Scenarios using a computable general equilibrium model of the New Zealand economy
2 Computable general equilibrium model
2.1 Balancing demand and supply
Economic models are sets of equations representing the major relationships between the various sectors and participants in an economy. These equations together form a coherent, but necessarily simplified, depiction of the workings of an economy. In essence, the modelling process illustrates the outcome of a balancing act (performed by the market) between the demands for goods and services and the resources necessary to produce those goods and services to satisfy such demands.
As depicted in Figure 2.1, the demands for goods and services can be simplified as originating from households, government, and exports. The resources required to produce goods and services comprise labour, capital (machinery, equipment, and buildings), land and other natural resources, and technology. Additionally, some demands are satisfied externally - through imports.
Figure 2.1 Schematic outline of relationships
2.2 Computable general equilibrium model
A computable general equilibrium (CGE) model is a standard and widely used tool to investigate the impacts of economic shocks or events, or to measure the contribution of sectors or industries to the wider economy. The model captures the inter-relationships between industries and between exports, imports, consumption, as well as their combined resource requirements (see Appendix A for further detail on structure).
The model follows standard neoclassical assumptions of market-clearing prices,[17] profit-maximising firms,[18] and utility-maximising consumers.[19] The equilibrium of the economy is determined by market-driven adjustments to the relative prices of production factors (resources) and outputs that ensure supply equals demand in each of these markets. In addition, embedded in the production structure of firms is the standard neoclassical assumption of zero pure (economic) profits.[20]
The Business and Economic Research Limited (BERL) CGE model of the New Zealand economy (called Joanna) used in this study, separately identifies 53 industries, 25 export commodities, 8 household consumption commodities, and 40 occupation categories (see Appendix B for further details).
The model has its origins in the models the Project on Economic Planning at Victoria University of Wellington developed in the early 1980s. Early applications focused on trade policy questions, with simulations of tariff removals and General Agreement on Tariffs and Trade outcomes, contributing to the 'gains to free trade' argument prevalent at that time.
The model was originally based on a model of the Australian economy,[21] so its structural framework is similar, arising from input-output relationships. This model can simulate the effect of a policy, world price, world demand, productivity, and/or behavioural shock and solves for the equilibrium outcome in a future identified year. The model used in this study calculates the results for a user defined year, in this case 2021, but does not report the path the economy takes in preceding years.[22]
The key elements of the model's structure are summarised in the sections below.[23]
2.3 Advantage of model is in its industry detail
An important feature of CGE models is that the equations to estimate demand and supply can be constructed at a detailed industry level. Furthermore, they are based on inter-industry relationships, which show the flows of goods and services between industries. Therefore, the model's estimates of employment and output growth by industry recognise that expansion or contraction in any one industry leads to a flow-on of demand into many other industries. Thus, we can explore questions such as, 'If the production of wood products were to increase 10 percent, how much would that affect the demand for the services of the transport industry?'. In other words, industries use inputs to produce goods and services and some of these inputs are goods and services produced by other industries. The modelling process captures and mimics the relationship between these inputs and outputs.
Furthermore, the ability of certain industries to change the amount (or type) of inputs they use is incorporated in the model. This ability to change (ie, to react to demand, supply, and price shifts) is limited by technological factors. And the extent to which industries change their inputs is guided by standard economic theory, which assumes producers strive to adopt the lowest-cost method of production.
2.4 Data and aggregations
The limited availability of data as well as practical limitations mean that any model involves a degree of simplification. The model, just like all simplifications of reality, is only as good as the information available. The CGE model in particular devours information (ie, data on the aforementioned relationship between inputs and outputs), but such up-to-date and detailed information is always difficult to obtain.
As stated in section 2.3 the CGE model is based on inter-industry relationships. This information comes from input-output tables. The most recent full-scale official input-output tables for the New Zealand economy from Statistics New Zealand describes the inter-industry relations as they were in 1995/1996. A new set of input-output tables was developed for the present study, so more up-to-date, realistic and accurate estimates can be made. However, updating input-output tables is a far from trivial exercise. Information from more up-to-date supply and use tables[24] were used to derive input-output tables for 2003/2004. From this information, inter-industry transactions tables for 2005/2006 were generated using the RAS method.[25] Of course, while not ideal, a RAS update from a 2003/2004 starting point is infinitely superior to a RAS update from a 1995/1996 starting point.
