levels.

Perverse Equity Effects of an EA?

Even an EA with an aggregate employment qualifying rule might have some perverse and unintended effects if not further constrained by other rules. For example, an EA could be viewed as giving more encouragement to the employment of new workers on high wages than to those on low wages. An EA of 10 per cent entails an extra tax deduction of 10 per cent for the costs of employment of a new worker. Clearly, the extra deduction would be far larger for a new worker on $50, 000 a year than for one on $20, 000 a year. Thus, the incentive provided to take on new workers (at least in absolute dollar terms) would, other things being equal, be larger the higher the wage of new workers.

This might seem to be the opposite of the incentive pattern which would be recommended by distributional and social equity concerns.

On the other hand, the percentage reduction in labour costs arising from a uniform EA will be the same for all workers across the wage spectrum. It is usually argued that the percentage reduction will be the better guide to enhanced incentives for higher employment of each wage category of workers.

In any event, if this incentive structure of a uniform EA is considered to be a major problem it might be avoided by setting the EA tax deduction at a higher rate for low-wage new employees. The EA rate could be manipulated so that it offered the same dollar value of tax relief for each worker, regardless of wage levels. Alternatively, an absolute dollar limit on the value of the extra EA deduction could used. Such weightings and limits could skew the incentives to take on additional workers towards low-wage employees.(25)

Both of these solutions were earlier canvassed as ways of combating abuse of an EA through overclaiming, so they can serve two useful purposes in the scheme. As with the aggregate employment qualifying rule, they would add somewhat to the costs of administering the EA and would require careful explanation to businesses so that the intended incentive effects were clearly understood by them. However, the beneficial social equity effects of encouraging more employment amongst low-wage workers would be substantial and valuable.

Perverse Pro-Cyclical Effects of an EA ?

If an EA with an aggregate employment qualifying rule has a positive effect on the Budget outcome, then it will tend to have desirable anti-cyclical effects during a recession. That is, during recessions when aggregate employment falls the number of firms claiming the EA would fall dramatically, thus reducing its impact on the Budget and tending to push the Budget outcome towards deficit. This is exactly what would be needed and would add to the strength of the other 'automatic fiscal policy stabilisers' such as the general tax revenue system and the unemployment benefits system operating to dampen the recession by bolstering private sector incomes.

However, an EA with a total employment qualifying rule which itself moves the Budget outcome towards deficit could have perverse effects in times of recession. During recessions employment growth throughout the economy tends to fall and thus the number of firms meeting the qualifying limit for the EA would again fall, thus reducing the total value of the EA tax concession and the net revenue forgone. This would tend to push the budget towards surplus at such times whereas automatic stabilisers should act to move the budget towards deficit to help to dampen the recession.

In this case the problem might be avoided by deliberately increasing the rate of the EA deduction (from, say, 10 per cent to 20 per cent) during recessions to encourage employment and to help to dampen the recession, and then dropping it back during times of economic recovery.(26) This variation of EA rates might be desirable, on counter-cyclical grounds, even in the case of EA schemes which improve the net budget balance.

Tax System Complexity

The Federal Government and the ATO are currently in the midst of a project to simplify the direct tax system to make it easier for taxpayers to understand it and comply with it. The question could be raised as to whether an EA would undermine this project by moving the business tax system back in the direction of more complexity.

The answer to this question is a qualified yes, but the costs of greater complexity would seem to be acceptable and tolerable if an EA could generate substantial and sustainable increases in employment and reductions in unemployment. In any event, the additional complexity would be minimised if the existing administrative machinery such as the Superannuation Guarantee compliance and matching system is used.

Does an EA Fit with Existing Active Labour Market Programs?

Overall, it would seem that an EA would complement many of the elements of existing Active Labour Market Programs (ALMPs).(27) The latter are essentially about improving labour market supply and overall labour market efficiency by increasing the skills and attractiveness to employers of the long-term unemployed (LTU). In contrast to an EA, labour market substitution (ie. getting subsidised long-term unemployed in to jobs, even at the expense of other workers) seems to be regarded as a more acceptable part of ALMPs.

An EA reduces labour costs, boosts labour demand and thus helps to bring the ALMPs to successful fruition by helping to create permanent new jobs for the unemployed who have been empowered and made 'job ready' by the ALMPs.

It is also possible to conceive of an EA which included specific help for the long-term unemployed. A higher rate of EA deduction could be allowed for the long-term unemployed as an extra incentive to take such people into employment (say 30 per cent compared to the standard EA rate of 10 per cent). This extra deduction could last for say 1 year, after which the standard EA rate would apply. In this case the wage subsidy components of current ALMPs could be abolished and replaced by this higher EA allowance for employing the LTU.

