• No results found

Initial result analysis at small area level of wards

Chapter 4 Development of the dynamic spatial microsimulation model

4.8 Initial result analyses

4.8.2 Initial result analysis at small area level of wards

The aggregated results of the MSM are useful in terms of facilitating the strategic planning or policy making. However, at the tactical level, local context plays an indispensible role in area based intervention measures. Therefore, this study assesses the initial results by small areas to explore the spatial differences.

4.8.2.1

Population pyramids for small areas

The simulation results over 30 years for all 33 wards have been analysed and a substantial difference is revealed between the small area projections of Cookridge and Headingley. In Figure 4.16, the population changes over time in wards have been presented in population pyramids: Pyramids in darker shade use the simulation results in year 2001 and pyramids in lighter shade use the simulation results in year 2030. Results from any year between 2001 and 2031 can be used in this analysis. Here results in 2030 are used to demonstrate the changes close to the end of the simulation.

Cookridge

Headingley

Figure 4.16 The age–sex structure of ward population: Cookridge and Headingley in 2001 and 2030

Source: Author’s computation using results from this version of model From the small area population pyramids illustrated in Figure 4.16, we can clearly see that characteristics of the local population evolve differently in small areas. In the projections in 2030, we can see the ageing trend of the local populations. The majority of the population in the ward is aged over 40 and there is a significant increase of the ages over 90. However, Cookridge seems to have a more serious problem than Headingley, where the population has stopped growing and there is a much smaller population in Cookridge in 2030 than in 2001.

The population in Headingley, on the other hand, keeps growing into a 146

much larger population. The main reason behind this may be that Headingley has a much larger number of younger population aged 20-30 and under 10 in 2001 (Figure 4.16). Such composition allows the Headingley population keep growing. However, it needs to be noted that this version of MSM models Leeds population within a closed system and there are no housing stock constrains for the areas. Therefore the MSM allows university students to stay and grow old in area after higher education. This has exaggerated the Headingley population growth and the effect of ageing in both wards. Omitting migration to and from the Rest of the UK outside Leeds and to and from the Rest of the World outside the UK is the main limitation, but other reasons are also possible. Such reasons and various issues with the initial model will be discussed and addressed in Chapter 5, 6 and 7.

4.8.2.2

Location Quotient Analysis in small areas

To explore more of these spatial differences and in a wider context, a Location Quotients (LQ) analysis has been used to assess the spatial variance in the local populations. The LQ technique is one of the most commonly utilised Economic Base Analysis methods (Wikipedia, 2010). The LQ technique is frequently used in locational analysis, economic geography, and population geography, but it has been applied to a much wider range of studies. The LQ technique compares a certain local characteristic to a reference characteristic, in the process attempting to identify specialisations in the local area. The location quotient technique is based upon a calculated ratio between the local and the reference unit. Generally speaking, the LQ is an index for comparing an area's share of a particular characteristic, e.g. industry concentration, demographic features, with the area's share of some basic or aggregate phenomenon. It therefore provides a way of quantifying how concentrated a particular industry, cluster, occupation, or demographic group is in a region as compared to the nation. It can reveal what makes a particular region “unique” in comparison to the national average and is useful for calculating and mapping relative distributions (Desai et al., 2009; Wikipedia, 2010).

The initial results from a selection of area groups that have similar characteristics have been assessed using LQs as measures of the concentration of a particular group within each geographical area at a point of time. The population structure is analysed in five year age bands. Where the location quotient is greater than 100, this indicates an over- representation of the population group in area, and vice versa. Some examples are shown in Figure 4.17.

Figure 4.17 Location quotient analyses of ward level projections

Source: Wu et al. (2008)

The pattern of changes has been found to be consistent in two types of small areas. Here results from four wards are presented: Aireborough and Cookridge represent the suburban areas and Headingley and University to represent city centre student accommodation areas. Aireborough and Cookridge are both established suburban areas in the northwest of the city. In these areas, the changing concentration of demographic groups over time looks reasonable. For example, in Cookridge there is initially an expansion in the very elderly population, but later the quotient for the elderly falls relative to the rest of Leeds as other areas begin to experience a similar 148

growth in the older age groups. However in wards where student migration has a great impact like University and Headingley, the MSM failed to reproduce the student population renewal. The peaks of young people aged 20–25 disappears after 10, 20 and 30 years simulation. The distinctive population structure in such areas has been lost in the small area projections. This indicates that the subtlety of the local migration patterns has not been captured successfully in our MSM (microsimulation model). The pure spatial MSM used here cannot differentiate students from other migrants in the migration process. The MSM is probability driven. Migration probabilities at small area scale have been generated using the 2001 Census Special Migration Statistics (Level 2) data. The data provides us the ward- based migration flows from one ward to another for migrants, but we cannot determine the number of the university students within each flow. Therefore student migrants are modelled exactly the same as the rest of the migrants in the pure MSM using the generalised probabilities.

Another factor affecting the results could be that the small area migration probabilities are not disaggregated by age. However, even if we refine the model with disaggregated ages, it still cannot capture the distinctive migration pattern of student migrants. In reality students only move around the area close to the universities where they study, not in the suburban areas. More importantly, most of them will leave the city once they finish their study, instead of settling down and growing old in the area. Due to the replenishment of the student population each year, the population of the wards in which university student stay tends to remain younger than that in other wards. Another issue with student migrants is that they often confuse their answers to term-time and home address in the census. Chapter 6 will discuss such issues in detail.

Given the above reasons, a hybrid modelling approach is adopted to better model the local migration patterns. In chapter 6, a hybrid approach combining MSM and ABM (agent-based model) techniques will be explored.