Relativity v 95% Confidence Band [2014] UKUT 319 (LC)

IN THE MATTER OF AN APPEAL FROM A DECISION OF THE

UT Neutral citation number: [2014] UKUT 0319 (LC)

Venue LEASEHOLD VALUATION TRIBUNAL FOR THE, LONDON RENT ASSESSMENT PANEL

Between

Relativity

95% Confidence Band

Model 1

86.79%

83.96% - 89.62%

Model 2

85.95%

82.38% - 89.51%

Model 3

86.43%

76.89% - 96.81%

Stan Gallagher , instructed by Farrer & Co, on behalf of the Appellant

Gary Cowen , instructed by Forsters LLP, on behalf of the Respondents

[1]This is an appeal, by way of a rehearing, from a decision of the Leasehold Valuation Tribunal for the London Rent Assessment Panel (the LVT) dated 10 June 2013 whereby the LVT decided certain matters concerning the price to be paid by the appellant to the respondents upon her purchase of the freehold of 47 Phillimore Gardens (the property) pursuant to the terms of the Leasehold Reform Act 1967 as amended.

[2]The appellant holds the property from the respondents upon a long lease which at the valuation date (13 October 2011) had 52.45 years unexpired. The price to be paid fell to be determined in accordance with section 9(1A) of the Act which provides, so far as presently relevant, that the price shall be:

“…..

(b) to (f) ……………”

In the present case in accordance with section 9(1D) a marriage value is to be included in the calculation of the price. In order to calculate that marriage value it is agreed between the parties that it is necessary to know the value at the valuation date of the existing 52.45 year lease assessed upon the assumptions in section 9(1A) – we hereafter refer to this value as the existing lease value.

[3]It is agreed between the parties that in order to calculate the existing lease value the proper approach is to start with the freehold vacant possession value of the property as at the valuation date and then to decide what proportion, expressed as a percentage, the existing lease value bears to the freehold vacant possession value. This proportion is referred to as the relativity. In the present case before the LVT the appellant contended for a relativity of 87.04% and the respondents contended for a relativity of 75.5%. In the result, the LVT adopted a relativity of 76%. It is against this finding by the LVT that the present appeal is brought by the appellant. As regards the freehold vacant possession value of the property at the valuation date the LVT determined this as £16,138,743 and there is no appeal against this determination.

[4]Before the LVT the appellant relied upon evidence from Dr Philippe Bracke (whose qualifications we refer to in more detail later but who in summary has a doctorate in economics and who at that time was a research economist and was also a teaching fellow at the LSE) who had devised a hedonic regression model to predict relativity for different lease lengths. The model was constructed upon a large amount of data regarding market transactions prior to the introduction of the Leasehold Reform Housing and Urban Development Act 1993 (so as to avoid the impact of Act rights upon prices). This data had been obtained from John D Wood (who have a large database of concluded transactions in which they were involved) and also from information available from Lonres which is a subscription-based website that records residential transactions in London. As regards relativity the appellant before the LVT relied entirely upon Dr Bracke’s evidence. The respondents relied upon valuation evidence from Mr O.J. French MRICS of Savills, who relied upon certain relativity graphs, in particular those prepared by Gerald Eve and John D Wood. The LVT concluded that Dr Bracke’s work was a serious minded independent analysis of market data. However as Dr Bracke’s work had not been peer-reviewed and as there was no valuation evidence on behalf of the appellant to support his work the LVT concluded that Dr Bracke’s work could not in isolation be relied upon, that Mr French’s evidence was to be preferred, and that the relativity was 76%.

[5]Before the Upper Tribunal the appellant has once again relied upon the work of Dr Bracke. The appellant submits that Dr Bracke’s work has now been considered, in the nature of a peer review, by Prof Sean Holly of Cambridge University who was called to give expert evidence in support of Dr Bracke. The appellant also called Mr Ingram-Hill MRICS of John D Wood, but it was made clear that Mr Ingram-Hill was not called as an expert witness and indeed he did not seek to give us any expert valuation evidence. Accordingly the expert evidence on behalf of the appellant consisted solely of the evidence of Dr Bracke (as supported by Prof Holly), with no valuation evidence submitted. Before the Upper Tribunal the respondents relied upon the evidence of Prof Colin Lizieri (whose qualifications we refer to in more detail later but who in summary is the Grosvenor Professor of Real Estate Finance in the Department of Land Economy at the University of Cambridge) who gave evidence analysing and commenting upon Dr Bracke’s work. Prof Lizieri did not seek to give any expert valuation evidence, but such valuation evidence was given on behalf of the respondents by Mr French. The following remarkable feature of the evidence as laid before the Upper Tribunal can be observed at this stage, namely that while at the centre of the present appeal lies the question of the existing lease value:(a) (A) the appellant has called no valuation evidence at all from anyone with formal qualifications in valuation of real property but instead relies upon Dr Bracke’s study (supported by Prof Holly) and his graphs showing how relativity varies with lease length based upon his statistical study of transactions between 1987 and 1991; and

(b) (B) as well as evidence from Prof Lizieri (analysing and commenting on Dr Bracke’s work) the respondents have called evidence from an expert valuer, namely Mr French, who has told us that there is no valuation assistance to be obtained in the present case from any study of comparable transactions in the market (with allowances to be made to convert to the no Act world) and who in consequence relies entirely upon various graphs which have been published by the RICS, being graphs which he did not prepare and regarding the preparation of which he has limited hearsay information to give.

[6]We shall return later in this decision to this lack of valuation evidence.

