|
|||
|
|||
|
 |
|
|
The Murphy Model is available for Members of Parliament who might
wish to look at alternative simulations using their own hypothetical
policy measures. |
During the course of 1997 there has been a good deal of interest in economic modelling in Australia with competing models being used in the recent public debate on levels of motor vehicle protection. One of the prominent models in that debate was an econometric model of the Australian economy developed by the Canberra economic analyst and former officer of the Commonwealth Treasury, Mr Chris Murphy. This model is commonly referred to as 'the Murphy Model'. This model has been made available for use by the private and public sectors for several years and has been utilised recently by Murphy to simulate the effects of car tariff reductions on the national and South Australian economies. This work was undertaken for the South Australian Government and became the basis of its submissions to the Industry Commission during the motor vehicle inquiry. Results of this work were also recently published as a research paper by the Information and Research Services (IRS) Branch of the Department of the Parliamentary Library.(1)
The IRS is sometimes approached by Senators and Members to undertake analytical assessments of the flow-on effects of actual or hypothetical economic policy initiatives, for example, the effects of a change in interest rates on unemployment or the economic implications of tariff rate changes or of changes in national immigration levels.
To assist it in undertaking such assessments, the IRS utilises the Murphy Model. The purpose of this paper is to demonstrate the Murphy Model's capabilities using several possible economic policy scenarios which illustrate policy trends and proposals recently advanced by commentators on Australia's economic performance. In this way, it is hoped that Parliamentarians will have a better appreciation of the Model's potential application and usefulness to meeting their own specific needs.
The Murphy Model (now known as MM2) is described by its originator as a fully-integrated macroeconomic computable general equilibrium model. By that he means that it not only addresses the relationships between macroeconomic variables such as inflation and unemployment, but it also analyses these variables at the industry level. Hence, it can track macroeconomic shocks through to specific industry outcomes. Further, it predicts the quarter-by-quarter path of the Australian economy over a nine year time horizon. However, it is not the intention of this paper to go through the nature of the Model.
All commentators on the economy use economic models of one type or another. In some cases, economic models may constitute just a few simple relationships (sometimes expressed as straightforward equations) which people use in their thinking. Indeed, some market economists seem to use little more than their own intuition. At the other extreme are the competing macroeconomic models which use a multitude of equations to try to capture the essence of the Australian economy in considerable detail.
Parliamentary attention is currently focused on the 1997-98 Commonwealth Budget. Some of the criticism of the Budget has been to the effect that outlays should have been cut further. Accordingly the first simulation presented in this paper is to model the macroeconomic effects of a $1 billion cut in Federal outlays for 1997-98.
The next simulation examines monetary policy; on 23 May 1997 the Reserve Bank of Australia announced a cut in official interest rates from 6 to 5.5 per cent. Accordingly this simulation considers a loosening of monetary policy by modelling the effect of a cut in interest rates. To try to approximate the Reserve Bank's action, this simulation models the effect of a 0.5 per cent reduction in the 90 day commercial bill rate starting in the third quarter of 1997.
Finally, there is often concern about labour market behaviour and policy measures which might be taken to improve labour market performance. The last simulations analyse labour market performance and model the effects of improved functioning in the labour market.
Each of these scenarios is examined for its implications for gross domestic product (GDP), the consumer price index (CPI), employment and the balance on the current account of the balance of payments. They are presented as illustrations of the type of current economic policy concerns which can be assessed by the IRS to meet Parliamentarians specific requests using the Murphy Model of the Australian economy. For GDP we use the Australian Bureau of Statistics' preferred measure, GDP(A) or GDP (average) as our preferred measure of GDP. The average is the average of the three other GDP estimates based on income, expenditure and production.
This section looks at the possible effect of a further tightening in the 1997-98 Budget context. For present purposes this is modelled by looking at a reduction in government current expenditures of $250 million in each of the four quarters of 1997-98, after which government expenditure remains permanently lower. A cut of $1 billion to government consumption expenditures represents a 1.53 per cent cut in 1997-98. It is assumed that the same proportionate reduction is continued through into the future.
