The Scottish Government

« Previous | Contents | Next »

Listen

DEVELOPING A METHODOLOGY TO CAPTURE LAND VALUE UPLIFT AROUND TRANSPORT FACILITIES

3. VALUE CHANGE MEASUREMENT METHODS USED

3.1 introduction

There does not seem to be any accepted methodology for assessing value change associated with transport infrastructure, and this is due in part to the complexity of the issues under analysis and partly due to the lack of suitable time-based data. This conclusion can be illustrated by reference to the Sherry Ryan (1999) review paper. On the evidence, Ryan refers to (1999:412) "Four decades of inconsistent findings" and (1999:413) "the inconsistent and perplexingly weak evidence of the transportation-property value relationship." Ryan concludes (1999:423) "the value of the properties where residents or firms are located will be bid up if travel time saving accrue to residents or firms". On methodology, Ryan notes (1999:418) "Regression analysis is used in all but one of the heavy rail transit studies [in North America]. ... there is considerable variation in other aspects of heavy rail studies, including the size of study areas, the length of the study periods, and the independent variables used to estimate property values." For example, some studies use access measured by distance, and others by time.

Ryan also notes the importance of accessibility and travel time (1999:422) "…it is more likely that the expected relationship between access and property values will be observed when an analyst's measure of access captures variation in actual travel time savings", and this point is reiterated in the conclusions (1999:423) "when transportation facilities provide travel time savings, and these savings are accurately measured, property values tend to show the theoretically expected relationship with transportation access." Ryan stresses again (1999:423): "A study that does not accurately measure changes in travel time will not accurately estimate property value changes". On key research questions, Ryan suggests: (1999:414): 'The real questions are, "Has this transportation facility improved travel times for travellers," and if so, "which travellers?" If the facility improves travel times, then we should expect those travellers who experience travel time savings to bid up property values. If the facility does not improve travel time, then we should not expect a land-market response."

In a sense, Ryan is arguing that all of the benefits are internalised through the transport time dimension and that there is no reason to investigate further into the property market effects. This is similar to the classic Mohring (1976) argument where it is assumed that all benefits are captured by the reduction in transport costs and that any additional benefits would amount to double counting ( see Footnote 1). The converse argument is that there may be additional benefits if, and only if, it is possible to demonstrate the existence of positive externalities, such as property and land value increases (Banister and Berechman, 2000:168).

Other literature notes the lack of clear evidence on the influence of transport on land use and development (in general). For example, Atkinson (1988) stated "It is difficult to measure the degree of development and economic benefit that is generated by transit, as some of it may have taken place anyway, and it is even more difficult to measure the wider economic benefits…". Walmsley and Perrett (1992:114) concluded "The extent to which a rapid transit system can directly influence employment is difficult to quantify. The question is whether these opportunities would have resulted if the system was not developed at all, and whether new jobs are created or are redistributed from other areas, thereby not increasing the overall level of employment opportunities." They reviewed a few US cases finding that these were inconclusive or there was no causative effect discernible, and that "the effect of rapid transit on house prices is difficult to determine, and any effect cannot definitely be ascribed to the rail system".

The overall picture given above is one of uncertainty and complexity, but this is not a reason for despondency. To give some structure to the methods that have been used, it is useful to consider the scale of analysis. At the macro level, there are methods that relate economic growth to transport investment, but these are not a concern here (see Banister and Berechman, 2000). At the meso level, the concern is primarily over labour markets, productivity, inward investment and the quality of the labour force. In the next section, we discuss some of the relevant literature (see Llewelyn-Davies and UCL, 2003). It is at the micro level that most interest is directed as it is at this scale that the effects of transport investments on property and land values can be measured, and an extensive review of the methods used is given here (Section 3.3).

3.2 The Meso Level

At the meso level, Linnekar and Spence (1992:45) conclude their paper on the M25 accessibility impact with the comment: "As to exactly what the implications of these findings are for the wider issues of sub-national development is not very easy to see. Clearly this major investment must be having some effect… But the mechanisms through which these accessibility dynamics are translated into those competitive margins of business which are so crucial to understanding development potential are as yet unclear."