[7]In summary the appellant advances the following submissions:(1) Dr Bracke is a competent and reliable economist and statistician who has carried out a careful and detailed analysis of what actually happened in the market and has come up with a graph showing how relativity varied with unexpired lease length at a date prior to the extension of leasehold enfranchisement rights which occurred in 1993 (such that the relativities shown are relevant when assessing the relativity of the existing lease value on the basis of no Act rights).

(2) Dr Bracke’s work has now been peer-reviewed (thereby removing the concern of the LVT).

(3) This work is a valuable response to the invitation given in various cases by the Lands Tribunal to the RICS (see in particular Arrowdell Ltd v Coniston Court (North) Hove Ltd (LRA/72/2005)) in the hope that the RICS may be able to produce guidance in the form of standard graphs of relativity that can readily be applied by valuers in carrying out enfranchisement valuations.

(4) The RICS did convene a Leasehold Relativities Group in an endeavour to comply with this request but were unable to agree upon definitive graphs.

(5) Dr Bracke has revealed much detail regarding his database, methodology and computer programs to the respondents’ experts for the purposes of this case and has dealt satisfactorily with their criticisms.

(6) The respondents have produced no comparable based valuation evidence, but instead merely rely upon graphs prepared by others and, in contrast to Dr Bracke’s approach, have provided no, or no significant, evidence regarding how those graphs were prepared and regarding their reliability.

(7) In consequence this Tribunal should adopt Dr Bracke’s approach, or alternatively should give substantial weight to his approach when deciding upon relativity.

[8]In summary the respondents advance the following submissions:(1) Dr Bracke has a financial interest in the outcome of the present case (because of his intellectual property rights in his study of relativity which he wishes to exploit commercially) and that in consequence little or no weight can be placed upon his evidence because he is not an impartial witness.

(2) Whilst an analysis of historic data based upon hedonic regression methods may in principle be an acceptable method of studying economic data regarding house prices, various points of concern can be made in respect of the accuracy and confidence levels of Dr Bracke’s study.

(3) Quite apart from such concerns, Dr Bracke’s study only measures relativity in respect of market transactions between 1987 and 1991 and this is of no relevance (or only of limited relevance) to the assessment of the proper relativity for calculating the existing lease value because (a) the data is from more than 20 years before the valuation date and the market has much changed in the meantime, (b) from substantially prior to the passing of the 1993 Act there already existed an anticipation of the likelihood of an extension of enfranchisement rights and this will have affected the price payable for leaseholds during at least part of the period studied by Dr Bracke, and (c) in any event, even if Dr Bracke’s study accurately measures relativity in a world where there are no enfranchisement rights, the assumption required to be made under section 9(1A) is in fact different and requires the assumption that there are no enfranchisement rights for the existing lease but that it is hypothetically for sale at the valuation date in a market where other long leaseholds do carry enfranchisement rights.

(4) In consequence no weight can be placed upon Dr Bracke’s evidence.

(5) The Tribunal should accept Mr French’s approach and calculate the relevant relativity by reference to the graphs published by the RICS. Alleged partiality of Dr Bracke

[9]It is convenient at this stage to deal with the point raised in paragraph 8 (1) above regarding the alleged lack of impartiality on behalf of Dr Bracke. The matter arises in this way. Prior to a case management conference on 28 March 2014 before the Deputy President an application was made to the Tribunal on behalf of Dr Bracke that there should be a confidentiality club restricting the disclosure of Dr Bracke’s report and, in particular, the data and methodology behind the report. The nature of the restriction sought was that any persons to whom such material was disclosed for the purpose of these proceedings should be prevented from disclosing that material to any further recipients unless that further disclosure was also for the purpose of the proceedings. Dr Bracke submitted a witness statement in support of the application which included describing the large amount of work which had gone into his preparation of the dataset (which was based upon confidential information from John D Wood as well as upon information from Lonres) and into the cleaning of this dataset (so as to make it in an appropriate form for being the subject of computer analysis) and into the methodology which he has adopted and into the preparation of his final results. He pointed out that the database and methodology and results were confidential and were the product of many years of hard work based upon a professional lifetime of experience. It was made clear that this confidential information was a valuable asset belonging to Dr Bracke and that it was desired to preserve that valuable asset so that it could be exploited in the future for commercial gain.

[10]The disclosure of the foregoing information led the respondents to argue at the case management conference before the Deputy President that the appellant should be debarred from calling Dr Bracke as a witness at all on the grounds that he had a commercial interest in the outcome of the proceedings and could therefore not be relied upon to express an independent expert opinion for the purpose of helping the Tribunal in accordance with the declaration required of all expert witnesses as to their duty to the Tribunal. The Deputy President declined to debar the appellant from calling Dr Bracke. The respondents decided to seek to appeal this decision and sought permission from the Court of Appeal to do so. This permission was refused, the learned Lord Justice observing that he particularly endorsed the Deputy President’s point that if the matters relied upon do in truth undermine the value of Dr Bracke’s evidence by impugning his impartiality, then that is most appropriately established by cross examination. As a result a significant amount of cross examination of Dr Bracke before us was directed to exploring his commercial interest and to challenging his impartiality.

