Key departmental publications, e.g. annual reports, budget papers and program guidelines are available in our online archive.
Much of the material listed on these archived web pages has been superseded, or served a particular purpose at a particular time. It may contain references to activities or policies that have no current application. Many archived documents may link to web pages that have moved or no longer exist, or may refer to other documents that are no longer available.
Environmental Economics Seminar Series
Department of the Environment, Sport and Territories, 1996
ISBN 0 642 24879 6
Centre for International Economics
I have been asked this morning to talk about an economic model called G-cubed. G-cubed is one of the models used to provide a framework for thinking through some of the interactions between industries and between countries that may or may not take place as a result of environmental regulations.
G-cubed, like any model, is an extreme simplification of the real world or the real trading system. But because we have simplified it, we can use it to trace through the effects of environmental regulations and to assess questions such as what will happen if one country imposes an environmental regulation and another does not. Will that have an effect on trade flows? What sort of effect will it have on overall welfare, and so on?
This morning I want to talk briefly about what G -cubed is and how we use models like this to think through these issues. First, I wish to put G-cubed in context because there are a lot of models out there that are used to analyse environmental issues. It is worth pointing out where they fit together and where this particular model fits in. I will talk first about how we model trade in G-cubed and then how we model environmental policies. Finally, I will discuss how the two - our trade and environment policies - interact.
In technical terms G-cubed is a multi-sectoral - that is, more than one industry - multi-country and dynamic - it has a time path extending out to about 2020. It is a general equilibrium model of the world economy. In the context of G-cubed, 'general equilibrium' means that we try to capture the interactions between sectors within the economy and, in doing that, we also capture the interactions between sectors in different economies. The equilibrium part of it means that we have to assume that at some point in the future the economy goes back to some sort of stable equilibrium growth path.
G-cubed is not actually an equilibrium model. The short term for G-cubed can be 10 to 15 years with significant disequilibriums, but at the end of the day we assume that the economy goes back to some sort of steady growth path.
G-cubed was originally developed to help the US Government analyse the issue of carbon taxes. It was developed by a colleague of mine, Warwick McKibbin, who is at the Australian National University, and a colleague of his, Peter Wilcoxen, who is at the University of Texas at Austin. At the time when G-cubed was developed, there were already a number of models looking at carbon tax issues, but the authors of G-cubed felt that these other models were lacking in a number of areas. First, many were just single-country models; they treated the whole world as a single country and thought about optimal greenhouse policies and so on in that context. They missed out the interactions between countries. That was an important limitation.
Secondly, some of the models did have interactions between countries but did not have any sectoral details - they assumed all these countries produced the one good. That was another limitation. Finally, those models that did have many countries and multiple sectors, did not account for capital flows that are used to compensate the trade flows of goods and services. As I will argue later, the way in which capital flows compensate for physical flows of goods and services can have a very important effect on the outcomes we project from environmental policies.
G-cubed was developed to cope with some of the limitations of those other models. G-cubed has some limitations itself - compared with some of the other disaggregated models, it is fairly aggregated. At Monash University they use a model called the Monash model, which contains considerably more sectoral detail of the Australian economy than does G-cubed. Also G-cubed contains less regional disaggregation than some other multi-country models. It has a little less disaggregation, for example, than the model developed by the Australian Bureau of Agricultural and Resource Economics (ABARE). G-cubed's compensation for this is that it tends to capture many more economic interactions than any other models.
Chart 1 gives an idea of the sort of regional and industry coverage of the model. The current version of the model distinguishes the United States, Japan, Australia, the rest of the OECD, China, Eastern Europe and the former Soviet Union and the other developing countries. There is another version of G-cubed currently under construction, which considerably disaggregates in the Asia-Pacific region.
CHART 1 - Regions and sectors in the G-cubed model
Its sectoral breakdown has five energy sectors - electric utilities; gas utilities; petroleum refining; coal mining; and crude oil and gas extraction. It also has seven non-energy sectors: the rest of the mining industry; agriculture, fishing and hunting; forestry and wood products; durable manufacturing; non-durable manufacturing; transportation; and services. Compared with other economy-wide models, this is a fairly aggregate industry classification. The problem is that once you combine a number of industries with a number of countries and you are extending 20 or 30 years into the future, computational constraints become a real problem, so you really have to compromise to a degree on your country and commodity coverage. This coverage is designed to allow a reasonably sensible analysis of greenhouse issues with more detail in the energy sectors.
Some of the key features of the model are:
G-cubed is like any economy-wide model in that it has a number of basic building blocks. First, it has economic agents - it recognises producers, consumers, governments and financial markets as distinct economic agents. Each of those economic agents has some behavioural assumptions which we use to drive their behaviour. For example, producers are assumed to maximise their profits. In the context of a dynamic model like G-cubed, they are trying to maximise the future stream of their profits. To put it another way, they are trying to maximise something that is roughly equivalent to the present value of their stock market valuation. In doing this, they have to choose the level of output and also the level of investment that they use for production. G-cubed recognises that if you want to change the level of investment, there are some adjustment costs involved and so it has explicit adjustment cost relationships.