Information from the 2006 Census of Population and Dwellings was used to update other data necessary for the model and allow the baseline to be as accurate as possible. As such the latest employment by industry, occupation, and household income data were incorporated using 2006 census figures.
Within the model elasticities determine the ability of industries to substitute between different types of labour occupations. Industries are assumed to undertake such substitution in response to price (wage) changes and/or constraints on the availability of different labour skill types. These elasticities were also updated for the model.
In addition, data on the physical stock of capital (machinery, equipment, land and building) in each industry was also updated. This information was obtained from the tables on supply and use, along with information from Statistics New Zealand's capital stock and productivity series.
Finally, information on household consumption and government fiscal accounts was incorporated into the model. This information came from another Economic Impacts of Immigration project, namely the one on the fiscal impacts of immigration.[26]
2.5 Interpreting model simulations
The CGE model allows us to perform computer simulations to investigate the effect of particular events on the economy. For example, we could estimate the changes in major economic variables (eg, employment or real gross domestic product (GDP)) resulting from a:
- change in population growth, which affects household spending growth
- technological breakthrough that results in increased productivity in particular sectors
- world event (eg, political turmoil) that reduces the demand for our exports
- change in policy (eg, increased government spending on hospitals)
- change in the price of commodities (eg, milk solids or oil).
In the analysis the CGE model first needs to establish a base case to which the results of various scenarios can be compared. This means there is a constant point of analysis between various scenarios. The base case (or baseline) is sometimes referred to as a business-as-usual scenario and is essentially what would happen in the absence of any significant shock. The model also needs a 'base' year (or starting point) and a 'snapshot' year to be defined. In this study, the base year is 2006 and the snapshot year is 2021. Essentially, the study is modelling the effect of a shock, such as an increase in the inflow of immigrants, on the economy in 15 years' time (2006 to 2021).
Figure 2.2 shows how the CGE model results should be interpreted. The example of real GDP is used. First, the level of real GDP in the snapshot year (2021), noted as Y1baseline, consistent with a baseline scenario needs to be established.
Thereafter, the CGE model experiment proceeds by changing one (or more) of the assumptions that have been adopted to determine the baseline or control level of real GDP Y1baseline. It is best to change only one assumption at a time so the impact of that change can be understood. If multiple assumptions are changed, it is not possible to understand the individual impact of each change or the impact as a result of the interaction between the changed assumptions.
Figure 2.2 Interpreting a computable general equilibrium (CGE) model experiment
If the annual flow of immigrants were changed, then this is the 'shock' that is to be modelled. For such an experiment to be modelled, a variety of variables is likely to be changed to mimic the 'shock' being introduced into the model.[27] For example, the labour supply is likely to be different in the baseline compared with in the scenario. In addition, government consumption demand may be changed to reflect different, for example, education and health spending associated with the changed flow of immigrants in the scenario.
The result of the model's simulation (experiment) of the impact of immigration would be a measure of the difference between Y1scenario and Y1baseline; that is, the difference between real GDP with shock (changed flow of immigrants) and real GDP without shock (baseline flow of immigrants).
The model provides results for a wide range of economic measures (eg, labour employed, gross output by various sectors, exports by different commodities, and imports and consumer spending by commodity). Each of these results should be interpreted in a similar way to that depicted in Figure 2.2. An alternative way of interpreting the CGE model experiments is to view them as answering, 'what ... if ... ?' questions. For example, the question being answered would be, 'what is the change in real GDP and employment, if productivity in agriculture increases by x percent?'.
2.6 Generating a baseline scenario
As noted above, the model generates a baseline scenario. Such a scenario should be interpreted carefully and, in particular, should not be confused with a forecast. Such projections are entirely contingent on the assumptions adopted for the key variables used to underpin the scenario. Key variables for which assumptions are required to generate a baseline scenario include:
- technological changes being faced by the industry or occupation (eg, which types of inputs (occupation skills or equipment) are more likely to be used than others) - the more we can find out about this, the more robust will be the model projections
- export market demand - expected global events and trends and whether these will constrain or aid the expansion in overseas sales of New Zealand goods and services
- terms of trade - movements in the relative world prices of goods and services that New Zealand producers are competing with on the global market
- demographics - growth in population, number of households, the working age population, and the labour supply available
- the relative rates of return and the savings to gross national product (GNP) ratio - assumptions on the savings to GNP ratio are required to establish the availability of the productive capital stock (physical machinery, equipment, buildings) for use by industry in the projection year.