Grieco states (1994:29) that:

• The relationship between transport investment and economic development is in need of more detailed and systematic research;
• The materials that exist do not allow us to directly address the issue of the transport investment preferences and perceptions of benefits amongst inner city residents and businesses [the subject of her study].

Dabinett also quotes Grieco alluding to 'limits to our understanding', 'dearth or material', suggesting that there is very little high quality evidence available on transport induced change. He also (1998:173) quotes Cervero on the tenuous relationship: "In isolated settings, rail transit seems to have produced value gains and induced growth. However, the circumstances that brought about these changes are not easily generalisable." Also, in discussing land use change Dabinett refers to problematic attribution and cautious interpretation.

The picture given here and in Section 2.5 is rather mixed, with some researchers questioning the need for any additional analysis (e.g. Ryan), whilst others suggest that there are additional non transport benefits, but there are no accepted analytical approaches (e.g. Banister and Berechman, Cervero and Vickerman et al). The second conclusion is that most of this research has concentrated on the meso level of analysis, looking for employment effects and other changes such as inward investment. Very few have taken the micro level to explore the effects on the local area around stations in terms of the property market effects.

In summary, at the meso level, prioritising objectives and criteria for project appraisal constitutes a set of issues that decision makers need to be very clear about. A three-step procedure has been proposed (Banister and Berechman, 2000):

• The majority of benefits need to be transport related, since otherwise why invest in transport facilities in the first place. Cost benefit analysis should remain the key method for transport evaluation.
• The need to avoid double counting in measuring non-transport benefits must be clearly recognised, and explicit measures taken to highlight situations where this occurs (see earlier discussion about allocative externalities).
• The need to show functional linkage between primary transport benefits (e.g. accessibility improvements) and potential economic development effects should be demonstrated on a project by project basis.

If transport investment is to take place, a twin approach could be adopted where conventional cost benefit analysis is carried out on the project to determine the user benefits and costs of investment. To achieve a given rate of return, this analysis may account for some or all the necessary returns. If there is a shortfall, then a complementary analysis needs to take place that takes a wider view of the investment proposal. It could be argued that this complementary analysis should become an integral part of all evaluation, not just where the transport analysis fails to meet agreed criteria. It would further develop the current best practice being operated by the Department for Transport in its New Approach to Appraisal (DETR, 1998) and the more recent DfT report on preparing Economic Impact Reports (2003). For further details see Banister and Berechman (2000:326-329).

3.3 The Micro Level

It seems that there might be potentially greater returns if more analysis was directed towards the micro level, as it is at this level that the local effects of transport investment should be measurable in terms of property and land market effects, and in terms of the environmental and distributional impacts. But even here, there seems to be no comprehensive research methodology in this area. A proposed methodology has been developed (ARW and UCL, 2002), and it is currently (2003) being tested in the Croydon Tram corridor in South London. It covers the necessary conditions for these additional impacts to be measured over time and to relate these changes in property prices to distance from the tram corridor. It is the additionality (or latent demand) and measurability of these benefits that need to be analysed (Table 5).

Table 5: Methods Used

Notes: The results are hard to compare and generalise. Some key points include:

• The studies are all measuring effects on different kinds of land value and property impacts. Surveys include business interviews, focus groups with stakeholders, attitudes to the investment, and in some cases local area surveys of vacancies.
• There are different measures of proximity (distance or time). Distance may be measured as the crow fly distance or network distance. In some cases, the affected zone is described in terms of particular land tracts or neighbourhoods. Different thresholds are used (measured in different units, metric or imperial). Sometimes simply terms such as 'near' or 'adjacent' are used.
• Monetary values can be absolute prices (e.g. per house) or rentals (per sq m or per sq ft, per month or other). The prices may be in pounds or dollars or cents.
• Increases or decreases may be absolute (e.g. $4000) or in percentage value.
• What the price/value impact is relative to is also important - is it the same thing at a different year, or a different thing (e.g. house further away) at the same time, etc.

