Government Kyoto Research Programme Critique

Government Research Programme

A Critique

Report to the Greenhouse Policy Coalition

by

NZIER

December 2001

NZ INSTITUTE OF ECONOMIC RESEARCH (INC.)

8 Halswell St. Thorndon

P O BOX 3479 WELLINGTON

Tel: (04) 472 1880

Fax: (04) 472 1211

1. INTRODUCTION

As part of its consultations on the ratification of the Kyoto Protocol, the Government has released a number of studies to support its contention that the economic impacts on New Zealand would be minimal, and could possibly even be positive. While the officials’ summary of this research cautions against its limitations, the Government’s public statements have relied on the results.

Four studies on the Kyoto Protocol have been prepared for the New Zealand government over the last two years. Two were conducted by Australian institutions - the ABARE (Australian Bureau of Agricultural Economics) study released in November 2001 and one by the Centre for International Economics in Canberra, (CIE) done a few years back. The other two studies were carried out by New Zealand consulting firms (Infometrics and PA Consulting). ABARE

All four studies were commissioned with the same general focus - an overall economic assessment. This raises the issue of what important economic research was not commissioned. The computable general equilibrium models used in three of the studies have a voracious appetite for a wide range of other research results before they can be useful - that underpinning research does not appear to have been done. The research strategy adopted by the government raises issues concerning the types of questions these studies were capable of answering.

This critique reviews the government’s overall research strategy as well as considers the results of the individual studies. In our view, the research available to the government does not equip it to make well-informed policy decisions.

2. POLICY RESEARCH STRATEGY

It has long been obvious that the Rio Summit and the Kyoto Protocol were going to have important impacts on the New Zealand economy if New Zealand governments were committed to implementing either the spirit of the agreements or sign formal treaties in the area. This is because New Zealand has had cheap energy initially based on hydropower over a long period of time. Other energy forms like coal, gas and fossil fuels have had to be competitively priced. Industrial development has accordingly proceeded along energy intensive lines, with much of it emitting significant quantities of greenhouse gases (carbon dioxide). New Zealand has also developed a strong comparative advantage in animal proteins that are based on methane and nitrous oxide emitting animals.

It has been nearly ten years since those international discussions were held. People have taken positions and various political stances have become almost “hardwired”. The economic policy research results are only now coming to hand. Economic policy research on climate change policy in New Zealand is almost too late for the ratification debate. As an experienced editorial writer once said, “New Zealanders don’t want to discuss things, they want to take sides!" Well-timed policy research can help produce good government.

When it comes to economic research, there are two reasonably distinct questions from a New Zealand viewpoint. The first question concerns the world market in tradable emissions permits; “what is the international emissions permit price likely to be after 2008?” From an economic modelling perspective the Kyoto Protocol will involve the invention of a new “product”; one that economies have never seen before. Put crudely, there is no historic data out there on Kyoto issues. The modellers difficult task is to estimate how for example, an increase in the cost of owning a sheep will change the number of sheep we have or how will a carbon subsidy change tree plantings?

Greenhouse gas emissions permits will be exported by some countries and imported by others. The question is what will their price be? To answer this question we need the best model of the world market for emissions. The best model is the one which best reflects the structure and behaviour of the greenhouse gas “superpowers”; basically the large countries - the US, EU, Japan, Russia etc. In short the model has got to be international. The GTAP model used by ABARE was developed at Purdue University and modified to deal with greenhouse issues in an Australian industrial context. The ABARE model mirrors the structure of the large countries arguably as well as anybody else does.

There are many authoritative studies (at least 12) of the above question that were available before these studies were commissioned. Some of these studies used GTAP based models. These comprehensive (and expensive) studies have predicted that the carbon charge under Kyoto will be between US$50 per tonne of carbon and US$215 per tonne. It was not necessary to analyse the likely price range again using the Australian models (ABARE and CIE). They added very little to the New Zealand issues. We now know that there are increasing possibilities that the US, Australia, Japan and the EU will likely either not ratify a Treaty or find a way around its constraints. The implications are obvious. In this sense the world price question asked was redundant to the New Zealand research agenda. The international permit price can be parameterised in New Zealand studies to study these eventualities. These studies added very little to what was already known from a New Zealand perspective.

