The Scottish Government

« Previous | Contents | Next »

Listen

DEVELOPING A METHODOLOGY TO CAPTURE LAND VALUE UPLIFT AROUND TRANSPORT FACILITIES

APPENDIX 3: RETROSPECTIVE COMMENTS ON CROYDON TRAMLINK

A retrospective commentary is given on the lessons that have been learned from the recent study in Croydon (published 2004). The purpose of the Croydon Tramlink Study (CTL) was to develop and test the methodology to measure value uplift. Although there were also empirical results, these were of less importance. The methodology consisted of three elements - contextual analysis, quantitative analysis and interpretation.

Contextual Analysis

The contextual analysis attempts to focus on market conditions over the recent past (up to twenty years) within the study area, principally through identifying the changes in the office, retail and industrial sectors, with a smaller review of the residential property market. This review was supplemented by an investigation of the available secondary data, such as the recent Census 2001, employment data, local planning applications and changes in travel patterns in the area. There were also some interviews carried out with stakeholders to assess views on the property and office market - here was some optimism in that the CTL had arrested decline in the office sector and that it had increased the catchment area of the retail outlets. Some suggested that there was a boost to industrial users through widening the recruitment area, with better access. But there were no significant increases in planning applications or changes in use in the corridor, but this must be set against a buoyant residential property market and the area is already well served by a high quality transport network.

This contextual analysis was extremely helpful in identifying (through secondary sources and interviews) the changes that were taking place in the study area, and the expectations of the impacts that the CTL was having. The fact that market conditions were buoyant and the quality of transport already available was high, may make it difficult to identify any CTL effects.

The Quantitative Analysis

This data has been assembled from the Land Registry to record all transactions in the second quarter of each of four years (1996, 1998, 2000, 2002) ¹¹. These dates are intended to capture the changes in residential property values before the decision was taken to build the CTL (1996), the anticipated change before the line was opened (1998), the actual impact when the line was opened (May 2000), and a short time after the line had been in operation (2002). The focus of the change is in the corridor up to 1 kilometre away from the CTL, but the wider area up to 10 kilometre from the route has been selected as a control area to establish the broader changes taking place in the housing market as a whole.

This "buffer" zone was found to be too large as it reflects the whole of the housing market in South West London. More recent runs of the models have reduced this area to a corridor of 2 kilometres so that the change near the CTL can be compared with that further away. The intention is to develop a series of "maps" that reflect areas of similar property values and to establish whether proximity to the CTL has influenced that value over time.

Figure 1: All residential transactions in the second quarter of 2000

As property values are extremely variable, even over very short distances, some smoothing has taken place. The total sample has been divided into four property types to reflect differences in property values - detached, semi-detached, terraced, and flats and maisonettes (Figure 1). The second process has been to use an inverse distance weighting algorithm ¹² to reduce the impact of very high or very low valued properties (outliers). Once the basic values are obtained, these have then been mapped according to whether the transaction values for the property type is higher than the average for the whole study area or lower.

As noted above, the corridor effect around the CTL has been swamped by the huge increase in the South West London property market as a whole. The initial maps produced have reflected the substantial increase in property values in Central London (1996-2000), where prices have risen by over 50 percent, and the subsequent cascading effect from the central part of London towards the suburbs (1998 - 2002) as the "price bubble" surges out from the Centre. Current runs of the models are controlling for this through taking a much smaller buffer areas around the CTL.

The location of our sample transactions was important. If the spatial distribution of transactions was different between our two time slices - before and after the CTL was constructed, then this could introduce bias to the results. It was important then to only include in our analysis those unit postcodes that had a transaction in both time-slices. Of course, the impact of this constraint would be to reduce the number of transactions in our sample, which inevitably would reduce the confidence we would have in the results. In retrospect, it may not have been necessary to be so rigorous in this assumption, and it is a case where the loss of data may have been more important than analytical elegance. It is recommended that in future all data transactions are used.

The table above gives us an indication of the impact of this constraint. We can see that the amount of transactions available to us has shrunk considerably. This reduction in the sample size varied between the property types; detached housing in particular saw a major drop. An obvious solution to this conundrum would be to combine the data across the four broad transactions categories - detached, semi-detached, terraced and flats & maisonettes. Statistically, this was not possible because when the sample distributions of transactions in the various markets where analysed, it became clear that they did not come from the same population and therefore could not be reliably combined.