The detailed assumptions imposed for the baseline projection are presented in Appendix C.
2.7 Baseline scenario for this study
The baseline projection should be interpreted as a business-as-usual scenario. As such, many of the variables used are similar to those observed in recent years. Productivity and export market growth are assumed to be similar to the levels experienced over recent years. The productivity[28] assumptions for the baseline vary across the different sectors and vary between capital and labour productivity. On average, labour productivity is assumed to grow at an annual rate of 1.2 percent over 2006 to 2021. This is comparable with the average for 1994 to 2006 of 1.1 percent per annum. As for capital productivity, this is assumed to grow on average 0.6 percent per annum in the baseline projection. This compares with the annual average of 0.5 percent over 1994 to 2006.
Similarly, export market growth also varies across the commodities. It is assumed that growth for primary (eg, agricultural) commodities grows at a slower rate than that for manufactured commodities. In part, this reflects market access, as well as capacity, constraints for primary products. Tourism market growth is assumed to be slightly higher than manufacturing sector growth.
The 2021 baseline terms of trade (measured as world export prices relative to world import prices) is assumed to remain relatively unchanged from the 2006 level. However, the world prices of oil and related energy products are assumed to increase at a faster rate than the prices of other goods and services.
The baseline also assumes the average net change in the overseas-born population resident in New Zealand (annual net inflow) to be 20,000 per year. Taking into account flows of the New Zealand-born population, this assumption is equivalent to a net annual inflow of permanent and long-term (PLT)[29] migrants of 10,000. This assumption places the business-as-usual scenario in the context of the 1991 to 2006 New Zealand experience when the net inflow of overseas born averaged just under 24,000 per annum and that of PLT flows averaged just over 12,000 annually, as shown in Table 2.1. This takes the resident New Zealand population from 4 million in 2006 to 4.5 million in the 2021 baseline.
The composition of this net inflow (in terms of the number of couples and singles and number of households) is assumed to be similar to that experienced over 1991 to 2006. The skill mix of this net inflow is assumed to be driven by the demands of the economy (ie, it is fully model determined). Together with the ageing profile of the population and heightened labour participation rates, this takes the labour available in 2021 to 2.2 million full-time equivalents. This represents an increase of 1.5 percent per annum in labour supply over 2006 to 2021.
The model uses broad age compositions to calculate the number of people aged under than 15 and labour market participation rates. These estimates are used to calculate the size of the labour force and the size of the population under 15 years and not in the labour force. Bearing in mind the predominant age groups for health expenditure are the very young and the very old, the 'not in the labour force' group is used as a proxy for those aged 65 and over.
In generating the baseline, it is assumed that the national savings ratio (ie, the proportion of GNP income that is not spent on consumption) remains unchanged from its 2006 level. In addition, the average tax rate on household income in the 2021 baseline is assumed to be 19.1 percent compared with 21.1 percent in 2006.
2.7.1 Baseline economy for 2021
The baseline projects GDP growth at an average 3.1 percent per annum over 2006 to 2021, with full-time equivalent employment growth of the order of 1.5 percent per annum. As part of this growth, export growth averages 4.1 percent per annum (see Table 2.2).
This growth gains some support from a slight decline in New Zealand's real exchange rate, which improves the competitiveness of the country's exports relative to other international producers. Among New Zealand's export categories, the long-term trend away from commodities towards services continues. Thus, within this expansion the shift continues to services and value-added manufacturing. These sectors are projected to grow considerably faster than commodity exports on the whole.
Growth in tourism activity remains at the forefront (at 5.3 percent per annum). Primary commodity export volumes continue to grow over the projection period, albeit at a moderate pace for most industries (eg, wool, meat, and horticulture). Moderate productivity growth allows wages to increase in real terms; that is, average wage rates are projected to grow more quickly than consumer price inflation over the projection period.
