The Scottish Government

« Previous | Contents | Next »

Listen

Technical Appendix A3 Simulation results: impacts of demographic change

A3.1 Introduction

This section (as in the former) involves exploring and extending previous applications of the AMOS modelling framework in order to allow us to provide some detail on the properties of the model, and on what types of simulations and sensitivity analysis can be carried out. In this section we extend previous work done at the Fraser of Allander Institute on the impact of demographic change on the Scottish economy. In recent work for the Scottish Government, colleagues at the University of Strathclyde examined the economic impact of demographic change on the Scottish economy, through linking a demographic model with the AMOSCGE model for Scotland. The findings and results of this work are discussed in Lisenkova et al. (2008).

Here, we extend the previous analysis by using the AMOSENVI model, rather than the AMOS model, for a set of anticipated changes to the Scottish total and working age population consistent to those modelled in this previous work. The AMOSENVI model has a more sophisticated treatment of energy inputs and a set of linked environmental accounts for Scotland. This provides considerably more detail on the relationship between economic activity in Scotland and energy and environmental impacts, and allows us to construct and report environmental and sustainability indicators. Further details on the AMOSENVI model of Scotland, and the use of environmental and sustainability indicators can be found in Section 1 of this report and in Learmonth et al. (2007) and Hanley et al. (2008). In this section, we describe an application of the type of analyses that can be carried out, and the type of results that can be obtained from, using the AMOSENVI model to explore the impacts of demographic change on the Scottish economy.

While the literature on the economic impacts of demographic change are well researched, the literature on the environmental and energy impacts of demographic change is small, indeed an initial literature search found little directly relevant material. A number of studies linking demographic change to energy consumption and environmental indicators have been carried out. However these tend to be statistical relationships between demographic variables and energy/environmental impacts. York (2007) uses regression analysis to calculate the impacts of a number of factors - per capita incomes, population, urban population and population over the age of 65 - on energy consumption for European Union nations from 1960 to the present day. Their results indicate that both total population and the age of that population are important for total energy consumption, with a 1% rise in total population indicating a 2.665% increase in energy consumption, while the population aged over 65 also has positive effects on energy consumption. Using projections for each of these variables York (2007) presents predictions for total energy consumption for European Union companies by 2025. While not using projections to predict future emissions, Cole and Neumayer (2004) similarly include demographic factors in their analysis of the factors driving emissions of CO ₂ and SO ₂. They find that a 1% change in total population produces a roughly similar change in CO ₂ emissions, but that the marginal impact of total population on SO ₂ emissions is an increasing function of the level of population. For both CO ₂ and SO ₂ emissions, the age structure of total population does not give statistically significant results. From our initial reviews of the literature, there appears to be little work focusing on the energy and environmental impacts of anticipated changes in working age and total population on the labour market, and, through wages, to economic activity. We seek to make a contribution to this literature with the simulations reported in this note.

Previous work (Lisenkova et al., 2008) found that for Scotland, forecasted changes to the level and age structure of the population of Scotland will produce significant impacts upon the Scottish labour market, and the competitiveness of Scottish industries, which will include energy industries. Changes in the size of total Scottish output, the composition of that output across industries and the structure of production within industries will have impacts on energy demand and environmental impacts, including on emissions. In this note, we explore these impacts of anticipated population change using the AMOSENVICGE model.

In Section A3.2 below we follow material presented in Lisenkova et al. (2008) and very briefly set out the anticipated theoretical impact of demographic change on Scottish labour market. In a sub-section of Section A3.2 we set out one set of predicted changes to total and working age population in Scotland, and which form the basis for the inputs to the AMOSENVICGE model. We term these predicted changes our "Central" scenario. Section A3.3 sets out the simulation strategy we follow, while in Section A3.4 we set out the key economic, energy and environmental results. Section A3.5 reports the results of sensitivity analyses, including to the assumed structure of the Scottish labour market, three alternative scenarios for population change - which we term "High", "Medium-High" and "Low" - and variations in the values of key parameters within the AMOSENVI model.

A3.2 Theoretical impacts of population change on the Aggregate Scottish labour market and details of the empirical "Central" scenario

Theoretical analysis

In this section we provide the conceptual underpinning for the simulation results which follow. We direct our attention at the labour market, given than we are interested in the way in which an ageing and declining population produces changes in the labour market. We focus on the long run, so that the equilibrium wage and employment rate are those towards which the economy is being attracted over time.