The second question concerns the social, environmental and economic impacts of emission restrictions in New Zealand. This is not an empirical question that is well answered by a model developed outside New Zealand, no matter what the expertise of the modellers. This is because overseas research institutes (and their models) do not have:

- the best data on New Zealand society and its economy [as will become clear later] and

- the best information on how New Zealand society and its economy are likely to react to issues like climate change policy.

These two points may be referred to as local knowledge.

The analysis of carbon taxes and subsidies on the production and alleviation of greenhouse gases in New Zealand industry involves a set of very complex issues. They include how each industry can change its production techniques to avoid emissions of such gases, how will the research and development sector react and produce new Kyoto friendly technology, how will Kyoto forest plantings respond to incentives, how will employees in vulnerable industries move to new jobs if they are displaced, and where will they move? The computable general equilibrium models used in three of these studies need this information to proceed. Yet much of the information has not been researched for the distinctive environment New Zealand presents and such research does not appear to have been commissioned here.

Commissioning an overseas model in a partial information vacuum, such as that just described, only increases this problem. This is because if a modeller doesn’t have a number she will simply have to make an informed guess; and informed guesses are best made by people with local knowledge. This is not a small issue. Computable general equilibrium models require hundreds of data points and parameters values for New Zealand alone. The general equilibrium analyses commissioned here are, accordingly, based on informed guesses of many parameters. This is worth bearing in mind because the quality of the results is simply a reflection of the realism of the assumptions.

In principle, the information problems could have been avoided in the two New Zealand studies that have been commissioned. In practice, however, this did not occur. The Infometrics model was not set up to deal with the creation of a market in emission rights. The results of the sectoral research by PA consulting, which could have been used to calibrate a general equilibrium model have instead been left hanging.

3. PA CONSULTING

This study was commissioned by the Ministry of Economic Development from PA Consulting Ltd, whose final report - Assessment of the Likely Impacts on Selected Sectors of a Domestic Emissions Trading Regime - is dated 29 July 2001. The report was released in November 2001, thus largely defeating its usefulness as an empirical input into general equilibrium modelling.

The officials’ Climate Change working paper on Assessment of Economic Modelling indicates the PA study was intended to allow a “reality check” of results from the general equilibrium modelling, with a more micro-economic assessment of the market characteristics and imperfections that different industries would face. But it also acknowledges that the different analytical techniques limit the ability to directly compare the ABARE and PA results. As a simple example, the PA study imposes an emission price assumption of NZ$20/t CO2 that is far removed from the emission prices generated by the ABARE model, of around $49-73/t CO2.

The PA study is largely qualitative, its estimates being confined to how emission prices would impact on industry costs or product prices. The sectors it examines are not aligned to the productive activities in the national accounts, but are rather defined by the sources of greenhouse gas emissions: energy emissions from the combustion of coal, gas, oil and geothermal fluid extraction; industrial process emissions from non-combustion uses of carbonaceous materials (e.g. cement manufacture and ironsand reduction), and from the minor synthetic gases used in refrigeration and switchgear (HFCs, PFCs and SF6); and emissions from waste management, agriculture and forestry (including carbon sequestration).

The study assesses the likely response of each of these sectors to a specific emission management instrument, namely emissions trading, under which those responsible for emissions would have obligations to report their emissions and acquire permits to match them. The placing of obligations would vary between industries according to administration and compliance feasibility: major users of coal and gas are assumed to have obligations, as are the operators of waste management facilities. But the obligation for oil emissions, which emanate principally from numerous and dispersed transport vehicles, would be placed “upstream’ of the emission point, on importers or producers of refined oil products. In agriculture the obligation could lie “downstream’ of the emissions points (individual farms), using production throughput of processing plant as a proxy for the various agricultural emissions.

In the first instance it is assumed that those with obligations would have to purchase all their permit requirements; but the study also considers what would be the effect of grandfathering permits to affected entities in proportion to their 1990 emission levels. It also gives brief coverage of the distributional and regional implications of such a regime, and how changes in the trading regime would change likely impacts.

3.1 Principal findings of the PA report

Emissions trading would have a discernible impact on the coal, gas and geothermal industries, and also on the closely associated electricity sector. PA expect increases in the price of gas, coal and electricity, with the result that more wind and micro-hydro generation will become economic. Gas would continue to be the principal source of electricity generation at the margin, but its increased cost would raise electricity prices, conferring on owners of existing renewable generation sources, in particular large hydro-generation, a windfall economic rent.