The main methods used in the studies as listed in table 5 are outlined below, both through a short description of the method and an example of the use of the methods (in boxes).

1. Accessibility and proximity levels and changes in these levels: this can best be operationalised through travel time thresholds with the use of a distance decay function, as the greatest impact is likely to be found closest to the public transport node. Two other issues are important here. It would seem that the distance thresholds are different for residential and commercial developments, with impact distances being larger for the former than those for the latter.

Evidence from the literature review has produced a variety of thresholds. For example, Riley (2001) has used 400yds, 800yds and 1000yds, Chesterton (2002) and Hillier Parker (2002) both used a 1km threshold, and studies in Tyne and Wear (200m), Helsinki (500-750m) and Toronto (<500m) used variable thresholds.

Secondly, the impacts may be different on existing developments as compared with new developments, as it takes time for markets to react to change. In addition, more than one location (for example in a corridor) might have an accessibility improvement as a result of a transport investment. If accessibility to a location is significantly improved, the labour market catchment area will increase, and the property and land values will also increase in some proportion to that increase in the size of the catchment area.

For example, in the Hillier Parker (2002) Crossrail study a 2km radius was considered appropriate in relation to more peripheral stations where residential land use dominated, compared with a 1km radii in more central locations which had a greater prosperity for employment land uses.

ACCMAP is a GIS based software package that links the Ordnance Survey OSCAR base mapping of the road network with a database of public transport information to give accessibility surfaces reflecting isochrones, which can be presented as changes over time. This approach was used in the Croydon Tramlink study (ARW et al, 2003) to map out changes in accessibility in the corridor before and after the Tramlnik's introduction.

2. Market Activity and Business competitiveness surveys have been central to many of the projects within the ESRC Cities Programme (Fainstein, 2002), and they have been used in assessing "impressions" of property market effects. Care should be exercised in terms on understanding the sampling process (how the firms were selected) and size of sample (often small). Comparisons made over time and location can also cause difficulties with interpretation. But most important is the difficulties of actually placing any confidence in the outputs. They should be used in conjunction with other methods as supporting or adding case study material to the more systematic analysis.

The Chesterton (2002) study uses this methodology, but came up with many problems as described above. They sought to improve the quality of the data sets used for the analysis, including the use of a longer time series (Q1-1997 to Q4 2001 as opposed to Q1 1999 to Q4 2001). Results from the study showed that price rises in residential property (flats & maisonettes) outperformed property in the reference areas.

3. Qualitative analyses, such as environmental quality audits and other analyses provide complementary information to help balance the more quantitative economic information and model outputs. Such analysis can include accessibility mapping, proximity analysis (distance to local facilities), and catchment areas for a range of activities (for example employment). But the real value of qualitative studies is in the interpretation of results and the resolution of attribution. It acts as a "softer" interpretation of the quantitative data.

This approach has formed part of the ARW et al (2003) study of the Croydon Tramlink where qualitative analysis has complemented the quantitative analysis through questionnaires and discussions with key stakeholders to interpret the results from the GIS analysis.

4. Descriptive statistics give a cross sectional view on what is happening in terms of property market and land value effects, and these factors can be cross tabulated with measures of change in travel and modal split to establish whether there is a statistical relationship (e.g. a correlation analysis). Such data helps to build up a picture of what is going on, but extreme care must be exercised in determining whether there is any implied causality between the variables. Even though there may be statistical association, this does not imply causality. In addition, property market impacts to cover the non transport impacts could be tabulated to include rent levels, land values, ownership patterns and land availability. This would provide a useful context for interpretation of impacts. A series of indicators of change should be developed to monitor these factors over time.

The methodology was developed by Chesterton (2002) with two main objectives in mind:

• To identify the dimensions of change in terms of the full range of impacts on both residential and commercial property markets;
• To initiate a time series, so as to assist in the attribution of cause and effect.