Figure A3.1 uses an aggregate labour supply and demand framework to represent the Scottish labour market. The wage is the real consumption wage. Labour is mobile between sectors we do not attempt to consider no skill or geographic "sub-markets" within Scotland. We take a comparative static approach here to describe the stages through which exogenous demographic changes might impact upon the labour market. The labour supply curve in this market is assumed to be upward sloping. As real wages rise, more people are attracted into the labour force. The labour demand curve slopes downwards as for higher levels of real wages, Scottish output becomes less competitive, and with falling output (including exports), household incomes and consumption fall and there will be a reduction in the labour intensity of production, and so a fall in the quantity of labour demanded.

Figure A3.1: The Scottish labour market in 2000 and 2050 under alternative population projections

In Figure A3.1 we compare the long-run labour market equilibrium after fifty years of demographic change with that in 2000, under the assumption that total and working age population changes are as anticipated and that real per capita government expenditure in Scotland remains constant. The initial equilibrium is represented by point A, where the base period labour demand and supply curves intersect. This generates the initial equilibrium employment and real consumption wage level as given by w _n,2000, N ₂₀₀₀.

According to recent population projections for Scotland carried out by Government Actuaries Department ( GAD), over the next fifty years the population of Scotland will decline and age. These exogenous changes in population size and age composition will have an effect upon both the labour supply and demand schedules shown in Figure A3.1. Firstly, the fall in the working age population reduces labour supply at each real consumption wage level, generating an inward shift of the labour supply curve. This is shown on Figure A3.1 by the new labour supply curve NS ₂₀₅₀, which lies to the left of the original supply curve NS ₂₀₀₀.

The change in population also affects labour demand. We follow Lisenkova et al (2008) in assuming that real government expenditure per head remains constant, so that Scottish real government expenditure varies in line with the changes in total Scottish population. In the AMOSENVI model, any exogenous change in product demand shifts the labour demand curve in the same direction. The demand curve shifts by the extent of the exogenous employment change plus the appropriate Type II IO employment multiplier. Because total population is lower in 2050, labour demand is also lower at each real consumption wage rate. The labour demand curve, therefore, is shifted to the left (ND ₂₀₅₀) compared to its previous level (ND ₂₀₀₀).

The new equilibrium is at point B, the intersection of the general equilibrium labour demand and supply curves NS ₂₀₅₀ and ND ₂₀₅₀. Both the labour demand and supply shifts lead to lower employment, but the impact on the real wage depends on the relative size of the shifts. Our prior expectation is that the reduction in labour supply will be much greater than the reduction in labour demand. This will be due to, firstly, the proportionate fall in working age population being less than the fall in total population and, secondly, government expenditure being only one element of Scottish final demand. This would lead us to expect to see a tightening of the labour market and an increase in real wages.

In sensitivity analysis, we have three alternative scenarios in total, one of which is for greater population decline, and greater population ageing - i.e. a scenario consistent with lower net migration or lower fertility rates ( ceteris paribus). Under such a scenario, the fall in working age population would be greater again, with greater tightening of the Scottish labour market following a larger fall in labour supply than outlined above. We would expect a greater rise in real wages and fall in employment under this scenario. Also in sensitivity analysis we have two scenarios for higher population growth than under the central scenario. Under these, such as would be consistent with increases in net migration or higher fertility ( ceteris paribus), we might expect the opposite to be the case.

Under our two population growth scenarios, the Scottish working age population and labour force is going to be higher than that assumed under the "Central" scenario, meaning that there will be an outward shift in the labour supply curve. Labour demand will also shift outwards as total population is higher. Under these scenarios, we would expect the equilibrium level of employment to be higher. With a greater increase in labour supply than total population we would expect the labour market to see a fall in the wage rate.

Central total and working age population scenario

The AMOSENVI model is currently calibrated for a base year of 1999. Colleagues in the Fraser of Allander Institute have estimated a number of alternative projections for the Scottish population from 2000 to 2050. These use the same assumptions for key demographic parameters as are used by the Government Actuaries Department ( GAD) in their projections, but allow us to make annual projections, and create alternative projections, annually to 2050. For the simulations reported in this chapter, we therefore assume that our models base year of 1999 also represents the Scottish economy in the year 2000.