The results of this simulation are shown in the following graphs which plot the economic consequences by looking at changes induced in GDP, the consumer price index, employment and the balance on the current account of the balance of payments. All results are compared with the baseline: a 'no policy change' simulation which predicts the behaviour of the economy in the absence of any policy changes. All the graphs in this paper set the baseline at zero so that all movements in the main variable represent induced movements around the baseline. For this reason the vertical axis in the following graphs refers to '% dev', the percentage deviation of the economic magnitude in question.
As can be seen from the first graph, the fiscal contraction has an almost immediate effect on the level of activity as measured by GDP. GDP falls immediately in the third quarter of 1997 and stays below the baseline for eleven quarters. At its worst, the Model indicates that GDP will fall by a quite significant 0.16 per cent below the baseline. That represents lost output equivalent to around $700 million per annum. Employment immediately follows the fall in GDP and remains below the baseline for sixteen quarters, much longer than the decline in GDP. At its worst there is a reduction in employment of 0.14 per cent or around 12,000 jobs in mid 1998. With the reduction in output and demand the balance of payments on current account follows by improving quickly at first and then keeping track with the overall level of activity.
The prices effect may not be immediately obvious. There is an initial increase in inflation followed by a later large movement below the baseline. The upward blip in the CPI is caused by a fall in interest rates which actually precedes the fall in Government expenditure. That reflects the forward looking character of the Murphy Model. Expectation effects are built in which lower interest rates and the Australian dollar in anticipation of the cut in Government expenditure. The depreciation in the exchange rate increases import prices and so feeds into an increase in the CPI. This effect is short-lived however, and after seven quarters the CPI falls below the baseline. The short term reduction in interest rates has the effect of stimulating investment, including investment in dwellings. This stimulus to investment and net exports is responsible for the recovery from the short term adverse effects of the cut in Government outlays. In the long run, output and employment are permanently higher.
The Murphy Model allows us to model the effects on individual States. In most respects the results are much the same across the States. However, the exception is the ACT where there is a permanent reduction in employment following the cut in Government expenditure. That contrasts with the national results which show a modest increase in employment following the decline over the first 4 years. In Australia as a whole the initial fall in employment is offset by later increases, like a pendulum swinging back the other way. However, in the very long run, beyond the 10 years examined here, there is no increase in employment. In the ACT, by contrast, the trajectory for employment does not recover from the initial reductions and, as a result, the ACT experiences a permanent decline.
This simulation has modelled the effect of a $1 billion cut in government current expenditure which take place in equal quarterly cuts. However, it is easy to infer the effect of different possible cuts. If the Government were to announce a $2 billion cut, for example, the above results could be multiplied by two to find the approximate outcome.
It is also worth noting that while the cut in Government expenditure gives short term adverse effects, in the Murphy Model it has no lasting effects in the long run. Shocks to the Model through changes in Government outlays have temporary effects, though those effects may last quite a long time relative to the political cycle. In particular there are quite adverse movements in GDP and employment for some years as a result of expenditure cuts.
As mentioned above, the Reserve Bank reduced increase interest rates on 23 May 1997. A simulation has been performed to look at a loosening of interest rates. Reserve Bank action would normally impact on the official money market, and changes there would subsequently flow through to the rest of the interest rate structure with, for example, banks being one of the first lenders to pass on the interest rate cut in the present round. The Murphy Model allows us to analyse the effect of manipulating the 90 day commercial bill rate. For the purpose of this simulation the 90 day rate was reduced by 0.5 of a percentage point below what it would have otherwise been starting with the 3rd quarter 1997. As it happened, the cut in official interest rates was quickly followed by similar cuts in commercial lending rates. The results of the simulation are given in the following four charts.
The immediate effect of the reduction in interest rates is to induce an increase in GDP and employment, as might be expected. The changes in employment track the changes in output very closely, albeit with a slight lag. The increase in employment is modest at first but moves to 0.66 per cent above the baseline by 1999. Note that employment is above the baseline for a total of 18 quarters before returns to just above the baseline. The response of the balance of payments on current account is more complex. The interest rate reduction and the induced depreciation of the $A discourages imports, encourages exports and so causes an improvement in the current account. This effect is short-lived and is soon dominated by a second mechanism which produces a deterioration in the current account. An increase in the level of activity in Australia will induce an increase in the demand for imports. In the case of capital goods imports, the reduction in interest rates will have the effect of increasing investment generally and, since investment is import intensive, this will produce a disproportionate increase in imports.