Are There Better Alternatives for Increasing Sustainable Employment ?

Let us now compare an EA using an aggregate employment qualifying rule with other means of encouraging employment, such as payroll tax cuts, general wage subsidies and labour market deregulation.

It would seem that the EA is superior to payroll tax cuts since the latter applies to both workers currently employed and those who would be drawn into employment by such tax cuts, whereas the EA with an employment qualifying rule is targeted specifically at additional workers at the margin. As well, many small businesses are already exempt from payroll tax because of the concessionary thresholds used by the state governments in collecting payroll tax.

Thus, the EA would achieve more in terms of employment (at least in the short to medium term) than the revenue equivalent payroll tax cut because the EA is better targeted to impact at the margin of current employment levels. Additionally, it would seem to be somewhat less vulnerable to generalised wage push than the payroll tax cut option because of its tighter focus and better targeting to influence behaviour at the margin.

In terms of general wage subsidy programs, it would be possible to design a marginal subsidy program which was effectively equivalent to the EA discussed above. However, it would seem that an EA would involve fewer administrative, monitoring and compliance costs than a wage subsidy scheme because the former would build upon the existing administrative structures of the business tax/ SG system and could thus use this system's valuable features of comparatively high quality data and substantial penalties for abuse, avoidance and evasion.

On the other hand, reduced tax liabilities through an EA will lag actual additional employment by at least one financial year because the business tax system does not settle such liabilities until the following financial year. Since the lag for cash payments would presumably be much less (possibly just a few weeks or months) it could be argued that such payments will have a quicker impact upon aggregate employment in the year or so after introduction.

This issue turns on the farsightedness of the relevant businesses and more empirical information is required for proper evaluation. Conversion of the EA to explicit marginal wage cash payments for new businesses experiencing operating losses by making the EA tax refundable, as was mentioned in an earlier section, might be even more highly desirable for this very reason.

Extensive labour market deregulation to increase competitive pressures in the Australian labour market and thus to lower RULC would also be effective in raising employment and lowering underlying unemployment. However, a central problem here is that deregulation could contribute to income inequality which is far greater than under the EA because of the increased wage inequality that it generates. As well, it would probably, in and of itself, reduce the level of real earnings of low wage and salary earners.

For these reasons this policy course would face very serious political barriers and obstacles. It also raises philosophical questions about the morality of squeezing those who are already at the bottom of the socioeconomic ladder, and who have already suffered real income reductions in recent years.

On the other hand, combining extensive labour market deregulation with tax-transfer compensation schemes for low wage and salary earners to ensure that their disposable incomes did not fall (or even to ensure that they increase) in the process might be both technically and politically attractive. For example, reduced market wage rates for low wage and salary earners could be combined with an extensive Earned Income Tax Rebate (EITR) Scheme, which actually made substantial cash payments to low wage workers.(28) This policy package could be very attractive and useful in reducing unemployment while not allowing any increase in inequality on the distribution of disposable incomes.(29)

Indeed, such schemes could use the extra national income generated by the rise in national employment and fall in unemployment to actually increase the real disposable income levels of low wage and salary earners.


Sadly, it must be admitted that unusual schemes such as the EA would face ingrained dislike from many political and ideological viewpoints. There will be those who dislike, and feel uncomfortable about, this scheme's use of the argument linking labour costs with employment and unemployment, but who will feel much more positively about the interventionist nature of an EA and its potential deployment in place of extensive labour market deregulation. Others will be uncomfortable about this interventionist nature of the scheme and thus will much prefer extensive labour market deregulation, but will welcome the use of the argument linking employment and unemployment outcomes to the labour costs facing business.

Thus, an EA scheme cuts across traditional political and ideological categories and is likely to receive a 'lukewarm' reaction for this reason alone, regardless of its inherent merits and faults. However, on the above analysis, a well-designed EA holds out the prospect of being able both to increase national employment and to reduce income inequality and poverty levels, without reducing real wages. This characteristic makes it a very attractive option.

The main problems of an EA revolve around the extra information and anti-abuse measures (and possibly, measures to prevent dissipation through wage increases) required to make such a scheme work effectively and efficiently. There would also be transitional problems about the scheme's effectiveness since businesses (and especially their managers in charge of employment decisions) would only gradually learn that the EA scheme was available to encourage extra employment at the margin.(30) The financial costs of an advertising campaign, which would need to be initially extensive, to publicise the scheme (especially to these crucial employment managers) would need to be taken into account in considering its implementation. Such schemes do not seem to have been implemented on any national scale in the advanced world before now, so there would always be the danger of some unanticipated fatal flaw arising which could not be adequately dealt with without fundamentally altering the scheme. This is always the case with policy innovations.