[11]We remind ourselves that Dr Bracke clearly does have a commercial interest in the outcome of these proceedings, not merely as any expert normally might have indirectly (such as from a possible increase in reputation from having their views adopted) but by reason of his intention commercially to exploit the work he has done to produce his database and methodology and results regarding relativity calculations, the value of which will be substantially adversely affected if this tribunal declines to accept his evidence. We were told that Dr Bracke was not being paid by the appellant for his appearance at and work in connection with the present appeal and that Prof Holly would be paid not by the appellant but in effect by Parthenia Research Limited, being a company in which Dr Bracke is interested and to which he has transferred the relevant intellectual property rights. We take these matters into account and we remind ourselves that the present situation is, in our view, most unusual. However we heard Dr Bracke being cross examined not merely upon his commercial interest and alleged partiality but also upon the case as a whole. We found Dr Bracke an impressive witness who presented his work in a fair manner, who was prepared to accept criticisms or other points potentially adverse to his work in so far as he thought there might be merit in them and who carefully explained the extent (if any) of the merits in these criticisms rather than seeking merely to brush them aside. He was in the witness box for a substantial time. We are satisfied that the evidence he gave represented his genuine and honest professional views upon the points in question. We reject any argument that we should ignore his evidence or downgrade the weight that can be attached to it by reason of some alleged partiality or lack of genuine independence.

[12]We should in this context add that at the commencement of the hearing before us we were addressed by Mr St Quintin on behalf of Parthenia with a request that we should extend the orders regarding confidentiality which had been made by the Deputy President at the case management conference, but which had been made on the basis that they would only extend until the hearing. We will not prolong this decision by repeating the reasons for the ruling which we gave upon this application, but in the result we did make an Order as set forth in Schedule 1 to this decision. However we made it clear that, whatever the position may be pending the publication of this decision, there was a high public interest, outweighing any commercial interest of Dr Bracke or Parthenia, that if Dr Bracke chose to give evidence as an expert to the Upper Tribunal then we must be entirely free to express our decision in whatever way we thought right, having regard among other matters to the role of the Upper Tribunal in promoting consistent practice in land valuation matters, see per Carnwath LJ in the Court of Appeal decision in Earl Cadogan v Sportelli

[2008] 1 WLR 2142

EWCA Civ 1042 at paragraph 99. It was therefore made clear that our decision might well refer to (and perhaps append a copy of) some part or parts of Dr Bracke’s work including his graphs. Dr Bracke was therefore giving evidence in circumstances where it had been made clear to him that the more merit we found in his work the more likely it was that certain details of his work would be expressly referred to in our decision and would in consequence lose the confidentiality he wished such details to have. Evidence for the appellant Dr Phillipe Bracke

[13]Dr Bracke is an economist. At the time of the hearing he was a Research Economist and Teaching Fellow at the London School of Economics (LSE) but was shortly to take up a position at the Bank of England. He had been working on a research project on the issue of relativity with a colleague from the LSE since 2011. Between October and December 2012, he was given access to the John D Wood & Co internal data set and the Lonres.com website, which jointly formed his data sources.

[14]In his expert report, Dr Bracke confirmed he had been instructed on behalf of the appellant for the purposes of this appeal to investigate the issue of relativity according to pre-1993 data or, as he put it, in a “no Act world”. In particular he was asked to provide expert evidence on the relative value of the appeal property, with 52.45 years unexpired on its lease at the valuation date, relative to its freehold value, by using real world transactions.

[15]At the end of this process, Dr Bracke arrived at three models. Model 1 was a slightly amended version of that which was presented to the LVT in May 2013, and contained 8,184 data points. Model 2 was based upon a newly extended set of variables, although a slightly smaller range of 7,476 data points. Model 3 was based upon the same set of variables as Model 2 but using only data points where the sale price was confirmed – of which there were 2,622.

[16]Dr Bracke outlined the steps taken to produce his relativity conclusions. In essence these involved taking data from the John D Wood & Co archives and the Lonres.com archives, processing this data to produce an adjusted data set which was then fed into a statistical model to arrive at a series of relativity figures and subsequent graph, from which the relativity for a lease with 52.45 years unexpired could be obtained.

[17]Dr Bracke stressed that he had no experience in valuing individual properties but specialised in extracting patterns from large data sets and has experience of working on the macro and micro economics of the housing market for some years. He said that the analysis presented in his report was “state of the art”, based upon well-established and uncontroversial statistical techniques, though ground-breaking in terms of the data set used. The focus of his report was property prices in Prime Central London (PCL) and the effect on value of unexpired lease length.

[18]He gave background to the traditional methods of accessing relativity, which compared a hypothetical or counter-factual scenario with the true state of the world. Existing evidence-based estimates traditionally relied upon one of three methods: open market transactions of comparable properties, where necessary adjusting to reflect benefits of rights under the Acts; settlements of claims under the Acts; and Tribunal decisions. Each method had drawbacks. The vast majority of documented transactions were post-1993 and the actual effect of the Act on prices was unclear. Settlement evidence was subjective, may be influenced by the "Delaforce effect" and may be self-perpetuating.

[19]Dr Bracke referred to the Tribunal’s hope expressed in Arrowdell that the RICS could produce standard relativity graphs but he noted that the working group set up by the Institution was not able to form an agreed position in its subsequent report (“the RICS report”).

[20]In terms of the use of historic transactions, and the difference from the use of other relativity curves, Dr Bracke commented that pre-1993 data had the clear advantage of avoiding any effect of the Act on prices. He considered it impossible to use post-1993 sales to estimate relativity because those sales were affected by the Act (buyers being aware that they had the right to extend or enfranchise). Current prices were affected by the existing relativity curves. Accordingly post-Act data was tainted and any attempt to adjust it would be subjective. It was impossible to estimate the benefit of the Act in current times owing to a lack of counter-factual sales.