The JLE corridor was split into four market areas for analytical purposes: City Fringe (Waterloo, Southwark and London Bridge); East of City (Bermondsey and Canada Water); Isle of Dogs (Canary Wharf); East London (Canning Town, West Ham and Stratford). The report uses radii of 1000 metres for 'catchment areas' and 3000 metres for 'buffer areas' - the outermost area of geographical influence - the intention here being to pin down any 'decay effects'. Data sets vary: those that are publicly available tend to be of poorer quality, while the high-quality data sets tend to be commercially confidential.

For the residential study, data was sourced directly from the Valuation Office Agency, and proved sufficiently robust for use in the study.

For the commercial study, however, the VO data set was deemed incomplete, and only 'beta' (unofficial) sources were used.

The qualitative methodology drew on information taken from articles in the press, research reports and property agents' perceptions of the market. Property Agents perceptions were gathered through a postal questionnaire survey. Individual catchments were integrated for the purposes of the final analysis. The point is made that the JLE itself contributes to a redefinition of existing property markets, and that consequently the qualitative assessment has been used to bolster this analysis.

It is important to understand the dynamics of the land value and land development process in this context, since land value uplift associated with new transport schemes will involve decisions on the nature of the development in the vicinity of the scheme. There is a 'virtuous circle' that can be described in the context of land use and land value that goes along the lines:

High land prices do not, in themselves, cause high property prices. Both are caused by restrictions in supply but ultimately high property prices feed back to high land prices. Generally speaking a developer will assess what he can afford for land after other costs and profit are taken into account (the residual land value). On this basis the direction of causality flows from property prices to land prices.

The nature of the development will have a significant impact on land value. A good example of this is in Edinburgh at the moment, where high residential values have resulted in high residential land values. Industrial developers cannot compete and are having to move out of the city to land where there is no competition for use and hence land values are not bid up.

The conclusion is that land values vary according to the type of development sector and care has to taken to eliminate the development factor from land value uplift estimates. We need to make sure we are comparing like with like.

5. Regression analysis is a formal technique for quantifying or establishing a relationship between different sets of data. A regression analysis is performed on the information available on property transactions where property price is the dependant variable and the vector of physical and neighbourhood characteristics are independent variables. The results of the regression then provide information on how much change a given property attribute would affect the price of the property, as well as the overall explanation of the variability in the data.

Regression analysis (and hedonic pricing) has been used in most of the North American studies, but there has been considerable variation in the size of the study area, the length of the study period, and the independent variables used to estimate property values. The outputs have also been variable, thus making comparison between projects difficult.

6. Hedonic pricing was originally developed in the United States in the 1950s. Since then, hedonic models have often been used to account for the prices of heterogeneous goods, including real estate rents and capitalised values for land and buildings. In this context, heterogeneity means that the properties of one good can differ markedly from the properties of another, but that they are seen to be members of the same product group. The hedonic model seeks to explain the price for which something sells as a function of the characteristics it contains. On the basis of recent prices that have been paid for property, a regression analysis is used to calculate the proportion of the total value accounted for by each of a property's individual features. Hedonic pricing is very demanding on both assumptions and data. Even where the most thorough analysis has been carried out, the results are not clear.

Cervero and Duncan (2002a and b) in their study of commuter and light rail impacts examined the impact of four vectors on the estimated price per square foot of individual "parcels" of land. It is interesting to note that they argued that the accessibility benefits were capitalised into land prices and not buildings. The four vectors related to transport (using a measure of proximity of under 400m), the neighbourhood (using measures of mixed land use and the median household income), the location and regional accessibility (using access to jobs), and control (using measures of density and land use to reflect fixed effect variables). The relationships were significant, but only 30% of the variance was explained by these variables.

The Chesterton study (2002) used the hedonic pricing approach, and three variables used (neighbourhood, dwelling group and walk time) explained less than 50% of the variation in transaction prices for both terraced houses and flats and maisonettes. This suggests other variables not identified explain as much about price as the ones identified. The neighbourhood variables seem to have by far the highest influence on prices, and this was confirmed by longitudinal analysis. There is limited evidence of price decay in some areas, but this varies and distance decay effects were unclear.