The consumer price movements are perhaps as might be expected. The CPI follows the movement in economic activity and employment, albeit with a lag. We would expect the CPI to move with the level of economic activity and employment, since price increases would be expected to strengthen as economic activity improves.
In the long run, monetary policy in the Murphy Model has no effect on real variables. Hence, the simulations show no long run changes to the real variables examined here: GDP, employment and the balance of payments on current account. However, monetary policy has a permanent effect on prices, here estimated as a 2.3% increase in the CPI.
When looking at the effect of monetary policy, Chris Murphy prefers to model monetary tightening by assuming this translates into a reduction in the Reserve Bank's inflation target, which translates into a reduction in the Reserve's target for nominal (current dollar) Gross National Expenditure. However, simulations for this type of scenario show roughly the same pattern of output responses in the Model, and so are not reported separately.
In this section the effects of reform in the labour market are examined by making the appropriate adjustments in the Murphy Model. Simulations in this area are different from the previous simulations performed here. For the earlier simulations we could directly feed in the policy changes we wanted to examine. However, for labour market reform, as will be seen below, some initial assumptions have to be made about the first round effects of labour market reform. To give the model something to work with, some assumptions about changes to the initial conditions have to be fed into the model to get it started.
A number of labour market reforms have been discussed from time to time which advocates claim will make the labour market more efficient. For example, efforts to increase the number of training places available to the long term unemployed should assist their return to employment. By better matching labour supplies with the jobs on offer, the labour market would be more efficient.
Other participants in the debate claim that further deregulation, such as the abolition of awards and reducing union power, will improve the efficiency of the labour market. In either case it could be argued that there will be substantial benefits from those policies. In order to quantify the benefits of such policies, we have to make some assumptions about how this might affect the workings of the economy.
The usual way of proceeding is to assume that the measure in question will lower the non-accelerating inflation rate of unemployment (NAIRU), which refers to a notion of the sustainable level of unemployment below which worsening inflation would be a real threat. The idea here is that increased labour market efficiency is supposed to improve the trade off between unemployment and inflation. Hence, with the introduction of labour market reform a lower level of unemployment should be possible without threatening higher inflation. As an advocate of a specific reform we might say that our preferred measure would have the effect of lowering the long term or equilibrium level of unemployment. For example, it might be thought that a specific labour market reform would lower the NAIRU by 0.7 per cent of the workforce. This case is examined in the following simulation. It is, of course, an assertion to say the NAIRU will fall, but it is necessary to assume some such linkage in order to make some initial change so that the effects of the labour market policies can be modelled and to simulate their impact on other economic variables. When we are looking at policy changes we have to make do with assertions as to how the policy will work and make the appropriate adjustment in the Model. The results of this simulation are shown in the four graphs below.
While the above graphs show the Model outcomes of the labour market reform it must be remembered that the Model is not saying the policy will necessarily work in the way indicated. The results are obtained by assuming that some features of the economy would change in a certain manner as a result of our policy change. It is really the effect of that assumption that is being modelled here. Hence the simulations are based on our view as to how the economy might respond to the policy measure. It would be open to a critic to say that the supposed changes to the Model are unlikely to result or may even go in the opposite direction.
Going through the results shows that the outcomes are generally positive on the indicators we have been using here. GDP and employment both increase, slowly at first but ultimately stabilise at significantly higher levels. However, the CPI effects are mainly driven by the behaviour of wages. Greater competition in the labour market leads to lower wages. Through the effect on production costs prices fall, but not by quite as much, with the result that real wages fall almost 1 per cent. In the very long run real wages rebound and leave real wages unchanged compared with the baseline, but that takes us outside the forecast period.
Important themes in labour market reform have been incentives to improve productivity and the introduction of productivity improvement measures. It is worth comparing the results we obtain by assuming a direct effect on productivity rather than immediate impacts on the efficiency of the labour market. Here we can model the effect of a 0.25 per cent 'skill' increase beginning in the third quarter of 1997.