The demographic changes estimated by our projections are used as the exogenous disturbances in the model simulations which follow. The population scenarios used therefore differ slightly from those produced by the General Registers Office for Scotland ( GROS). We use a scenario of net migration to Scotland of 5000 per year as our "Central" projection ¹⁷. This is similar, although not identical, to the assumed rate of net migration in the "Principal Projection" for the Scottish population used by GROS. Assumed changes in the total and working age population for Scotland from 2000 to 2050 under our "Central" scenario are shown in Figure A3.2.

Figure A3.2: Percentage changes from base year for working age and total population under "Central" projection

In 2050, under our "Central" scenario, total Scottish population is 1.68% lower than in 2000, however this is a significantly older population than in 2000, with the working age population down 14.91%. We describe the assumed changes in three alternative sensitivity scenarios in Section A3.5.

A3.3 Simulation strategy

We follow the method employed in Lisenkova et al. (2008) in estimating the impact on Scotland of population decline and ageing. Beginning with the labour supply effect, there will be changes in the labour supply schedule as the fall in the working age population reduces labour supply at each real consumption wave level, implying that the labour supply curve would shift inwards. As the model is currently configured, we enter the changes in the labour force by means of a linear trend between 2000 and 2050, so that the change in the working age population over the 50 years is modelled as a linear reduction.

The labour demand effect is treated in the identical way to Lisenkova et al. (2008), where it is assumed that real per capita government expenditure remains constant, so that the level of government spending changes with the size of the Scottish total population. As noted in the earlier paper, this assumption is realistic since Government expenditure in Scotland is mainly financed through the Westminster Parliament and the experience of the Barnett formula over recent years is that per capita Government expenditure figures for Scotland have remained fixed relative to the level in England.

As in Lisenkova et al. (2008), any changes in the composition of government and household consumption demand which occur because of demographic changes described above are not considered in this analysis. The results presented here will be driven by general demand side factors, such as movements between public and private consumption as population structure changes, as well as supply-side factors operating through the tightening of the Scottish labour market and the impact of this on the competitiveness of individual sectors.

A3.4 Results from "Central" population projection with AMOSENVI

Aggregate economic impacts

We present the change in total and working age population under our "Central" population projection in Figure A3.2. The demographic data represented in Figure A3.2 are used to convert exogenous disturbances to labour supply and labour demand in the AMOSENVI model as discussed earlier. From running the AMOSENVI model for these exogenous disturbances to total and working age population, we get the simulation results summarised in Table A3.1.

The results in Table A3.1 should be interpreted as variations away from what would have occurred but for the changes in total and working age population. As expected following the earlier theoretical discussion, in the results for 2050 we see a fall in employment of 9.89% with a corresponding fall in GDP of 9.30 %.

Table A3.1: Percentage change of aggregate economic and demographic variables under the central projection, bargaining labour market closure

Two important points can be noted from the results in Table A3.1. Firstly, the fall in employment (9.89%) is less than the fall in working age population (14.91%). This suggests that there is an increase in the labour market participation rate, and a fall in the unemployment rate. The tightening of the Scottish labour market is clear from the 7.89% rise in real wages by 2050. Secondly, the decline in GDP closely follows the observed reduction in employment. The reduction in GDP is driven by the reduction in the labour force, and increase in real wages, causing a reduction in Scottish exports generated by the reduced competitiveness of Scottish output.

In 2050 the consumer price index is 2.93% higher, but the increase in the export price index (Competitiveness Index) is higher at 4.00%. As a consequence the demand for exported goods falls in the central projection by 7.55%. The capital stock will adjust to changes in output demand but this will occur more slowly than the change in employment in particular sectors so that the change in GDP will slightly lag the change in employment. There will also be a tendency for production to be more capital intensive given the increase in the nominal wage rate, so that there is some substitution of capital for labour.

Public consumption, e.g. by Government in Scotland, is exogenously shocked in line with total population, but private consumption, e.g. by households, is endogenous within the AMOSENVI model, and can give a useful indication of the welfare of Scottish households. By 2050, the fall in private consumption is 6.08% - less than the fall in GDP and employment. This reflects the increase in the real wage for those in employment. As in Lisenkova et al (2008), private consumption falls by more than the reduction in total population, meaning a decline in per capita private consumption.