But from the preliminary analysis, it has proved very difficult to isolate any effects from the data. In the detached sector (small sample), the London effect seems dominant with few CTL effects, except at the end of the line in Wimbledon, which is a major interchange point for the suburban rail and tube systems. For semi detached houses, the London effect is still there with the demand for housing close to Central London having a premium value. It was in the terraced housing type that some CTL effect was discernable, but further examination is required of the impacts. The flats and maisonettes seemed to show little positive uplift in the corridor or elsewhere. This may be because the data in this category is even more heterogeneous than that elsewhere in the dataset.

Property Market Effects

Change across South London - The house price data surface for South London is shown for the four types of property and it can be seen that there was a significant uplift in values across 1996 and 2002 (Figure 2). Areas which are shaded white are those which have seen an average increase over the period; those shaded in red have seen an above average increase in value, the darker areas being where prices are much higher than average. Those areas in blue are those where there has been a below average increase in values. The degree of increase in value was different across the study area, with central London at appearing to be a major driving force in the market. It is difficult from the maps shown above to determine how the increase varies through space, and to identify whether or not there has been a demonstrable increase in values within one kilometre of the CTL.

Detached Market - There doesn't appear to be any obvious pattern to the increases in the detached market. This is primarily because the sample size for this market was too small, and distributed unevenly across the study area. While there does appear to have been an above average increase in values close to Wimbledon, it would be difficult to attribute this to the CTL.

Semi-detached market - Proximity to central London and central Croydon appears to be a bigger driver in the market than the CTL.

Terraced market - Of the four sub-markets, the terraced market had the most sample points and is probably the most reliable. Areas to the north of the CTL have seen the largest increase and although there has been some uplift along the Beckenham branch and around Wimbledon

Flats and Maisonettes - Proximity to central London again appears to be the major driver of market increase and although there is above average increase in values close to Wimbledon with the 1 km zone. How much of this is due to CTL, and how much to do with the Wimbledon postcode is a moot point. In parallel value impact is markedly sub-average in Central Croydon and towards the New Addington end of the line.

Figure 2: The impact of CTL across South London

The local impact of CTL - To control for the central London effect, the relative increases in property values in closer proximity to the CTL. Figure 3 again show those locations, which have seen an above or below average increase in values between 1996 and 2002, but this time the average is calculated over a much smaller area - those locations that are within 2 km of the CTL. The aim of this is to remove some of the macro effects (such as the proximity from Central London) and to isolate CTL's more immediate impacts. A 2km zone was chosen since while is it much smaller that the south London study area, it should also be possible to detect any uplift as we enter the 1 km zone around the stations.

Figure 3: The impact of CTL within 2km of the stations

Again, there appears to be no consistent pattern emerging, which suggests that CTL has had a major impact on property values at this scale. Indeed, the overall average increase in value along the New Addington spur - one of the areas which sees the most dramatic improvement in accessibility - was consistently below the local average in all four sub-markets. In contrast, properties on the northern parts of the Wimbledon and Beckenham spurs have increased more than the average for the area.

Qualitative Interpretation

A range of methods was used to obtain an understanding and interpretation of the results from the quantitative analysis. Questionnaires were sent to 125 local property developers to gauge their views on the CTL impacts, but this exercise proved unsuccessful as the response rate was very low and because the respondents were unwilling to make clear judgements. Generally, the respondents identified a positive impact from the CTL, and that accessibility improvements were key to positive uplift. It also gave civic pride and "credibility" to Croydon, but the actual impacts were difficult to identify in a buoyant local market.

Of more interest was the Colin Buchanan and partners (2002) report for the South London Partnership on value uplift. Their analysis was also based on property transaction data at the postcode level (too high for individual effects) as compared with the Greater London changes as a whole. They identified a 4 per cent uplift in Croydon, but not elsewhere in the corridor - those near the CTL were better off than those away from the line in Central Croydon, before, during and after opening. There were no net effects found in Merton and Bromley, even though there was some uplift before but none after opening - a general repositioning effect. It should be noted that this analysis was at the post code sector level (8500 households) and that Greater London was taken as the control area. This seems to be too crude an analysis to identify CTL effects, and many assumptions are being made about what was going on more generally.