7. Transactional analysis monitors the changes in property and land values from actual transactions. This type of analysis is very useful, but usually depends on information from the valuation office and the land registry, often at a very localised level. Problems may occur with confidentiality, but it can (and should) be used to determine the spatial patterns of change around stations and the changes in these values over time.

The Land Registry in England and Wales contains the transaction data for every residential property since January 1996. It records the transaction price and date, the unit postcode, and the type of property sold. The Valuation Office Agency also collects this information for both residential and commercial transactions. In addition to the variables above, there are data on floorspace, age of property and tenure. In Scotland similar data is available from the Report of Sasines and Scottish Property Network.

8. Projected rateable values also give a good indicator of property values in both the residential and commercial markets, and a revaluation is currently being undertaken (to be completed in 2005). Related to this is growth assessment, which requires experts to determine the way the market is likely to move in terms of yields, occupancy rates, the demand for different types of space, and the rents to be paid. It seems that much of this assessment is dependent on the property sector's own knowledge of the market and as such is dependent on the quality of data input, support systems and the skills of the key researchers. But it can act as a valuable additional means by which quantitative and qualitative approaches can be brought together.

CB Hillier Parker (2002) report contains an analysis of the levels of yields and rateable values likely to be available to contribute to the funding of Crossrail, via a range of value capture mechanisms including:

• Section 106 (planning gain) charges;
• Use of the rating and Council Tax systems to secure contributions from occupiers who benefit from an improved business environment and transportation links;
• The introduction of a levy on freeholders to reflect the benefit from what could be considered an "unearned" capital gain in the value of existing property stock.

The report anticipated that some 10.87 million square metres of additional commercial floor space would be realised by 2025 within the study area, a little over half of which would be generated by redevelopment. Within the residential sector the report anticipated 54,804 new dwellings to be constructed within the study area by 2025, all of which was assumed to be new development.

3.4 Conclusions

As we have seen, there are many available methods, but each situation needs to be carefully investigated prior to a particular method or set of methods being selected. This decision depends on the objectives of the analysis, the scale at which it is to take place, and whether time series analysis is required. There also seems to be no clear set of methods explicitly directed at land values and transport investment in particular modes of transport. Transport congestion is now much higher on the agenda for businesses, as it is affecting their efficiency (Banister, 2002), and so the linkages between land values and transport investment need to be better understood, both methodologically and empirically.

In all situations, there is a case for mixing quantitative and qualitative analyses, with the more technical analysis being supplemented by surveys and interviews to help with interpretation and to infer some causality in the relationships. One of the weaknesses in much of the analysis has been the over emphasis on ex ante studies (forecasts), which have helped in trying to establish whether a transport investment should be made. There is much less weight given to monitoring or ex post analysis which would help to improve methods and to learn from a comparison between expected and actual outcomes. This means that data should be collected at several points in time so that repeated cross sectional analysis can be carried out, or even some longitudinal analysis where individual firms or households are followed over time.

One new approach developed in the last year, has been the use of GIS mapping techniques to draw "value surfaces" for different time periods around the stations along a public transport route to reflect the actual transaction prices related to residential and commercial properties (ARW et al., 2003). This approach is discussed further in the conclusions to this review, but subject to data availability, it does seem to accommodate most of the empirical and methodological challenges presented in Sections 2 and 3.

The basic requirements of the method are that there should be high quality transactional price data at the local level that measures change over time (including before the opening of the transport investment and for a sufficiently long period afterwards to measure change), including other variables that might also be important. Secondly, the project should be of a sufficient scale to be reasonably confident that measurable change might occur. Thirdly, it seems that GIS Mapping approaches and hedonic pricing with a spatial dimension are the two key methods available - other approaches should be seen as supporting.

« Previous | Contents | Next »