There would be increased pressure on marginal coal mines, and some may shut, especially if oriented to the domestic market. Export coal production would be largely unaffected, because no obligation attaches to exported coal.

There would be price increases on refined petroleum fuels, faced by both commercial and household transport users. But the demand for transport is inelastic, and these price impacts would not be large enough to significantly affect the demand for transport or modal choice.

Because of its large use of energy and coal for iron sand reduction, the viability of the steel industry could be threatened. Aluminium also faces a major energy cost increase, but could be shielded from this by its long term electricity supply contracts.

The report suggests cement and lime industries will be less affected, despite substantial emissions from calcination of lime, because of the absence of readily available competitive suppliers. The cost increase for the synthetic gases would be small and unlikely to significantly affect their use in refrigeration and switch gear.

Emissions trading is likely to have very little effect on wastewater management. In solid waste management it would reinforce the recent trend towards larger, more modern landfills with methane gas collection, and the closure of older, smaller sites.

The report finds pastoral agriculture would be one of the sectors most adversely affected, although the impacts would vary across farm type. Dairying would face the least negative impacts, because most competitors are also Annex 1 countries facing obligations under Kyoto, and because of a degree of market power for some products increases the likelihood of dairying recovering more of the extra costs from consumers. (As we discuss later, this is in direct contradiction to ABARE and others’ findings, and raises important issues of interaction between various elements of the research programme). The impacts on meat would be more severe, with loss of competitiveness to non-Annex 1 suppliers and other meats such as poultry. Deer farmers face substantial cost increases, with little likelihood of higher recoverable prices.

PA suggest the benefit of Kyoto forest sinks is only temporary and unlikely to provide a sustained stimulus for increased planting. Obligations for harvesting non-Kyoto forests and greater energy and transport costs for processors imply a negative impact on forest industries.

The report finds a grandfathered allocation of emission permits would have different effects across the sectors. It would alleviate the cost imposition on basic metals manufacture, but have little impact on deer farming and refrigeration.

The principal finding with respect to distributional impact is that emissions trading may provide a stimulus for rural depopulation. This is because of a combination of negative impacts on primary industries and increased transport cost reinforcing the tendency for industry relocation and consolidation around urban markets and ports, particularly towards the top of the North Island.

3.2 Commentary

A fundamental question in assessing the PA report is how useful are its findings in informing the debate about likely impacts of climate change policies? Answering this requires addressing a number of subsidiary questions about how reliable are the assessments, how accurately does the report depict the various industry structures and the likely policy applied to them, and within the limitations of its method and terms of reference, how fully has it identified the likely effects?

The authors stress the report is an “initial investigation” and was prepared in relatively short timeframe. This itself is troubling since it is not clear why the government would be commissioning an “initial investigation” so late in the decision-making process.

The study claims to apply a structure-conduct-performance framework across each of the sectors examined. Yet, it does not explore the causative links between each of these aspects of industry organisation, nor does it consider external competitive threats. The structure-conduct-performance paradigm is a tool for considering market competition, yet the report does not provide a thorough competition analysis.

The level of quantification is uneven across sectors: the energy sector has a reasonably detailed breakdown of the emission cost impact against different fuels and stages in the supply chain; agriculture has emissions costs compared against returns from each livestock enterprise; other sectors compare emission costs against turnover, which is not necessarily informative about the likely impact on profitability, and hence continued operation. There is little analysis of export orientation of different industries or the scale and origin of their main competitive threats.

3.2.1 Depiction of industry structures and policy settings

The report makes a reasonable assessment of the emission policy setting, given that there was no detailed proposal available at the time the study was undertaken. However, there remain some issues that arguably could change the interpretation of the assessment.

The energy market analysis may understate the impact of emissions trading. In particular, the finding that coal exports would be relatively unaffected may be over-optimistic. Although export coal would not face emission obligations, even coal producers dedicated solely to export could lose competitiveness with overseas suppliers, because of the obligations they could face on fugitive emissions of methane from their mining operations. A similar issue arises with the processing emissions of the refinery at Marsden Point, which would represent a cost additional to the emissions implicit in product use, that is not faced by competing refineries in countries outside Annex 1, such as Singapore.

As the report is not an inter-industry study, it has little to say about the inter-actions between primary and secondary production sectors. This may be a reason contributing to its conclusion that impacts on dairying may be less severe than those on other pastoral production (particularly meat). In contrast, general equilibrium modelling indicates the impact of a carbon charge measure on energy prices for both transport and processing can lead to bigger output reductions for dairy processing and dairy farming than for the corresponding meat industry sectors.