Sectoral economic impacts

Looking at the pattern of sectoral impacts, note firstly that the sectoral disaggregation of the AMOSENVI model is different to that used by Lisenkova et al (2007). The AMOSENVI model is calibrated around a SAM for Scotland in 1999, and for a set of economic and environmental accounts built around a consistent sectoral aggregation. We are particularly interested in the energy and environmental impacts of population change. We begin by discussing the sectoral economic results.

Figure A3.3 shows that by 2050 the output of, and employment in, all sectors in the Scottish economy are negative affected. There is, however, wide variation in the impacts across sectors, with the output of 'Education' and 'Public and Other Services' sectors falling by 5.9% and 4.9%, while 'Coal Extraction' and 'Construction' see a decline in output by 2050 of 14.2% and 12.9% respectively. The sectors in which government demand is concentrated in the base year IO - 'Education' and 'Public and Other Services' - are least affected since government expenditure per capita remains constant over the period simulated, and in total falls by 1.68% by 2050 (in line with the fall in total population).

Figure A3.3: Impact on sectoral output and employment, % changes from base year values by 2050

The extent of the negative impact upon other sectors is determined by two factors. Firstly, labour intensive sectors are worst affected because of the now increased cost of labour. Second, the sectors which are more exposed to international trade feel the negative effects on competitiveness more strongly. For example, sectors such as 'Sea Fishing', 'Fish Farming', 'Oil and Gas Extraction', 'Chemicals' and 'Transport and Other Machinery' suffer these negative export competitiveness effects, with each of these sectors having exports constituting more than 80 per cent of sectoral output in the base year SAM. 'Sea Fishing', which is the most export intensive sector, sees the biggest decline in output of these sectors because is it also the most labour intensive.

Of the five energy sectors identified in AMOSENVI, the largest fall in output by 2050 (14.2 per cent) is observed for the 'Coal Extraction' sector. Of the five energy sectors, the output of the 'Oil Refining' output falls by the smallest amount, caused by it having the lowest employment intensity of all sectors in AMOSENVI. In all sectors, employment falls by more than output because as the price of labour rises, firms substitute capital for labour. Also, it takes more time to optimally adjust the capital stock.

Energy and environmental indicator impacts

Changes in the energy and environmental indicators can be seen in Figure A3.4. Looking firstly at GDP, we can see that under the central simulation for the change in total and working age population, GDP reduces by 9.30% per cent by 2050. As observed above, the output of each sector contracts by 2050 as competitiveness suffers, particularly for export- and labour-intensive sectors. The level of energy demands also fall as output declines, as shown by the two red lines in Figure A3.4. Electrical energy consumption (measured in GWh) and non-electrical energy consumption (measured in tonnes oil equivalent) fall by 11.09% and 8.63% respectively.

Figure A3.4 : Energy indicators, % changes from base year under the central population projection, bargaining labour market closure

The other indicators of sustainability, detailed in Figure A3.4, show mixed results. These two measures relate the amount of energy consumption divided by GDP, and use electrical energy and non-electrical energy as the respective numerator. Note in these measures that GDP is the numerator, rather than the denominator as in the 'CO2 intensity of Scottish production' measure. A positive change in these indicators therefore indicates a positive movement in sustainability of economic activity, while a negative change indicates the opposite. As mentioned above the fall in electrical energy consumption is greater than the fall in GDP, and so the GDP/electrical energy consumption indicator moves in a positive direction, indicating greater sustainability. On the second measure, the fall in non-electrical energy consumption is less than the falls in GDP, and so on this indicator, there is a negative movement showing a fall in sustainability.

Figure A3.5: CO ₂ emissions and CO ₂ intensity of production indicator, % changes from base year under the central population projection, bargaining labour market closure

Figure A3.5 shows the changes in GDP and CO2 emissions as well as the CO2 intensity of Scottish production. Emissions of CO2 are 8.76% lower by 2050, a smaller fall than the decline in GDP. This means that the CO2 intensity of production - defined as CO2 emissions divided by GDP output (£million) - shows a small increase, i.e. consistent with decreasing sustainability of output. The carbon intensity of Scottish output is rising; however this is due to the greater relative decline in output than decline in CO2 emissions by 2050.