This EA scheme fits well with the 'conventional wisdom' of many economists that the tax-transfer system should be used to address income distribution issues rather than this being addressed directly through the wage-setting mechanisms of the industrial relations/ awards system. The EA scheme does this in a way which does not generate any disincentive to work, which seems to be a central problem with many types of transfer payments, because it operates through businesses to generate higher employment and more jobs and not through direct payments to individuals and families as in EITR schemes.

EITR schemes also address distributional issues through the tax-transfer system but they suffer from problems of discouraging work effort for certain wage classes of workers. These disincentive effects are much reduced (but not eliminated) in universal EITR schemes which apply to all workers. However, the latter types of EITR are, in consequence, very expensive to the budget and are thus far more difficult to finance.


  1. Johnson, G.E. and Layard, P.R.G. 'The Natural Rate of Unemployment: Explanation and Policy' in Orley Ashenfelter and Richard Layard (eds) Handbook of Labour Economics, volume 2. Amsterdam: North Holland, 1986: 954-956; and Layard, Richard, Nickell, Stephen and Jackman, Richard. Unemployment: Macroeconomic Performance and the Labour Market. Oxford: Oxford University Press, 1991: 490-492. During the second half of the 1970s Britain operated a marginal wage subsidy scheme for manufacturing firms in depressed areas. Retrospective evaluation indicated that it had a significant positive impact on employment in eligible firms: Layard, Richard. 'The Costs and Benefits of Selective Employment Policies: the British Case'. British Journal of Industrial Relations, 17(2), 1979: 193-194. The United States also operated a similar scheme, the New Jobs Tax Credit Programme, at the same time: Haveman, Robert. 'Reducing Poverty While Increasing Employment: A Primer on Alternative Strategies, and a Blueprint'. OECD Economic Studies, no. 26, 1996/ 1: 37.
  2. Economic Planning and Advisory Commission (EPAC). Future Labour Market Issues for Australia. Canberra: Australian Government Publishing Service, July 1996: chapter 4.
  3. EPAC (Future Labour Market Issues for Australia. Canberra: Australian Government Publishing Service, 1996: 60) uses the term 'structural unemployment' to cover both mismatch and classical unemployment.
  4. RULC is calculated by adding nominal labour on-costs such as payroll tax and superannuation contributions to nominal wages and salaries, and then dividing first by some measure of output prices and then by some measure of labour productivity.
  5. For the most recent results, see: Phipps, A.J. and Sheen, J.R. 'Macroeconomic Policy and Employment Growth in Australia'. Australian Economic Review, no. 109. 1st quarter, 1995: 86-104. Phipps and Sheen actually report results for the effects on employment of real wages, and not RULC per se.Other notable Australian studies which report statistically significant effects for either real wages or RULC include: Russell, Bill and Tease, Warren. 'Employment, Output and Real Wages'. Economic Record, no. 196, March 1990: 34-45; Pissarides, Christopher. 'Real Wages and Unemployment in Australia'. Economica, no. 229, February, 1991: 35-55; Ooi, Soon Huay and Groenewold, Nicolaas. 'The Causes of Unemployment in Australia, 1966-1987'. Australian Economic Papers, no. 58, June 1992: 77-93; Dao, Dan. 'A Model of Unemployment' in Committee on Employment Opportunities. Restoring Full Employment. Canberra: Australian Government Publishing Service, December 1993. Background Papers: 155-162.
  6. Sloan, Judy and Wooden, Mark. 'Labour Market Programs' in Freebairn, John, Porter, Michael and Walsh, Cliff (eds) Spending and Taxing: Australian Reform Options. Sydney: Allen and Unwin, 1987: 152-153.
  7. I thank Bernard Pulle for pointing this out to me.
  8. Lindbeck, Assar. 'The Welfare State and the Employment Problem'. American Economic Review (Papers and Proceedings), 84(2), May 1994: 74.
  9. I thank Chris Field for pointing this out to me.
  10. Fittoussi, Jean-Paul. 'Wage Distribution and Unemployment: the French Experience'. American Economic Review (Papers and Proceedings), 84(2), May 1994: 63. The British scheme of the 1970s also used a flat level of subsidy payment (see endnote 1 above).
  11. The important point should be noted that such long run analyses assume away unemployment and any changes therein: Freebairn, John. 'The GST and Payroll Tax Abolition' in John Head (ed) Fightback! An Economic Assessment. Sydney: Australian Tax Research Foundation, 1993: 97.