3.2.2 Identification of likely effects

Within the limitations noted above, the report identifies effects primarily as price increases and infers likely industry responses. However, two general issues critical to its conclusions bear closer examination: the extent to which cost increases are passed into product prices, and the likely effect of grandfathering permits.

The report makes a basic assumption that energy industries will pass on increased cost of fuel created by emission permits, but that this may not hold for higher prices than the $20/tCO2 base assumption. This begs the question of what happens to those costs at higher permit prices? Although there may be short term price stickiness, in the medium term, any permit price can be expected to be passed on, since the alternative of subsidising consumption out of reduced profits would not be sustainable. To the extent that passing the price on causes consumption to contract suppliers will face some loss, but they would make a larger loss if they did not recover the full costs they incur on marginal supply.

Suppliers enjoying a monopoly could conceivably absorb some of the cost if this is less damaging to their profitability than passing the cost on in full. They may, for instance, already be charging above what a competitive price would be, and absorb some of the permit cost to avoid larger contraction in sales. However, the report does not identify any sectors in which this appears likely. Further, although higher permit prices are likely to result in more substitution away from affected products, the conclusion that price pass through is variable with permit price level is unsubstantiated.

One of the key problems of the analysis is the confusion between whether marginal or average costs will determine prices. In describing the impact of grandfathering, the report notes that “while the average cost of gas will not increase by the permit cost, the marginal cost of additional generation ¡K will face the full cost of permits’ [page 4-11]. The implication is that if the average cost is lower than the marginal, the effect on prices will be reduced. Hence, at this point and elsewhere, the report implies that grandfathering lowers the likely effect on prices.

However, this is inconsistent with the literature on grandfathering of permits, or with the broader economic understanding of average and marginal costs. Even if gas suppliers face no financial outlay for some portion of their emission permits, each emission permit carries an opportunity cost in the sense that, if not used, its value could be realised by sale elsewhere. Hence there is incentive to recover the cost of permits in the price of all units of gas supplied. All that grandfathering does is provide an effective lump sum rebate that compensates for lost profitability, without affecting the price of gas supplied.

3.3 Reliability of results for informing debate

Given the limited quantification in the report, and inconsistent coverage of different sectors, it is hard to ascribe the report with a high level of authority. It is, however, an initial assessment, and cannot be expected to provide the last word on these impacts.

The report contains some surprising results. It suggests relative immunity for cement and lime production. This appears to ignore recent developments in cement transportation options (bulk bags) and the already visible effect of the threat of import on the domestic prices. Competitive incursion by imports are more likely than the report makes out.

As we mentioned, the relatively low impact on dairying largely reflects a failure to consider how the energy costs on transport and processing are likely to feed back into reduced returns in dairy production.

The specific conclusions that some sectors benefit more from grandfathering than others is largely a result of the assumption that grandfathering would be on the basis of 1990 emissions. Neither deer nor synthetic gases were as plentiful in 1990 as at present, and their grandfathered allocation would be a smaller proportion of emission requirements than those of longer established industries, like cement or steel. There are other bases on which permits could be grandfathered, and the mere fact of a gratis allocation of permits does not ensure minimal impacts. If an activity becomes unprofitable because its marginal costs increase, a supplier may simply cash in its gift of permits to ease the cost of closing down.

Overall, the PA study adds little to the government’s ability to make informed decisions.

4. GENERAL EQUILIBRIUM MODELLING

The results of computable general equilibrium (CGE) models critically depend on the underlying assumptions, model architecture, and key parameter values. When it comes to such issues, the frequent refrain of economists - “it’s an empirical question” - could not be more true. Some assumptions and parameter values are justifiable and others simply are not in the context of any one country. When configuring a CGE model, the analyst must rely on an amalgam of economic theory, historical observation (i.e. data), statistical and empirical evidence, informed judgements, and plain common sense (i.e. ask the simple question: does this seem right?).

We are concerned that the research commissioned by the government does not pass the test of justifiability. In order to make this assessment, we compare Infometrics and ABARE modelling to our own work. This is not a sign of hubris - we do not claim that our modelling is flawless. Rather, we have carefully considered the assumptions and architecture of our model, and this comparison helps us highlight the issues.