A3.5 Sensitivity analysis

Alternative population scenarios

As in Lisenkova et al. (2008), the assumptions made about demographic parameters are important for the shape of the projected total population and working age population profiles for Scotland. The birth rate, male and female life expectancy and the rate of net migration will all be important demographic parameters for the scale of the economic impacts. Lisenkova et al. (2008) found that the demographic parameter with the biggest economic impact is the assumed rate of net migration.

For the simulations in this project, three alternative scenarios for Scottish population change have been modelled. taken two extremes of population change for Scotland to demonstrate the usefulness of the modelling approach to understanding the dynamics, through the Scottish labour market, of changes in total and working age population. The variants are consistent with the central scenario, but in each scenario one of the demographic parameters has been adjusted. We label these three scenarios "High", "Medium-High" and "Low" respectively. In the first of these - the "High" scenario - the rate of net migration is revised upwards from 5000 per year to 30000 per year. The second ("Medium-High" scenario has the rate of net migration at 20000. The final ("Low") scenario keeps the rate of in migration constant at 5000 but lowers the birth rate from 1.65 births per woman to 1.45 births per woman. The profile for total and working age population under our Central, High, Medium-High and Low population scenarios are shown in Figure A3.6.

Figure A3.6: Percentage changes from base year values for working age and total Scottish population under four population scenarios

Under the "High" scenario for Scotland, total and working age population is higher in 2050, up 26.3 and 16.4 per cent respectively compared to 2000, while under the "Medium-High" scenario, total and working age population in 2050 is lower than the "High" scenario, up 15.3 and 3.9 per cent respectively compared to 2000. The "Low" scenario, total population in 2050 is 13.1 per cent lower than in 2000, while working age population is 26.2 per cent lower.

In Figures A3.7 to A3.13 we present the changes in GDP, employment, real wage, consumption, CO ₂ generation and electrical energy and non-electrical energy demands for these three alternative population scenarios.

Beginning with GDP and employment figures, Figure A3.7 and Figure A3.8 show that for the "High" scenario, GDP and employment are higher in 2050 by 11.2 and 12.7 per cent respectively, compared to the base year, and the "Medium-High" scenario also shows increases in GDP and employment relative to 2000. In the "Low" scenario, GDP and employment fall by around double the fall seen for the "Central" case, by 18.5 and 19.9 per cent respectively. As with the results presented for the "Central" scenario above, the mechanisms driving these results stem primarily from the labour market and the subsequent impact on the competitiveness of Scottish industries.

Figure A 3.7: Trends of Gross Domestic Product for "Central", "Medium-High", "High" and "Low" population scenarios

Figure A3.8: Trends of employment for "Central", "Medium-High", "High" and "Low" population scenarios

Figure A3.9 shows the changes in the real wage under the three population growth scenarios. It is the tightening in the regional labour market in response to the changes to working age population which drives changes in the real wages and the competitiveness of production, leading to the falls in output discussed above and shown in earlier figures. In the "High" scenario, where working age population and total population in 2050 are significantly higher than in 2000, these pressures are not seen. Conversely, in the "Low" scenario, the wage increase is much greater than the "Central" scenario.

Figure A3.9: Trends of real wages for "Central", "Medium-High", "High" and "Low" population scenarios

The impact of the alternative population scenarios on Scottish (private) consumption are shown in Figure A3.10. Under the "High" scenario, the change in total consumption in 2050 is positive, rising by 10 per cent. Consumption is also greater under the "Medium-High" scenario, but is lower in both the "Central" and "Low" scenarios. As in Lisenkova et al. (2008), the variation in consumption is less than the variation in GDP.

Figure A3.10: Trends in (private) consumption under "Central", "Medium-High", "High" and "Low" population scenarios

Figure A3.11 shows the impact of the alternative population scenarios on Scottish CO2 emissions. Under the "High" population scenario emissions of CO2 increase by 9.76 per cent, compared to an 8.76 per cent reduction observed under the "Central" scenario. With the "Low" population scenario, CO2 emissions fall further (down by 16.9%) due to the greater fall in economic activity.

Figure A3.11: Trends in CO ₂ generation under "Central", "Medium-High", "High" and "Low" population scenarios

The movements in the "CO2 intensity of production" indicator are shown in Figure A3.12. As we have a greater increase in GDP than CO2 emissions for both the "High" and "Medium-High" scenarios (see Figure A3.7 and Figure A3.11), here this indicator moves in a downward direction, consistent with increasing sustainability of economic activity. Although the decrease in the CO2 intensity of production is small in both cases - 0.7 per cent and 1.4 per cent in the "Medium-High" and "High" scenarios respectively - this is an important finding. Recall however, that total CO2 emissions are 2.9 and 9.8 per cent higher in these scenarios.