Consumers are assumed to maximise utility. Again, it is in the context of a dynamic optimisation problem. Some want to maximise the present value of all their potential future consumption. Others are assumed to simply react according to short-term liquidity constraints - that is, they will consume now if they have the money; if they haven't got the money, they will not. The mix of optimising and non-optimising consumers in the model was determined by looking at historical data and trying to get the model to track it and fit it into that parameter.
As well as producers and consumers, the model recognises governments. Governments are assumed to do what governments normally do - they tax and spend. They can run deficits if they want to, but if they run deficits the model explicitly says that at some point in the future those deficits must be paid and because the agents in the model are forward-looking, they know that as well. If the government chooses to go into debt now that affects today the interest rates, prime rates and so on. Thus, future government liabilities are factored into those decisions, much as you see in financial markets in the real world.
There is a separate financial sector in the G-cubed model and this has two purposes: first, households can hold some of their wealth in money and use their wealth to buy goods and services, or they can buy financial assets; secondly, the financial sector is the intermediary by which trade flows and capital flows compensate for each other.
How does trade come into the model? Let us look briefly at the production structure of the model. See Chart 2. What we have is that any one industry - whether it be mining or manufacturing - produces its output by combining energy, capital, materials, labour and in some cases a sector-specific factor, which I will deal with in a moment. The industry buys electricity, gas, petroleum or whatever to produce an energy bundle that uses capital to build its capital stock. It combines materials from other industries - for example, the mining sector must buy inputs from the manufacturing sector, such as trucks and so on; similarly, the agricultural industry must buy tractors from the manufacturing sector. All the sectors buy from the transportation sector to transport their goods and services. Within this structure, there are all the input-output links between industries within a country.
However, as well as buying all this domestic output, each firm also buys imported goods and services. The mining industry not only buys Australian trucks, it also buys German trucks, or perhaps trucks from all around the world. When buying imports, the domestic industry has a choice as to where it gets its imports. It can buy its trucks from Germany, Japan, Korea or wherever. This is where trade comes into the model because the decision on where to buy products is based on relative prices, as well as on preferences about the same good from different countries. So if the relative price of trucks from Korea goes up, Australian companies will buy fewer trucks from Korea and more from Japan, assuming that those trucks perform the same function. Trucks from Korea and Japan are not perfect substitutes. They are imperfect substitutes, recognising that they are not necessarily the same product, even though both may be called 'trucks'.
Similarly, for the household sector, people can buy either imported goods or domestic goods and they choose on the basis of relative prices. That is essentially how trade comes into the model.
There is one other important aspect for trade in the G-cubed model - and that is the relationship between trade flows and capital flows. Chart 3 shows the relationship between trade flows and capital flows and also between domestic savings and investment. Three things must add up and must be equal. First, in any country the excess of imports over exports must be compensated by the capital. You have to pay for your imports somehow. So that will be equal to the inflow of foreign savings - that is, the inflow of capital that foreigners have accumulated. That relationship is mediated by the exchange rate.
Similarly, the difference between domestic savings and domestic investment must also be related somehow to the inflow of foreign savings. It is related by the interest rates. When there are insufficient domestic savings in the economy, the domestic interest rate will rise and that will attract foreign savings. The interest rates mediate the flow of foreign capital.
Similarly, there is a direct relationship between savings and investment in a country and that country's balance of payments -that is, the excess of its imports over exports. This chart also shows that specific policies retarding the balance of trade will not work because you cannot improve the balance of trade unless you do something about the savings and investment in the economy. If savings and investment are unbalanced, all you will do by subsidising exports is to affect the exchange rate; you will not improve the balance of payments.
That is the other side of the coin that Charles used. There is no point in using an environmental or any other subsidy to specifically target your balance of trade because there are a couple of other things out there that will compensate. The balance of trade is driven by the relationship between domestic savings and domestic investment. These links are all present in the G-cubed model.
Let us work through an exercise. What would the G-cubed model say about certain situations. Let us imagine we have three countries - A, B and C. A is Australia, and unilaterally we impose a carbon tax. If you work that through in the model you will find that that carbon tax will, firstly, increase the price of coal. The increased price of coal will most likely, in the first instance, feed through into higher prices for electricity. That, in turn, will feed through into higher priced manufactures or whatever else uses electricity. The end result of this is that we are producing some product - X. Country C will choose between Xs made in Australia and Xs made in country B. After the carbon tax, it will see that the Australian price has risen relative to the price in country B. Because country C is substituting on the basis of relative prices, it will import more from B, but less from A. In this instance, a unilateral carbon tax in Australia will affect the trade flows between these three countries.
There are two qualifications for this. First, just because it affects trade flows does not say anything about whether the carbon tax is good or bad. The carbon tax needs to be evaluated against completely different criteria. Secondly, this effect may be very seriously washed out by the time the price of coal goes through the price of electricity and up the chain. Some empirical studies have shown that with some sorts of environmental regulations this effect really is washed out, so it does not affect trade flows.
In the case of Australia, the big trade flow that is obviously affected when we unilaterally impose certain sorts of carbon taxes is the export of coal direct. That is if we do it alone and no other country does.