NZIER and ABARE have similar model architecture. Hence, as we discuss later, the differences can be easily explained by the data and the assumptions used.

The comparison with the Infometrics model is more difficult. It is not well documented, and its architecture is somewhat dated, making it more difficult to explain the differences in results. In order to attempt an analysis of the Infometrics results, we conducted an experiment using the NZIER and Infometrics models We should point out that while Adolf Stroombergen of Infometrics was not commissioned to work with us in comparing the two models, he did put in a considerable amount of time in undertaking the modelling exercise, providing us with the summary results, and answering our questions as we sought to reconcile the differences.

5. INFOMETRICS

Both the NZIER and Infometrics models, as used for this experiment, are single region (i.e. New Zealand), static CGE models with a focus on the energy sectors. Both models include a household sector disaggregated on the basis of five income quintiles; a government sector which collects taxes and demands public goods on behalf of households; a foreign (rest of world) sector from which New Zealand purchases imports and to which it sells export goods and services; and a collection of industry sectors which combine both intermediate and primary inputs (capital, labour, and, in the case of NZIER, land) to produce outputs.

The NZIER model is based on the 1996 126-industry/210 commodity inter-industry tables released by Statistics New Zealand in August 2001. For the purposes of this analysis we have aggregated that data to 28 industry sectors, each producing a single commodity. The Infometrics model is based on the 1995 50-industry inter-industry tables. For the purposes of this analysis it was aggregated to 32 industry sectors.

The 28-industry NZIER model was constructed specifically to look at climate change policy issues. Hence, we have built in a market to enable trading of the right to emit greenhouse gases, i.e. permits. If a sector wishes to emit, it must purchase permits in a predetermined proportion to the amount of the fossil energy it purchases, i.e. such energy supplies are the source of emissions. The Infometrics model has no such market. Instead, it simply taxes the use of fossil energy supplies. The two approaches collapse to the same thing in the case of the simple experiment we conduct here, but would generate quite different results in actual policy simulations.

5.1 The Experiment

Whilst the two models are based on different data sets, embody some different design philosophies, and are formulated and solved using different software packages, we used the models to quantify the macroeconomic impacts of a charge on CO2 emissions under the restrictive and unreasonable assumption that the industry sectors have no substitution possibilities with respect to their input mix. In other words, while each industry still controls the choice of how much output to produce following the imposition of the charge on emissions, it must continue to use inputs, both intermediates and primary factors, in the same fixed proportions that it used before the charge was applied. We call such a production process a Leontief production technology.

Restricting the substitution possibilities to zero is unrealistic. For example, if the price of emitting were to increase, one would expect firms, where possible, to compensate by replacing equipment with more energy efficient vintages, or to switch to less emitting energy sources. Nevertheless, imposing the zero substitution restriction should limit the range of responses available to each model, and thereby yield results which are more likely to be similar.

The tax imposed was $13.50 per tonne of CO2 emissions.

Both the NZIER and the Infometrics models incorporate a feature known as the “Armington” assumption, after Paul Armington (1969). The essence of this feature is that imported goods are considered to be imperfect substitutes with domestic varieties. For instance, imported cheese is a different good than New Zealand produced cheese. A set of parameters (one for each good in the model), which are unsurprisingly known as the Armington import parameters, determines the degree of substitutability the model will permit between imported and domestic goods. We configured the NZIER model such that it used the same Armington import parameter values as the Infometrics model.

The values used in the Infometrics model are, in general, somewhat lower than those adopted by NZIER. The implication of this is that the Infometrics model assumes domestic prices can change relative to import prices without causing changes to import shares more than the NZIER model does. As an empirical observation, we hold the view that many New Zealand goods can be easily substituted by imports. This certainly applies to such products as steel and cement.

By using the same values in both models we have eliminated this as a source of explaining the differences in results. To keep things simple, we assumed a fixed supply of labour and capital in both models. The supply of land, of relevance to the NZIER model, is, of course, also fixed.

In both models, the size of the government sector was held constant by recycling the revenue raised from the carbon charge back to households as a non-distorting lump sum. However, we can not be sure that both models in fact represent such recycling in exactly the same way.

In the NZIER model, the balance of payments deficit was held fixed at its benchmark (BAU) level. Although not quite equivalent, the current account in the Infometrics model was held fixed as a percentage of GDP.