Figure A3.12: Trends in CO ₂ intensity of production indicator under "Central", "Medium-High", "High" and "Low" population scenarios

Roughly proportional falls are observed in electrical energy and non-electrical energy, as seen in Figures A3.13 and A3.14.

Figure A3.13: Trends in electrical energy demand under "Central", "Medium-High", "High" and "Low" population scenarios

Figure A3.14: Trends in non-electrical energy (tonnes oil equivalent) demand under "Central", "Medium-High", "High" and "Low" population scenarios

Alternative labour market structures

In order to test how sensitivity the results are to variation in assumptions about the nature of the labour market, we conducted simulations for the same population changes, but under two limiting case. Firstly, we assumed within-period fixed labour supply. This means that labour supply is a given proportion of the labour force, where the labour force is adjusted period-by-period through demographic changes. There is therefore assumed to be no adjustments in unemployment, or the participation rate, as the labour market tightens. This implies a wage curve that is infinitely elastic. In any individual time period therefore, this is represented by a vertical labour supply curve.

The second alternative assumption is that the labour market is characterised by excess capacity, such that any changes in labour demand can be met by a corresponding change in the level of employment, but with no upward pressure on the real wage. In this case, in each time period the labour supply curve would be horizontal, and employment adjusts in labour demand through changes in the unemployment and participation rates. We saw the largest declines in GDP above in those scenarios in which real wages increased as a result of anticipated demographic change.

In the fixed labour supply case, we expect the employment reduction and the increase in the real wage to be larger than under the bargaining scenario. In the fixed real wage scenario, the opposite results should hold: we expect employment to fall and the wage increase to be less than in the bargaining scenario. However, when we run the model with a fixed real wage, the model fails to solve for a long-run equilibrium, and stops when unemployment rate falls to zero - although the dynamics of the model are wanting the unemployment rate to continue to fall. As noted in Lisenkova et al. (2008), the population constraints implied by the "Central" projections, combined with fixed per capita government expenditure, must cause real wages to rise. The fixed real wage scenario is therefore not feasible.

When labour supply is completely inelastic on the other hand, the whole adjustment to the labour force contraction comes through higher wages. Employment and output falls by more under this closure than under the bargaining closure as participation and unemployment rates remain fixed - i.e. firms are not able to substitute capital for now more expensive labour. In Table A3.2 we present the percentage changes in the main aggregate indicators under the fixed labour supply and fixed real wage labour market closures. When the model fails to solve under the fixed real wage labour market closure, we mark these cells with an asterisk.

TableA3.2: Percentage changes of aggregate economic and demographic variables under the "Central" population projection, for Exogenous Labour Supply and Fixed Real Wage labour market specifications

With a fixed labour supply, employment is predicted to fall by 14.91 per cent, identical to the assumed reduction in the labour force and working age population. Under the bargaining closure, employment only reduces by 9.89 per cent (see Table A3.1). GDP is lower by 14.22 per cent (9.30 per cent in the bargaining case) and a real wages are 13.06 per cent higher (7.89 per cent higher in the bargaining case).

By 2050, under the fixed labour supply specification, the fall in economic activity and employment is manifested through a greater fall in CO2 generation than under the Bargaining case for the "Central" scenario. CO2 generation falls by 13.10%, however the greater fall in GDP (14.22%) means that the CO ₂ intensity of output increases. CO2 emissions do not fall by as much as Gross Domestic Product is predicted to fall, and so the environmental impacts of economic activity worsen.

Energy demands, on the other hand, show a varied response under the exogenous labour supply case. Reductions in electrical and non-electrical energy demands are observed, and these are larger than the reductions in these variables under the bargaining scenario, as would be expected given the greater increase in the real wage and reduction in employment and activity. By 2050, electrical energy demands are 16.97% lower than the base year, while non-electrical energy demands are down by 12.24%. Electrical energy demands have fallen by more than GDP, so the GDP/energy demand indicator with electrical energy demands in the denominator shows a positive movement - consistent with increasing sustainability. Non-electrical energy demands, however, have fallen by less than GDP so the GDP/non-electrical energy demand indicator decreases, indicating negative movements in sustainability.