[21]Dr Bracke noted that the Gerald Eve graph featured in the RICS report, which was originally based on pre-1996 data, had not since been amended. Additionally the report said that it was considered there had been no reason for relativity to have changed over time.

[22]He indicated that his analysis, using 8,184 data points, produced a relativity curve that was generally higher than most of those mentioned in the RICS report. But none of the contributing companies to the report had published the data underlying their analysis, which on average each used around 200 points.

[23]Dr Bracke had taken his raw data from two sources, each within a date range of 1987 to 1992. From the John D Wood & Co records, he used transactions of properties in Belgravia, Chelsea and Kensington. From Lonres, he had used data for postcodes SW1, SW3, SW5, SW7, SW10, W1, W2, W8, W11 and W14, which he had been advised made up the PCL area. The Lonres records included PDF copies of original sales particulars from which Dr Bracke was able to pick out features such as whether the property had a garden.

[24]This combined data was then processed to render it suitable for proper analysis. Various transactions were deleted from the analysis, for example suspected duplicate sales, data points where information was missing or John D Wood sales in the Lonres data that would have caused duplication. Various modifications were made to unify the two data sets, which were then merged.

[25]The combined data set was then subject to further processing. Properties that were susceptible to enfranchisement were deleted (based on their rateable value), as were properties for which the enfranchisement status was unknown. Sales in 1992 were also deleted to avoid any effect on price of the possibility that the leasehold reform would be bought onto the statute book by that year’s incoming Government. As a robustness check Dr Bracke also ran his analysis including 1992 sales but this did not change the result in a material way. Sole properties on streets were also deleted and some variables were created and formatted, for instance floor level.

[26]Following this merger and processing, the final data set had 8,184 sales. From the John D Wood data, there were 1,006 houses and 1,040 flats. From the Lonres data there were 1,643 houses and 4,495 flats. Dr Bracke commented that whilst all of the transactions from the John D Wood data were verified contract exchange prices, 86.7% of the Lonres data had not been similarly verified. Dr Bracke did not consider this to be problematic for two reasons. First, non-verified properties were found in very similar proportion between freehold and leasehold properties, which therefore did not affect a relativity analysis between the two. Secondly, he had run his statistical model only on verified properties as a check method and obtained a similar relativity estimate. The data was spread over five years with the largest number of transactions being in the third quarter of 1988. In respect of leasehold unexpired terms, there were many data points for leases with 55-65 years left, and for 85-100 years left.

[27]The attributes used in the statistical analysis were then grouped and listed in three ways. The most important, or primary, attributes were whether or not the property was a house or a flat, the number of bedrooms (this being a proxy for the size of the unit but then being subject to a further robustness check using square footage when available); the quarter of the relevant year in which the sale occurred and the street upon which the property was located.

[28]The secondary or additional attributes, based on data from the John D Wood and Lonres tables, included floor level, whether or not the property was a maisonette or a purpose-built flat, whether the sale had been verified, whether there was an onerous ground rent (which Dr Bracke considered would be when the rent was above 0.1% of the sale price), or when the property was a freehold flat. These attributes were used in Models 2 and 3.

[29]The third group of attributes were those which Dr Bracke termed extracted attributes. These were based upon the "notes" column from the John D Wood website, or the PDF sales particulars from Lonres. The extracted attributes included whether the property was a detached house or a mews house, whether it had two or more bathrooms, whether it had a garden or the use of a communal garden, a balcony, terrace or patio, or whether it had been refurbished or was in need of refurbishment. Of the 8,184 transactions, six did not have floor area information, and only 7,476 transactions had sufficient information to use.

[30]This revised data was then subject to a form of statistical modelling known as hedonic regression. The methodology was complicated but in essence was based upon the premise that the price of each individual property would be based upon the average property price in that sample plus or minus an error factor which signified by how much that price diverged from the average price. This was carried out for all of the factors of the variables outlined above both for freehold and leasehold properties to ensure that there had been a mixed adjustment prior to relativity being calculated. The data was then subject to a logarithmic model. Logarithms were used because they could be effectively added together for analysis purposes even though attributes between the various factors were multiplicative. The resultant figures were then converted back from a logarithm using an exponential calculation to enable them to be interpreted as relativities.

[31]Hedonic regression was used by a number of organisations including The Nationwide, Halifax and a recent RICS study in respect of green building certifications. Dr Bracke stressed that hedonic regression was not a model to show how individual house prices were determined but was to compare average leasehold prices at various lease lengths with average freehold prices, having adjusted for the various factors outlined above.

[32]The results were, as to be expected, scattered around on a graphical basis and accordingly a polynomial curve was used on a best fit basis – similarly to that traditionally used on relativity graphs. The relativity of the interest was then computed as the value of the polynomial curve at a specific lease length. To allow for the fact that at certain unexpired lease lengths, for instance around 60 years, there were more data points, the regression that fitted the polynomial was weighted using the number of data points corresponding to that specific lease term. Dr Bracke confirmed that this had not been done in the case of the model submitted to the LVT but had been done for Models 2 and 3.

[33]To estimate the confidence level of his relativity estimates, Dr Bracke used a method known as "boot strapping". This involved re-shuffling the sample and drawing different data points so that a slightly different sample was obtained, estimating the relativity on that new sample and then repeating this multiple times (he had done this over 10,000 times). All of the relativities obtained were then ranked from highest to lowest and the 2.5th and 97.5th percentile were used to construct a 95% confidence band around the original estimate. In this way a narrow confidence band, where the 2.5th and 97.5th percentiles were close to the original estimate, was taken as a sign that the statistical information procedure was robust and that results did not change drastically from one sample to another.