The situation changes quite dramatically if all three countries impose a carbon tax. If this happens then approximately the relative prices between these three countries will not change. There will be differences in the composition of production in different countries - carbon-based fuels may be a little more important in Australia than they are in a country such as Norway. At the margin there will be some differences, but because you have this washing through effect, at the end of the day you will not necessarily have big changes in relative prices.
That is half of the trade story; that is the story about the effect of a specific sort of environmental regulation - the carbon tax - on the price of goods and services traded between countries. The other half of the story is what this carbon tax is doing in the domestic economy and the effect it is having on interest rates, and therefore on exchange rates. All this assumes that the nominal exchange rate does not change. We have changes in the real exchange rate because we have changes in relative prices between countries, but so far we have assumed that the nominal exchange rate does not change to affect this pattern.
In G-cubed we have found that there are a number of ways in which environmental policies can affect nominal exchange rates and interest rates, and that is to do with how the revenues from those policies are used. Obviously not all environmental policies generate government revenue, but the carbon tax is one that does.
We can illustrate this in the case of the United States because the US is a big enough country to have an effect on world interest rates and so on. We did some experiments where we imposed a carbon tax of $15 a tonne of carbon dioxide, and we looked at ways of using the revenue from that tax. The first way you can use a revenue is to reduce the government's budget deficit - the US has a big deficit, so they could simply use that revenue to reduce it.
An alternative is to give a lump sum rebate to householders in terms of lower tax bills. Another way is to have a general income tax rate cut and a cut in company tax. The fourth option is to use the revenue to give an investment tax credit to certain industries, such as the energy industries so that they can look for ways to produce alternative fuels.
Chart 4 shows us in each year how much US real GDP would vary from an underlying growth rate after the imposition of a carbon tax. The standard sorts of simulation that people use show that if you use the revenue to reduce the deficit, GDP drops initially quite a bit, then slowly comes back, but remains permanently lower over the long term.
The exact opposite occurs if we use the revenue from the carbon tax to fund an investment tax credit. GDP initially falls, but as the industries that are getting investment subsidies start to invest and produce more capital, GDP rises and remains permanently higher. I should caution you not to interpret this GDP figure as having a welfare meaning; the other side of the coin is that because GDP is higher and because investment is higher, consumption is actually permanently lower. We have just switched the mix of investment and consumption in the economy. The point of this is to illustrate dramatically different paths depending on what you assume about the use of the revenue from the carbon tax.
How does this affect the rest of the world? How does it affect international trade flows? Firstly, there is a dramatic difference between the two simulations in the effect of the carbon tax on the US current account. See Chart 5. In one simulation the US current account remains permanently below the underlying trend rate and in the other it remains permanently above. That, in turn, is being driven by dramatic differences in the effect on the long-term interest rate. (Chart 6 ) In one alternative, where the US uses an investment tax credit, the rate permanently rises; but in the other, where the money is used to pay off fiscal deficit, the rate permanently falls. So there is quite a big difference between those two options, just by simply changing the way in which the revenue from the tax is used.
There is one other interesting effect on the US current account. A carbon tax can improve the balance of trade. See Chart 7. In the option where the tax was used to reduce the deficit, the current account actually improved and the same is true for the balance of trade. Thus, if the US imposes a carbon tax, at least initially the balance of trade goes into surplus. Most people would think that if you imposed a carbon tax you would expect the balance of trade to go into deficit and stay there. The reason it does not do that for the US is the effect that the US has on international capital markets.
- Consequences for current account of a carbon tax under alter native revenue recycling assumptions
- Consequences for interest rates of a carbon tax under alternative revenue recycling assumptions
Let us think through the mechanisms. If you impose a carbon tax and you use that money to pay off the government's budget deficit, you are immediately increasing savings within the economy. If the government pays off its deficit, that is a net increase in domestic savings in the economy. Economy-wide savings are made up of what households save and what government saves. Secondly, you are reducing aggregate domestic demand. Thirdly, you are reducing the marginal productivity of capital. All the capital you have in store that uses carbon-based technology is now less productive for you because of the carbon tax. These three things together reduce interest rates. That, in turn, does two things: it leads to a capital outflow from the United States and to a depreciation of the US dollar, and therefore an improvement in the balance of trade. Because the US is a large country which has an impact on world interest rates, the way in which it uses the revenue from this environmental policy can have a dramatic effect on its trade flows.
That is just a brief introduction to the G-cubed model. I have tried to show that within it there are some standard, real mechanisms by which particular environmental policies may change relative prices and therefore affect trade flows. There are also some non-standard mechanisms for these kinds of models in which environmental policies may also affect interest rates and exchange rates, which will have a different set of effects on trade policy. It is fair to say we have not fully explored these trade flow effects, particularly bilateral trade flow effects with the G-cubed model and it remains to be seen whether particular environmental policies have a big effect on trade flows or whether these two compensating sides of the argument cancel each other out. That is a piece of research yet to be completed, but I guess the idea is that this kind of framework can be used for tracing through these sorts of interactions.