A difference in models that we could not easily (due to time and resource constraints) control for was the assumptions relating to the export sector’s interactions with the rest of the world. The NZIER model adopts a format that can be viewed as the opposite of the Armington assumption which applies to imports. That is, each sector produces an output that can be used domestically or exported, and the ability to substitute between the two is governed by the so-called elasticity of transformation. The Infometrics model, on the other hand, includes a rest-of-the-world demand function for New Zealand’s exports. The key parameter in the Infometrics specification is the demand function’s own price elasticity. To gauge the importance of the difference in export specifications, we simply ran the NZIER model for a range of values for the elasticity of substitution.

5.2 The Results

Summary results are shown in the table below.

Table 1 Macroeconomic impact of an emissions charge

Percent change from BAU

NZIER Infometrics

įdx=2 įdx=4 įdx=6 įdx=8

CO2 emissions -0.69 -0.99 -1.27 -1.53 -1.45

GDP -0.50 -0.50 -0.50 -0.50 -0.03

Terms of trade -0.12 -0.12 -0.12 -0.12

Private consumption -0.02 -0.02 -0.02 -0.02 -0.07

Investment -0.01 -0.01 -0.01 -0.01 -0.01

Exports -0.16 -0.19 -0.21 -0.23 -0.02

Imports -0.17 -0.20 -0.22 -0.24 -0.06

Real wage 0.22 0.35 0.39 0.40

Notes: (1) įdx denotes the elasticity of transformation between domestic output and exports.

Source: NZIER and Infometrics

It is clearly evident from the table that the results are quite different for each model, despite our attempts to engineer similar results. The remainder of this paper discusses reasons for the differences.

With respect to the decline in emissions, the two models are similar if the NZIER results with a value of 6-8 for the elasticity of transformation are selected. We have not econometrically estimated this parameter but based on the available literature a value of 6 to 8 is not unrealistic. For this experiment we have held the elasticity value constant across all sectors simply for the sake of simplicity (we need not have).

However, when we get down to the GDP impacts, there is a marked difference between NZIER and Infometrics. Quite simply, we do not see how, when the possibilities for changing the input mix are zero, emissions can be reduced by almost 1.5% at practically no cost to the economy, which is what the Infometrics result is suggesting. If nothing else, the Infometrics result must surely imply a significant change in the output mix of the economy.

One possible explanation for the very low GDP impacts yielded by the Infometrics model is that there may be some switching in the fuel mix of the thermal electricity generation sector, e.g. coal replaced with gas. Evidently, the model allows substitution among the fuel types in this sector even when all other intermediate and primary factor substitutions are set equal to zero.

There are some basic differences in the data used by each model as reflected by measures such as sectoral output, import, export, and factor shares. Until Infometrics upgrade the inter-industry data used in their model to the most recently available, it is difficult to guess at what differences this might account for.

The Infometrics model considers oil and gas as a single commodity as that is how Statistics New Zealand present it in the inter-industry tables. But the emissions coefficients associated with these two commodities are substantially different. NZIER disaggregated these two commodities. This too can account for lower impacts.

In addition, there are some significant differences between the NZIER and Infometrics model in the area of base case emissions by sector and fuel source. For example, Infometrics allocates an unrealistically low proportion of base case emissions to the transport sector. This too would have an impact on results.

Overall, the stripped down comparison of the two models suggests that the Infometrics model would be likely to understate the economic impacts of a carbon charge by allowing a much greater degree of substitutability between energy sources than actually exists in the New Zealand economy.

5.3 Infometrics report

The Infometrics report concludes that a “low level” carbon charge, recycled in the form of lower taxes would improve economic welfare. This is generated through two effects: the direct impact of a carbon charge is found to be low, while the reduction in distortionary taxation (such as the income tax) has a positive effect. As we pointed out, the direct impacts of the carbon charge appear to be understated. What of the positive effects of lower taxes?

Unfortunately, the results produced by the Infometrics model are not consistent with the economic literature or with the empirical research conducted elsewhere. The positive effect would occur if the tax that is being cut is more distortionary than the charge that is being introduced. The literature would predict that the most positive result would come from a cut in the business tax, since such tax causes most distortions when capital is mobile. The next best impact would be from a cut in labour tax. The theory and other research would predict that a cut in GST - the least distortionary of all taxes - would have the smallest positive effect. In fact, substituting a carbon charge for GST could have a negative effect since a carbon charge is likely to be more distortionary than GST.