Alternative parameter values

Sensitivity to elasticity of substitution between labour and capital

One key parameter is likely to be the substitution elasticity between labour and capital in the production of value added at the sectoral level. In previous simulations, this parameter was constant for every sector in each simulation, at 0.3. There is a wealth of recent empirical work concerned with estimating the appropriate value for this parameter. In this subsection, we impose alternative values of 0.8, 0.999999 (approximately assuming a Cobb-Douglas function) and 1.2. We expect that as we make substitution between labour and capital easier ( i.e. impose higher values of this elasticity), employment will fall more rapidly compared to our previous simulations.

Table A3.3: 2050 results for sensitivity analysis for elasticity of substitution between labour and capital, bargaining labour market specification under "Central" population scenario

The main aggregate economic, energy and environmental results for these three alternative values of this parameter are shown in Table A3.3. As expected, when there are higher values of the elasticity of substitution between labour and capital, employment falls by more (down by 10.69% in 2050 for an elasticity of 1.2). To the contrary, the greater ease of substitution between labour and capital means that the wage rate does not increase by as much (up 6.19% in 2050 for the elasticity of 1.2). The fall in GDP seen in the previous "Central" simulation, (9.30% by 2050) is less under higher values of this elasticity, down 9.17%, 9.01% and 8.85% for elasticities of 0.8, 0.999999 and 1.2 respectively. Thus, even large changes in the value of this elasticity have small impacts upon the aggregate economic indicators.

CO2 generation is lower for all three sensitivity simulations carried out compared to the simulation with an elasticity of 0.3. CO ₂ generation falls by more than the decline in GDP for values of this elasticity greater than 1, meaning that the CO ₂ intensity of production falls. This is, however, again associated with a lower level of economic activity and employment. Energy demands on the other hand, show a mixed result. As the elasticity of substitution between labour and capital is increased to 0.8, there are lower electrical and non-electrical energy demands. For values of 1 and 1.2, electrical and non-electrical energy demands still fall, but not by as much as where the elasticity of substitution is 0.8. The non-linear relationship between the elasticity of substitution and energy (electrical and non-electrical) demands suggests that this could be an interesting area for future research.

Sensitivity to elasticity of substitution between value added and intermediate inputs

Another key parameter is likely to be the substitution elasticity between value added and intermediate inputs, for the production of gross output of each sector. In the previous simulations, this parameter was held constant at 0.3 in each sector. In this subsection, we change this parameter across values of 0.1 to 1.2. We would expect that increasing the elasticity of substitution from 0.3 to higher values would lead to greater substitution away from more expensive value added (given higher real wages) and towards intermediate inputs. Employment will likely be lower for higher values of this elasticity, which is also likely to produce larger falls in GDP than simulations with lower elasticities. The converse is likely to be produced by simulations where this elasticity is lower than 0.3.

Table A3.4: 2050 results for elasticity of substitution between value added and intermediate inputs sensitivity analysis, bargaining labour market specification under "Central" population scenario

The results from varying this parameter are shown in Table A3.4. The results change in line with our prior expectations, with higher elasticity of substitution generating larger falls in employment, and so smaller increases in the real wage than under simulations where this parameter takes a lower value. With this elasticity at 0.9, employment falls by 10.14% (compared to a 9.89% for this scenario previously) and the GDP impact is greater with a decline of 9.50%. CO2 generation falls across all sensitivity simulations, but across all simulations falls by less than GDP declines, thus the CO2 intensity of Scottish production (CO2/ GDP) increases, showing a declining sustainability of output.

Electrical energy and non-electrical energy demands fall by 9.89% and 7.69% respectively, smaller declines than under the scenario for lower elasticities of substitution. The GDP/energy indicators are lower than where the values of this parameter are lower. Positive values of these indicators suggest that economic activity is becoming more sustainable. For our previous results, the GDP/electrical energy indicator showed a positive movement, while the GDP/non-electrical energy indicator moved in a negative direction. As we increase the elasticity of substitution, both indicators move towards and become further negative respectively. For values of the elasticity greater than 1, both GDP/energy indicators show an absolute decline in sustainability.

« Previous | Contents | Next »