The two different sets of simulations on the labour market are interesting to compare. The pattern of changes is broadly similar in the two: GDP and employment both increase above the baseline, although the employment effect is not permanent in this case. The CPI falls below the baseline and the current account of the balance of payments tends to follow GDP in the expected manner. There is little apparent difference in the graphs shown above. However, while the CPI is below baseline in both cases, in the first simulation wages are below baseline by even more than the CPI. That means that real wages have declined relative to the baseline scenario over the forecast period. Effectively, wages are falling and bringing down prices but with the former outpacing the latter. In the other labour market scenario the opposite happens. The deviations below the baseline of the CPI are much larger than for wages meaning that real wages are increasing. This is shown in the final graph immediately above. This graph shows that the wage deviation line falls much less than the prices deviations. It is also worth noting that, while real wages go in opposite directions in the two scenarios, long term employment levels increase under both scenarios, although the second scenario may not entail any reduction in sustainable unemployment. However, the modelling suggests that a package of labour market reforms which incorporated both types of reform could possibly increase both employment and real wages in the long term.
These results are very interesting. Evidently, industrial relations and labour market policies which create more flexibility and greater competition between workers are likely to result in reductions in real wages compared with what might have otherwise happened. While not permanent in the Murphy Model the reduction in real wages nevertheless persists for a long time. On the other hand, measures aimed at boosting the productivity of workers result in increases in real wages. Because real wages can go either way it becomes critically important that particular labour market policy measures are modelled in the most appropriate way. However, there are no hard and fast guidelines in these sorts of exercises. Much depends on the skill and artistry which the economic modeller is able to bring to the task. Unlike the first two simulations, labour market reform is less amenable the mathematical representation. New values for certain economic variables cannot be simply entered and the Model allowed to run. Instead, to simulate labour market reform we have to assume or assert particular effects which can be fed into the model. Much depends on the accuracy of such assumptions or assertions.
This paper is intended primarily to demonstrate the types of economic analyses which the Information and Research Services Branch can undertake using the Murphy Model of the Australian economy.
It is emphasised that the scope of such analyses can be tailored to the specific needs of individual Parliamentarians and may address overall macroeconomic concerns (for example, national inflation, interest rates or unemployment issues) as well as more specific concerns such as immigration levels, State-specific economic performance or the consequences of structural adjustment in particular industries, for example, the effect of tariff changes on the car industry of South Australia. The latter was examined as part of Chris Murphy's paper on tariffs commissioned by the IRS.(2)
The particular case studies illustrated in this paper show the usefulness of the Model in showing the possible effects of different policy choices. One of the case studies relates to the effects of a contractionary stance in fiscal policy causing a cut in Government expenditures; a second case study analyses an expansionary monetary policy using a reduction in interest rates. The Model's results suggest that contractionary fiscal policy has the effect of inducing short to medium term reductions in output and employment. Prices tend to follow GDP since the cut in Government expenditure is associated with lower interest rates and, as a result, some temporary depreciation in the exchange rate which shows up as short term price increases in the Australian economy. By contrast, the output of the Model indicates that expansionary monetary policy induces a fairly strong expansion in the economy over about four years. However, a permanent expansionary monetary policy of the type examined here gives a permanent increase in prices.
Labour market policies have always received a good deal of interest. An interesting finding from the Murphy Model analysis is that those policies which claim to increase labour market efficiency and improve the trade-off between inflation and unemployment have many effects similar to those which involve increases in labour productivity. Both are associated with increases in national output. However, the implications for real wages and employment are different. Labour market efficiency measures tend to be associated with reductions in the real wage compared with the baseline. Measures designed to increase the productivity of workers are different in that they involve an increase in real wages, but in the long run, no permanent change in employment. Against that, labour market efficiency measures which reduce the sustainable level of unemployment will thereby also increase employment.
These illustrations are intended to demonstrate the Murphy Model's usefulness in addressing some typical economic policy questions; they are presented in order to give Parliamentarians a better appreciation of the scope and types of analytical economic services which the IRS can provide by using the Model's capabilities.