Yet, the Infometrics result is almost exactly the opposite. The best result is achieved by introducing a carbon charge and recycling the revenue through lower GST. This result raises significant questions about the architecture of the model - it appears (in the absence of documentation) that the tax treatment in the model requires further work.

6. THE ABARE STUDY

The ABARE results confirm that a number of New Zealand industries are high emitters of greenhouse gases and that these sectors will be forced to scale back operations when carbon taxes are introduced. The livestock industries (dairy, wool and other livestock in the ABARE model) are among the most severely affected along with the coal, steel, aluminium, gas and electricity industries. In the most extreme case, the output of dairy products would have to be reduced by around 25 percent from 2008 and wool output by around 28 percent. Being a general equilibrium model, these declines in output are offset by increases in the output of the light manufacturing industry (3.6%), the crop sector (12.1%) and the forestry sector (2.5%).

The study assumes that New Zealand will have carbon sinks (Kyoto forests) equal to 22.5 million tonnes of carbon dioxide by 2008 and that we will be a net exporter of emission permits to the rest of the world valued at $600 - $1300 million.

The study examines the range of possibilities for New Zealand that flow from the US and Australia not ratifying the Treaty and alternative responses to key exporters of permits - the Ukraine and the Russian Federation.

As referred to earlier one feature of the ABARE/GTAP modelling system is that the data are quite old and for some issues like this one, the results could be biased. The GTAP 4e database which is used by ABARE basically uses the 1987 input-output table to define the structure of the New Zealand economy. While this input/output table has been partially updated it does not take the effects of New Zealand’s radical economic reforms into account. These structural changes are significant. Better data is available. Earlier this year a new 126-sector survey-based input/output table was released by Statistics New Zealand. This input/output table forms the basis of some of the New Zealand computable general equilibrium (CGE) models in current use.

The out-of-date data makes considerable difference. The structure of the New Zealand economy is now materially different, particularly with the growth of dairy and road transport sectors. Both of these sectors are significant emitters, and therefore would substantially alter the results of the research.

The ABARE/GTAP model system breaks down each country’s economy according to a classification system devised to suit other countries, not New Zealand. For example, there is no sheep industry nor horticultural industry in the ABARE/GTAP model. This is another potential source of bias because it means that the distinguishing features of some of our important industries cannot be dealt with.

ABARE have a data set based on 1995 (except the 1987 input/output table) and in updating that data to 2010 their researchers have assumed that emissions (per unit of output) decline by 1.5 percent per year over that 15 year period. Now, actual New Zealand statistics are showing that emissions (per unit) have been increasing by 1.5 percent per year since 1995. The New Zealand economy has been getting more vulnerable, not less vulnerable, to carbon taxes. ABARE would appear to be underestimating the effects of carbon taxes on New Zealand. This error clearly stems from lack of local knowledge.

The ABARE results do not appear to add up in some important respects. In one of their extreme scenarios dairy, livestock, wool, coal, gas, steel and aluminium production are projected to decline by up to 28.5 percent. At the same time, exports only decline by 2.68 percent and GDP declines by only 0.24 percent (scenario 5, pages 17-18). Given that these reductions are suffered by some of New Zealand’s largest and most efficient industries (were we have our strongest comparative advantage), these results are very surprising. This result would ring alarm bells to anyone familiar with the New Zealand economy. We would expect such a massive reorganisation of the tradable sector of the New Zealand economy to involve a much larger economic cost than the ABARE model is showing. It is not at all clear why the model produces these results (a perennial problem with non-transparent modelling systems). However, indirectly it appears that these results are influenced by the choice of Armington variables (as discussed above, these variables describe the likely substitutability between domestic and foreign goods). Yet again, the choice does not reflect the empirical realities of the New Zealand economy, and appears to assume much greater degree of market power by New Zealand producers than actually exists.

The reallocation of land in the ABARE model also seems to be problematic. Scenario 5 has a large decline in land use for some of our largest farming industries with a relatively small compensating increase in crops and forestry. On the surface it would appear that a large acreage of land is taken out of production and not used. This would not be a realistic result.

Overall, the ABARE study uses a highly sophisticated model, but fails to adapt it to the empirical realities of the New Zealand economy.

7. CONCLUSION

Our overall conclusion is that the government is embarking on a major policy initiative on the basis of remarkably limited, poorly co-ordinated and empirically risky research. This does not appear to be a good basis for informed policy making.

© Scoop Media