  12. Indeed, there are a number of econometric studies on Australia which report that unemployment levels have very little (or no) effect on nominal and real wage outcomes. For example, see: Mitchell, William. 'The NAIRU, Structural Imbalance and the Macroequilibrium Unemployment Rate'. Australian Economic Papers, no. 48, June 1987: 101-118; Watts, Martin and Mitchell, William. 'Australian Wage Inflation; Real Wage Resistance, Hysteresis and Incomes Policy, 1968(3)-1988(3)'. The Manchester School, 58(2), June 1990: 142-164; Dawkins, Peter and Wooden, Mark. 'Labour Utilisation and Wage Inflation in Australia: An Empirical Examination' Economic Record, no. 173, June 1985: 516-521. However, there are also a number of studies which find statistically significant effects: Simes, R.M. and Richardson,C.J. 'Wage Determination in Australia'. Economic Record, no. 181, June 1987:144-155; Ooi, Soon Huay and Groenewold, Nicolaas. 'The Causes of Unemployment in Australia'. Australian Economic Papers,. no. 58, June 1992: 85. Some of these studies would seem to be suffering from a number of methodological flaws which cast doubt over their results. For example, surely the relevant variable for regression analysis of effects on wages is not unemployment per se but the deviation of actual unemployment from the current NAIRU level with the latter being able to vary over time. Only the 1987 Mitchell paper and the 1987 Simes and Richardson paper examine the issue in this way. The divergent results of these two papers is somewhat mysterious; the first found the deviation variable statistically significant but the second did not, although the test was close in the latter and in both studies it had the expected sign. These differences may be partly due to differences in their (relatively crude in both cases) calculation of the NAIRU variable. These empirical results are relevant in that we are arguing that an EA can reduce both the NAIRU and actual unemployment without increasing the deviation between them, rather than just increasing the deviation by pushing unemployment temporarily below some predetermined NAIRU.
  13. In any event, it is interesting to note that there exists no Australian statistical evidence to support the argument that the burden of the payroll tax is borne by workers through lower real wages. Indeed, there is statistical evidence to the contrary, albeit of only of a short run nature, that it does not generate real wage adjustments: Ooi, Soon Huay and Groenewold, Nicolaas. 'The Causes of Unemployment in Australia 1966-1987'. Australian Economic Papers, no. 58, June 1992: equation 17; Lewis, Philip and Kirby, Michael. 'The Impact of Incomes Policy on Aggregate Wage Determination in Australia'. Economic Record, no. 181, June 1987: 159.
  14. Layard, Richard, Nickel, Stephen and Jackman, Richard. Unemployment: Macroeconomic Performance and the Labour Market. Oxford: Oxford University Press, 1991: 485-490.
  15. However, it could be argued that a universal lower limit of 0% job growth should be set for the EA in order not to delay structural change which is generating job losses in industries in 'natural decline'. I thank Geoff Winter for pointing this out to me.
  16. Piggott, John and Chapman, Bruce. 'Costing the Job Compact'. Economic Record, no. 215, December 1995: 313-328.
  17. Phipps, A.J. and Sheen, J.R. 'Macroeconomic Policy and Employment Growth in Australia'. Australian Economic Review, no. 109, 1st quarter, 1995: 95; and more generally, EPAC, Future Labour Market Issues for Australia. Canberra: Australian Government Publishing Service, July 1996: 17-18. A question can be validly raised as to whether it is appropriate to use an elasticity estimated from data on general averages for real wages and RULC in an examination of the effects of changes in real wages and RULC at the margin only. Since the relevant econometric studies control for the effects of many other variables on employment and unemployment, we can be confident that the estimated elasticities measure the effect of changes in real wages or RULC at the margin, for other determining factors held constant. This is what our calculation requires. As well, we know that it is changes in costs at the margin which is needed, for other factors held constant, to affect business behaviour. However, it is also clear that general changes in averages for real wages and RULC will have larger long run total elasticities of effect than for marginal changes when we add up all the channels of influence. For example, general changes will have much larger effects on total business profits than marginal changes, and this will feed into stronger investment effects through the influence of financial/ cash flow constraints on business investment plans. A higher capital stock will in turn encourage more employment.
  18. A reviewer has argued that account needs to be taken of the reduction in employment by unsubsidised businesses as subsidised businesses use the EA to reduce costs and prices and thus to take customers away from the unsubsidised. However, this jobs transfer or 'displacement effect' (which will very probably be small) does not seem to have any implications for aggregate employment effects of an EA, precisely because it is just a transfer between businesses and will be 'netted out' in terms of the national economy. The employment effects in the worked example can be thought of as net increases, over and above any displacement effect.