[34]Dr Bracke explained that Model 3 relied on fewer data points than Models 1 and 2 and therefore leasehold terms had been split into three year "bins" - for example a leasehold term of between 1 and 3 years, between 4 and 6 years etc. This resulted in a much wider confidence band for Model 3.

[35]The results were as follows: Relativity 95% Confidence Band Model 1 86.79% 83.96% - 89.62% Model 2 85.95% 82.38% - 89.51% Model 3 86.43% 76.89% - 96.81%

[36]In conclusion from the three relativity figures Dr Bracke arrived at an average of 86.39% which he therefore confirmed as his opinion of the best estimate of the market relativity of a 52.45 year unexpired leasehold interest in the no-Act world. He confirmed that his research was still ongoing but that it was a significantly more detailed and rigorous technique than anything represented so far in the field of leasehold relativity.

[37]In answer to a question from the Tribunal, Dr Bracke agreed that it was unexpected that his model 3 curve appeared to show a very similar level of relativity for a leasehold unexpired term of around 50 years as it did for one of around 80 years.

[38]In a supplementary report, Dr Bracke gave further support to his choice of a fourth-order polynomial curve. This choice involved a trade off between smoothness and accuracy. He accepted Professor Lizieri’s contention that other forms of curve would produce different estimates of relativity, but submitted that these differences would not be large. In support of this he produced a table showing relativities for each of the three models, using a third, fourth, and fifth degree polynomial, together with a “local” polynomial (which used a restricted “bandwidth” of data, 15 years on either side of the relevant point).

[39]His supplementary report also dealt with his interpretation (which was slightly misplaced as we mention below) of Professor Lizieri’s apparent contention that the true result may equally likely lie at any point in the confidence band. However, Dr Bracke submitted (which Professor Lizieri accepted while giving evidence) that the most likely true value always corresponds to the estimate at the centre of the confidence band.

[40]During the hearing, Dr Bracke provided a third report, in which he illustrated that the distribution of unverified sales was relatively uniform across varying lease lengths and freehold interests – at between 80 to 90% of sales for all groups. Professor Sean Holly

[41]Professor Holly is the Executive Director of Research in the Faculty of Economics at the University of Cambridge, a University Reader, and a Professorial Fellow of Fitzwilliam College. He has more than 30 years’ experience in economics and econometrics. He was instructed to conduct a second review of the data, methodology and results generated by the relativity research project that formed the basis of Dr Bracke's expert report.

[42]In his expert report, Professor Holly gave a brief background to the use of hedonic indices which he said dated back to at least 1928 and which were a well-established international standard for constructing price indices, most commonly house price indices. He explained that a hedonic price index weighs each transaction in terms of the attributes of that property, bedrooms, garden, location etc. In order to calculate by how much an average index of house prices had changed in the course of a year, an allowance had to be made for the different mixes of the types of property that were included. It may be that in one year a larger number of smaller properties are on the market and the following year, a larger number of more expensive properties. Reliance on this raw data may suggest that there has been a rise in the price index when all that had happened is that the mix of properties had changed.

[43]Professor Holly had re-run the programmes provided by Dr Bracke, using the 8,184 transactions, from which he was able to replicate the results and estimate the relativity arrived at by Dr Bracke. He considered the method to be entirely right for uncovering the effect of differing lease lengths on price. The methodology was a mathematical process that, by isolating and holding constant the effect of the other determinants of price, enabled the effect of lease length on price to be calculated as a mathematical certainty within the constraints or level of confidence of the model.

[44]Dr Bracke had carried out some further analysis in the light of previous concerns expressed by Professor Holly that house prices had been correlated spatially over and above the other controls that had been included in the analysis. Dr Bracke had found no evidence for a neglected spatial effect.

[45]In conclusion, Professor Holly considered that the empirical evidence contained in Dr Bracke's report was robust and after several alterations still showed a 52.45 year lease with an associated average relativity of approximately 86%. He stressed that the application of hedonic indices to understand the contribution that the unexpired term of a lease on the price of a property was uncontroversial. It had been used many times in many different countries to address similar questions but it was almost certainly new to the UK. Mr Alexander Ingram-Hill

[46]Mr Ingram-Hill is a chartered surveyor employed in the valuation and surveying department of John D Wood & Co in London. His department specialises in the valuation of residential property and acts on behalf of leaseholders in Leasehold Reform valuation and negotiations. He has been at the firm since 2007 apart from a small period in 2014 at Knight Frank. His evidence was given as a witness of fact.

[47]Mr Ingram-Hill confirmed that his company had provided assistance to Dr Bracke by examining historical rateable values of the properties that formed the data set. Approximately 25% of these were deleted from the data set as the rateable value indicated that they may have been enfranchiseable prior to 1993, or alternatively that records were missing or inconclusive. He said that all houses with five bedrooms or more could be considered conservatively to be non-enfranchiseable as it could safely be assumed that they exceeded the then applicable rateable value limits. This led him to identify all of those houses in a data set with four bedrooms or fewer, of which there were just over 1,000 (Dr Bracke said 1,006). The 1990 rateable values for those properties had been checked at the relevant local authorities.

[48]He accepted that this approach would have resulted in the inadvertent exclusion of a number of houses which were not enfranchiseable and therefore in an ideal world should be retained in the data set. Additionally it was possible that a small number of enfranchiseable houses may not have been excluded. But there would, at the very least, be a cancelling out effect.