With investment spending growing in line with overall GDP, consistent with the required expansion in capital stock, the import-to-GDP ratio reduces little in real terms. This restrains the improvement in the balance of trade over the period.
Full-time equivalent employment is projected to expand by approximately 425,000 over the 15-year period, which equates to an annual increase of approximately 28,300. Employment in primary industries, such as agriculture, is projected to remain static or decline. Growth in government investment and exports will drive ahead employment in the higher value manufacturing industries (eg, machinery and equipment manufactures) and the building and government sectors.
This model projection assumes the labour resource required is available; that is, the skill (or occupation) composition of the labour resource supplied is totally demand driven. As shown in Table 2.3, the labour required by the baseline 2021 New Zealand economy is led by a greater than average expansion in professionals, trades workers, and machine operators. Further detail of the model results indicate that the increase in the required number of professionals is concentrated in scientific, computer, engineering, and business professionals. This, along with the increase in machine operators, is consistent with the increase in the manufacturing industries noted above. Some in the trades workers category will also be required by the manufacturing sector, while the expansion in the building sector is also relevant here. At the other end of the spectrum, little employment growth in agriculture is reflected in the low number for the increase in primary sector workers.
Accompanying this employment growth, are capital requirements that expand an average 2.7 percent per annum over the projection period. However, the savings arising from the income over the period are insufficient to fund the increase in capital resources required. Consequently, net foreign liabilities increase to be equivalent to nearly 98 percent of nominal GDP in 2021.
Table 2.2 compares the projection for the two productive factors, labour and capital resources, and shows that capital stock expands faster than employment. This means an overall shift over 2006 to 2021 to a relatively more capital-intensive economy. This is clearly reflected in the primary sectors where output growth is achieved by an expansion of capital but little, if any, employment growth. In addition, some of the services sectors (eg, education and transport) as well as the higher-value manufacturing sectors (eg, machinery and equipment manufactures) record noticeable increases in demand for capital.
As for the government accounts, the tax revenue accruing from the 2021 income is sufficient to fund the spending in line with demographic and final demand projections. This results in the projected core Crown balance declining slightly, relative to the size of the overall economy, to be 5.7 percent of nominal GDP in 2021.
Footnotes
[17] The price at which the level of demand equals the level of supply in a particular market.
[18] Profit maximisation is the process of obtaining the highest possible level of economic profit through the production and sales of goods and services.
[19] The process or goal of obtaining the highest level of satisfaction of wants and needs obtained from the use or consumption of goods and services.
[20] A firm earning zero economic profit is doing as well by investing its money in capital as it could by investing elsewhere.
[21] Dixon et al (1982).
[22] A dynamic version of the model has also been developed (Nana, 1999) that enables the path of an economy over time to be modelled. Comparing a baseline path with a path that incorporates the response to a shock or shocks enables comparative dynamic (as opposed to comparative static) analysis to be undertaken.
[23] The detailed model structure closely follows Dixon et al (1982) and is described in Poot et al (1988).
[24] These use 2003/2004 data. The supply table shows the origin of the resources of goods and services, and the use table shows the uses of these goods and services and the cost structure of the various industries.
[25] The RAS method is a method used to update existing input–output tables to relate to a year for which intermediate input (column) sums are known but not the intermediate deliveries themselves. See Parikh (1979) for an overview of this method.
[26] Slack et al (2008).
[27] In technical terms, a set of variables is not determined by the model – such variables are termed ‘exogenous’. These variables must be set, or ‘shocked’, by the user, depending on the experiment or scenario being simulated. On the other hand, variables that are determined by the model are termed ‘endogenous’. The outcomes for these variables are obtained as a result of solving the model’s equations. This solution process occurs after the introduction of the shock, through changes to one or more exogenous variables, to the model.
[28] Productivity is defined as output per unit of combined labour and capital inputs used.
[29] PLT arrivals include people who arrive in New Zealand intending to stay for a period of 12 months or more (or permanently), plus New Zealand residents returning after an absence of 12 months or more. Included in the former group are people with New Zealand residency, as well as students and holders of work permits. PLT departures include New Zealand residents departing for an intended period of 12 months or more (or permanently), plus overseas visitors departing New Zealand after a stay of 12 months or more.