  19. This parameter requires very careful empirical study since it will have a major effect on the financial viability of an EA scheme. Data sources are limited but estimates for the manufacturing sector are available: Borland, Jeff and Home, Richard. 'Establishment-Level Employment in Manufacturing Industry: Is Small Really Beautiful?'. Australian Bulletin of Labour, 20(2), June 1994: Table 1. Borland and Home estimate that this job creation rate was about 16% over 1984-85. However, more recent work by Borland for the period 1978-79 to 1991-92 shows an average job creation rate for manufacturing of just 2.3%: Borland, Jeff. 'Job Creation and Job Destruction in Manufacturing Industry in Australia'. Economic Record, no. 216, March, 1996: Table 1. Now, if we adjust for factors such as that the job creation rate would probably be lower in the public sector, mining and agriculture, but probably higher in the services sector, that national job creation and destruction rates will be typically be lower now, and that various eligibility constraints in an EA will reduce the deadweight loss factor (eg. minimum size rules for eligibility of new businesses), it would seem reasonable to argue that a waste factor of approximately 5% will be applicable. This estimate is supported by recent calculations by the Industry Commission on rates of aggregate job creation and destruction in Australia, using industry-level data (rather than firm-level data): Industry Commission. 1995-96 Annual Report. Canberra: Australian Government Publishing Service: Box 2.1, page 16.
  20. Similar fiscal results have been reported, on the basis of plausible assumptions, in the case of policies to improve the 'job readiness' of the long term unemployed: Piggott, John and Chapman, Bruce. 'Costing the Job Compact'. Economic Record, no. 215, December 1995: 313-328. For a contrary and critical perspective on such policies, see: Kenyon Peter. 'The Job Compact: What Does the International Evidence on Active Labour Market Policies Suggest about the Likelihood of its Success?'. Australian Bulletin of Labour, 20(4). December 1994: 272-297.
  21. Chapman, Ross and Vincent, David. 'Payroll Taxes in Australia, Part Two: An Economy-Wide Approach to Estimating the Effects of Their Removal'. Economic Analysis and Policy, 17(2), September 1987: 149-177. Chapman and Vincent's simulations using the ORANI model indicate that a direct tax surcharge to fund the abolition of payroll tax will have larger positive employment effects than funding through government spending reductions, and still larger effects than for indirect tax increases. However, in the case of government spending reductions they do not seem to have modelled the possibility of combining this with some relaxation of monetary policy in order to bolster aggregate demand. The latter case would probably turn out to produce the largest employment effects of all the scenarios. For a critical perspective on their results see: Ryan, Mathew. What Future for Payroll Taxes in Australia? Canberra: Treasury Research Paper. no. 10, September 1995.
  22. Freebairn, John. 'The GST and Payroll Tax Abolition' in John Head (ed) Fightback! An Economic Assessment. Sydney: Australian Tax Research Foundation, 1993: 106-108.
  23. Blandy, Richard. 'Submission to the Committee on Employment Opportunities', October 1993. See also his article in The Australian, 13 October 1993. A good discussion of the Blandy scheme can also be found in Kenyon, Peter. 'Restoring Full Employment: Backing an Outsider'. Australian Economic Review, no. 105, 1st quarter, 1994: 43.
  24. Committee on Employment Opportunities. Restoring Full Employment: A Discussion Paper. Canberra: Australian Government Publishing Service. December, 1993: endnote 13, page 72.
  25. Alternatively, wage subsidy schemes exclusively targeted at low wage earners have recently been proposed: Pencavel, John. 'British Unemployment: Letter from America'. Economic Journal, no. 104, May 1994: 630-631; Phelps, Edmund. 'Low-Wage Employment Subsidies versus the Welfare State'. American Economic Review (Papers and Proceedings), 84(2), May 1994: 54-58.
  26. This counter-cyclical role for wage subsidies is also a central part of the Blandy scheme.
  27. Active Labour Market Policies are those which actively seek to help the unemployed to find jobs, as opposed to the passive alternative of just paying unemployment benefits to them.
  28. Hanratty, Phil. 'Helping the Working Poor: An Earned Income Tax Rebate for Australia?'. Parliament of Australia, Parliamentary Research Service, Research Note, no. 17, November 1996.
  29. Alternatively, an EITR might be combined with an EA scheme because these are complementary ways of raising the employment, after-tax incomes and living standards of the working poor and the unemployed. For a relevant discussion, see: Haveman, Robert. 'Reducing Poverty While Increasing Employment: A Primer on Alternative Strategies, and a Blueprint', OECD Economic Studies, no. 26 1996/1: 7-42.
  30. I thank Mark Wooden for pointing this out to me.