[49]Adopting Dr Bracke's 86.39%, Mr Alexander Ingram-Hill arrived at a valuation of £1,925,521. To assist the Tribunal he also provided separate valuations on the basis of Models 1, 2 and 3 which arrived at the values of £1,893,244, £1,961,026 and £1,922,293 respectively. Evidence for the respondent Professor Colin Lizieri

[50]Professor Lizieri is the Grosvenor Professor of Real Estate Finance within the Department of Land Economy at the University of Cambridge. He is a Fellow of the RICS and of the Royal Geographical Society. He was instructed by the respondents to comment upon the report of Dr Bracke.

[51]Professor Lizieri agreed with the general descriptions of the technique of hedonic modelling provided by Dr Bracke and Professor Holly, and commented that hedonic models date back to the 1920’s but were developed on a more formal basis from the 1960’s. There was a long tradition of using hedonic models in housing markets. Whilst the Nationwide and Halifax UK house price indices draw upon hedonic models, it was not the sole or indeed dominant way that such indices are estimated. Increasingly, indices are constructed using repeat sales methodology which analyses pairs of sales of the same property. The Land Registry’s house price index also uses a repeat sales method, which was developed partly to deal with problems with hedonic models.

[52]Professor Lizieri explained that in order to run a hedonic model, a researcher must make a number of decisions which critically affect the outcome. These include the selection of the data set, which must be an accurate as possible representation; the functional form of the model to be estimated, for example, the log-linear form adopted by Dr Bracke (which he accepted was made frequently in hedonic models); the selection of variables that are to be included in the model, as excluding missing variables can have a significant impact on the results; and the estimation method has to be selected which again will affect the reported outcome.

[53]Professor Lizieri highlighted some of the errors that can occur when the form of model does not accurately represent the process by which prices are formed in the market which may mean that the price of the attributes are inaccurately represented. Statistics may also be incorrectly stated, for example the functional form may be incorrect, the different attributes may interact, or the relationship between characteristic and price may be non-linear. The market may be segmented geographically or by type of purchaser or particular variables included in the model may be strongly correlated. He gave various examples of how this can happen.

[54]These problems can be compounded by missing variables that may be unavailable or not recognised, and all of these issues need to be considered in weighing the significance of a particular hedonic model. Professor Lizieri did not accept Professor Holly’s contention that missing variables would only be a problem in the current case if “the omitted feature is not correlated with the fact that a property is leasehold or not or, if leasehold, the length of the unexpired term of the lease”. That was unknown and the issue remained that the model would be misspecified in the presence of omitted variables that were relevant to the pricing process.

[55]A further issue was potential sample selection bias which, if uncorrected, could result in bias co-efficient estimates or spurious relationships between dependent and independent variables.

[56]Spatial autocorrelation should also be considered. Neighbourhoods often have similar structural characteristics and if in close proximity share the same location and amenities. A related problem is spatial heterogeneity, where the impact of a variable varies from one location to another for example an additional bedroom may have more impact if the property is located in the centre of a city than at the edge.

[57]Professor Lizieri accepted that none of these issues invalidate hedonic regression as a technique but they did indicate the range of problems that could exist in applying hedonic models to particular markets, and the potential biases that could occur if such problems were untreated. Detection was difficult which left uncertainty as to the reliability and robustness of any individual modelling specification applied to a particular data set.

[58]Commenting on Dr Bracke’s data set, and his method of extracting information from sales particulars, Professor Lizieri had no reason to believe that this process had not been carried out conscientiously. Others would be better placed to comment upon whether the blended John D Wood and Lonres data would be representative of sales from 1987 to 1991, but this would not be the case if for example records of significant agents in the PCL market were not present on Lonres or if the John D Wood transactions were tilted towards a particular type of property or client.

[59]The relativity analysis was strongly dependent on the accuracy of the price estimate. In some instances in Dr Bracke’s data there was an indicator that the selling price had been verified but not in all cases which presented a problem because inaccuracies in price cast doubt on confidence intervals for the individual co-efficient estimates. A further problem could arise if the sale price was estimated from the asking price. Not only was this inaccurate but it varied over time dependent on the market cycle – in a strong market sales prices may be greater than asking prices for instance. In the data set used, 95% had identical asking price and sale prices but those records where both were recorded there was an average absolute difference of 16%.

[60]Model 3 only including verified sales prices had a substantially smaller sample of 2,622 data points which, as Dr Bracke had noted, had a major impact on the confidence intervals around point estimates. Model 3 had a confidence interval of 19.90% with the lower bound less than 1% above the value adopted by the LVT. The Model 3 result could be interpreted as saying “I am 95% confident that the true relativity falls in the range of 76.89% to 96.81%” and not “I am 95% confident that the correct relativity is 86.43%”. Professor Lizieri noted that the confidence bands came from smoothed fitted relativity curves and Dr Bracke’s boot strapping procedure, and not from the estimates from the original hedonic model. In answer to a question from the Tribunal, he confirmed that he was not suggesting that confidence fell equally across the confidence band, and that it was more likely than not that the correct relativity would lie towards the middle of the range.

[61]The final data set was distributed unevenly across time with a strong peak in the third quarter of 1988 and periods having low sales transactions. This may have implications on the results. Similarly the distribution of unexpired term was highly clustered with a strong peak of around 60 years unexpired. This has implications for the curve fitting exercise. Less than 8% of the observations had unexpired terms of 25 years or less, but the estimated relativities for those data points were vital in determining the curve fitted to the data.