Finally, perceptions of the key stakeholders were again sought on the CTL impacts. It was generally felt that the CTL was a greater success and that accessibility had been improved. There was a greater interest in central Croydon for development, and that the CTL had prevented the drift away. The rental market seems to have been less affected than the private housing market.

These three approaches to interpretation (questionnaires, evidence from other studies and focus groups) have all helped to interpret the findings from CTL, but they have provided less clear messages than expected. It seems that the respondents are unable or unwilling to be positive (or negative) about the impacts. This is probably due to the variable impacts, the time required for change, the effects of other changes in the local economy, and their own information and knowledge. This would suggest that the inputs to the interpretation need to be very clearly specified in terms of the actual changes found in the quantitative analysis, so that the stakeholders can relate clearly to those outputs. Otherwise the responses seem to be too general to be of real use.

Comments on the Croydon Study

At the end of this extensive attempt to investigate and understand the effects of a major transport infrastructure investment on property values in one corridor in South London, what can be concluded?

1. The data quality is crucial, as it needs to be locally based, geocoded and to report actual transaction prices. The Land Registry data only records private housing transactions in the four categories identified. It does not cover the public housing sector or the rented (mainly private) sector. There are no details on the properties (e.g. number of rooms, central heating, garage etc) or on the quality of the property. The housing market is extremely heterogeneous and it has been necessary to normalise the data through the use of a simple density variable to reflect local quality. This is where other measures of quality should be included, but the data set used does not have this information.
2. Further analysis will be carried out on the residential and commercial sectors where the Valuation Office has detailed information on transactions with more variables that reflect the size and the quality of the property.
3. The effects of proximity to the Central London property market has been the major driver of residential property values over time, together with relatively low interest rates for mortgage loans. This is apparent from the red areas in Figure 2, and over time it is possible to see this effect spread out from the centre in a cascade of value increase. . Matching changes in values within the 1 kilometre and the 2 kilometre distances from the tram stops seems to be one way to isolate tram-related impacts from London wide effects, but as the study area is redefined, there is less data to be used.
4. Change takes place over time and this means that information is required for at least three (or four) points in time to reflect the before, during and after situations, together with control data for the property market as a whole.
5. Supporting evidence is also required for other changes in the local economy over time, as other changes may also influence property values. This was carried out here through a survey of local property agents, but the results were inconclusive, as they seemed reluctant to place a value on the CTL effects, or on other changes that had taken place in the local economy. Their conclusion was that accessibility was the main determinant of the local property market values, with the CTL being mentioned along with access to local schools, access to Central London, access to sources of employment and access to the local rail network. The other main factor commented on was that there had been a substantial increase in the numbers of transactions taking place (up by 50%), but again it was difficult for the property agents to isolate the CTL effects.
6. The latest time period used for the analysis was for 2002, and this may be too soon for the impacts on prices to be realised (2 years). This is reflected in the relatively small number of transactions in a residential location over a year, and so a five year span may be more suitable.
7. Development does not take place continuously along the transport corridor, as it is concentrated at specific points. This makes an investment such as the CTL harder to identify change. It may be better to focus on particular locations in the corridor (e.g. interchanges or the main points of tram use) where impacts can be expected. At several points along the route there is no development, as the land has been designated open space. The impacts of a heavy rail system may be easier to study, as each station is clearly located in a spatially discrete environment.

In summary, the methodology developed is robust, and the use of data surfaces allows the identification of local change and to see how this change fits into the overall picture. It has clear advantages over the use of predetermined zoning systems and each data point is fully used in the analysis. The fact that little value uplift has been identified that can be attributed to the CTL, is disappointing, but this reflects the far greater importance of other factors and the need for a longer period over which to measure change. It would also have been helpful to have had more information on the properties that were being sold, so that the data could be normalised or analysed through other statistical methods (e.g. geographically weighted regression or hedonic pricing models).

« Previous | Contents | Next »