[62]In terms of the robustness and reliability of the results, Professor Lizieri noted the results of the three models. He had no reason to doubt that the models had been estimated competently and he had been able to replicate a number of the results. The co-efficiency reported seemed to be correctly found. The results suggested that, other things being equal, larger properties had higher prices than smaller ones. However some results were more difficult to explain, for instance having a patio reduced the value of a property relative to not having a patio. There may be some common factors between the variables and some effects may interact. Professor Lizieri constructed a time series of residential prices from Dr Bracke’s results which showed a pattern of rises to the end of 1989 and then a fall thereafter. He represented this graphically compared with other available series for Greater London and PCL. He noted that whilst the general shape was consistent with both the Halifax and Knight Frank PLC Index, the two latter appeared to show sharper falls than Dr Bracke’s and it was not possible to suggest any reason for the disparity.

[63]Examining the individual year coefficient and their error bands from model 1 – that had the largest sample size – it was evident that the relativity estimates were volatile and subject to large confidence intervals. Some results were implausible for example a point estimate for the relativity of the property with 75 years unexpired was 105% of the value of owning a freehold. In other places sharp falls occurred for example at 65 years the relativity is 96% but at 68 years it fell to 79%. The curve fitting process smoothed out these volatile estimates but their existence raised some doubt as to how much weight could be given to the relativity generated in this way.

[64]Professor Lizieri constructed a graph showing the individual point estimates of relativity from 42 years to 62 years – 10 years either side of the subject unexpired term. This graph illustrated the point estimates together with the 95% confidence intervals and showed a high degree of variation that was masked by the curve fitting exercise. Dr Bracke had used a polynomial curve, which was a “theory-free” approach. There was no explanation as to why Dr Bracke used a fourth order polynomial and various changes in slope could clearly be seen in Dr Bracke’s graphs. It was not clear that the polynomial was the correct form of curve to use conceptually even if it provided the best statistical fit for the estimates produced by the specific hedonic model for that data set. Professor Lizieri stressed that other forms of curve, which may equally show a good fit to the data, would produce different estimates of relativity.

[65]Professor Lizieri went on to examine the implications of Dr Bracke’s use of historic data. With a minimum of 20 years between the transactions and the subject valuation date, there had been substantial shifts in asset markets, the nature of investors and their attitude to risk and return.

[66]The ten year nominal zero coupon yield at the end of the first quarter of 1987 stood at 9.2%. At the end of the third quarter of 2011 it stood at 2.5%. Even allowing for reduced inflation expectations and distortions related to the financial crisis, that represented a substantial shift in a variable that was a key component of discount rates and hence the value of a reversionary freehold element. Purchasers in today’s market may have very different time horizons from those purchasing in 1987.

[67]Dr Bracke’s model was intended to control the characteristics and attributes that influenced price and his broad approach was not invalidated by the change in the conditions but it was necessary to assume that relativities did not change over time and that they were invariant to structural changes in asset and financial markets and in the nature and the motivations of purchasers.

[68]Commenting upon the nature of Professor Holly’s peer review Professor Lizieri commented that there had been direct interaction between Dr Bracke and Professor Holly which had clearly been an iterative process. This was more akin to a research collaboration than to a normal form of peer review. Professor Lizieri noted that Professor Holly had arrived at similar results but it was important to stress that different specifications, for example the introduction of further variables or specification of an alternative functional form would result in different results each of which would also be a mathematical certainty within the constraints of those models and specifications.

[69]In conclusion, Professor Lizieri had been able to replicate some of Dr Bracke’s results and had no reason to question the technical application of the regression models. However he had a number of concerns. Hedonic regression was critically dependent upon the specification chosen on data inputs and the large range of possible biases. There was considerable uncertainty about the reliability of much of the data, the extent of omitted variables, the appropriate functional form of the relationship between explanatory variables and price, interaction events and about possible spatial relationships in the data.

[70]The curve fitting procedure masked variability in the individual year by year estimates and there must be doubts about relativities calculated from data that was more than 20 years old when applied to today’s very different market.

[71]In considering the reliability of a new method, it was important to check the results against prior work or market practice. That was not to say that change should not occur, but when the results were substantially different from many other studies then that may be of some concern. The relativity graphs and tables collected within the RICS report, despite using different methods and data sources showed strong similarities. Of the six or seven relativity graphs listed, all but one indicated relativity at below 80% for a leasehold property with a similar unexpired term to the subject property. The majority showed a relativity of around 75% which was substantially below the 86.39% advanced in Dr Bracke’s report. This must cast some doubt on the applicability of his results. Dr Bracke had indicated that his work was still ongoing and as and when new specifications were introduced or new variables added the results would change. Whether the point estimates of relativity would change significantly is unknown, but the results had already changed from those put forward to the LVT. This must be relevant in considering whether reliance should be placed on a “work in progress” which produced results out of line with prior models.

[72]However during the course of his evidence Professor Lizieri commented that respect was due to Dr Bracke for the amount of time and diligence he had devoted to his work; that there were many strengths in Dr Bracke’s work; that he had no reason to dispute that Dr Bracke had adopted a technically very competent way to analyse the time period he has chosen; and that if one was seeking to establish how relativity varied with unexpired lease lengths during the period 1987 to 1991 it would be appropriate to consider Dr Bracke’s evidence but one should not solely rely upon it. Professor Lizieri stated that his main concern regarding Dr Bracke’s work was the shape of the curve which he ultimately obtained, which did not look rational from an economic point of view having regard in particular to the way in which the line of the graph followed an almost horizontal track above about 60 years (especially on the graph relating to Model 3).

[73]In a supplementary report, Professor Lizieri outlined his analysis of the confidence intervals of Dr Bracke’s Model 3. He noted that Model 3 grouped unexpired lease terms in bands of three years and that, even with this grouping, the confidence intervals were much wider than those of the other models. For lease terms of 49-51% the upper bound was as high as 117%, and for 55-57 years the lower bound is as low as 67%. There was a wide variation and volatility of the estimates. Professor Lizieri considered that the masking of the variability by the curve fitting procedure was even more of a concern in respect of Model 3 than the other models. Mr Oliver French

[74]Mr French is a Chartered Surveyor and is an associate director in the valuation team of Savills based in the West End. He is part of the team responsible for the strategic asset management of the Phillimore Estate and was especially involved in providing recommendations and implementing actions to maintain and where possible to enhance the portfolio’s value. This incorporated all valuations, advice and negotiation of leasehold enfranchisement claims and rent reviews. Mr French gave evidence before the LVT. His approach to valuation was influenced by the Tribunal’s decisions in Arrowdell , and Nailrile , and numerous graphs of relativity. Accordingly Mr French tried to determine a reliable relativity from comparable evidence but had also had regard to graphs of relativity.

[75]In the four years prior to the valuation date, Lonres recorded only three houses in the entire W8 postcode which had been sold leasehold. Since the valuation date there had only been five. None of these were on the Phillimore Estate and Mr French considered none to be comparable to the subject property. Therefore he was not able to deduce a relevant relativity from real world comparable evidence. Even had leasehold transactions been available, he would have had serious concerns about their application due to the subjective adjustment that would be required to discount for the “no Act world”. Mr French referred to the RICS report which commented that open market evidence of enfranchiseable sales was so tainted by the effects of the Act that an accurate adjustment could not be made and that any adjustment would be subjective and not based upon fact.

[76]Owing to the lack of leasehold evidence around the valuation date, Mr French considered transactions on the Phillimore Estate in the timeframe adopted by Dr Bracke of January 1987 – December 1991. However he highlighted his serious concerns over the historic nature of this evidence. Even though leases in that area may not have qualified for enfranchisement, values would have been affected by the ability to extend or acquire a freehold on a voluntary basis. Mr French understood from valuers and agents that were active prior to 1993 that this was not unusual. Indeed he had assumed that it had been the case in the subject lease.

[77]Knowledge of possible enfranchisement was in the market well before the 1993 Act came into force which inflated the value of short leases. Enactment of legislation took some time and the Landlord and Tenant Act 1987 included the first steps towards collective and enfranchisement. Other documents were in existence including consultation papers and by March 1991, with a general election looming tenant pressure groups had waged an effective campaign demanding action. The Department of the Environment had produced a leaflet on enfranchisement in July 1991. It was clear that there was a real and significant expectation towards enfranchisement of leaseholders of long leasehold flats. Sales within the 1987 – 1991 period would have been inflated because of this. Arguably there should therefore be a discount for this implicit but very real expectation. Mr French said that he had realised that the “no Act world” should assume that the lease which is the subject of the claim does not have enfranchisement rights but other leasehold properties do. The evidence would therefore not accurately reflect the hypothetical “no Act world” basis because in such a market the subject property would compete with other leases which had the right to enfranchise. A greater discount would then be applied to that subject property. This is evidenced by a schedule of transactions which showed that in the five year period concerned there were only seven freehold transactions within the Phillimore Estate compared with nineteen leasehold transactions. Whereas within a five year period around the valuation date, 2009 to 2013, there were 30 freehold house sales and no leasehold. There was therefore a fundamental difference in the market between the two periods.

[78]Within the period of 1987-1991 there was a significant number of transactions on leasehold interests in W8. Acquiring the lease of a house was very common. Mr French drew up a schedule of transactions on the Phillimore Estate by street to attempt to draw conclusions as to relativity. However the results were not logical in that a leasehold house in Argyll Road or Essex Villas would appear to sell for more than a freehold. On Campden Hill Road it is only after adjusting by Dr Bracke’s bedroom analysis that the indicated relative change suggested that 31 year lease was only worth slightly less than a freehold which clearly could not be correct.

[79]Other difficulties arose and from this Mr French concluded that no reliability could be placed on the transactional evidence identified on the estate between 1987 and 1991. He considered that this gave force to Professor Lizieri’s comment that some of Dr Bracke’s data points suggested relativities that were nonsensical.

[80]Next, Mr French researched a number of open market sales of houses which were unenfranchiseable which would potentially give a better representation of the statutory assumption to ignore enfranchisement rights. He was able to point to eight transactions of unenfranchiseable leases, bench marking these against the Gerald Eve/John D Wood relativity table. The results were wide ranging, ranging from 30.7% below the table to 7.4% above, or after adjusting for market growth, condition and amenities, 26.6% below to 8.1% above. These results also identified that leasehold values may be more than freehold values which clearly could not be correct. Mr French concluded that it was not possible to draw any reliable conclusions from the transactional evidence of unenfranchiseable sales.

[81]Mr French then turned to the graphs of relativities and the results of the RICS report following Arrowdell , together with the 1996 Gerald Eve/John D Wood graph and working paper.

[82]The Gerald Eve/John D Wood table indicated a relativity at 52.5 years unexpired of 76%. The RICS Report suggested the following relativities, in each case interpolated from data at 50 years and 55 years. W A Ellis 72.75% Knight Frank 77% Cluttons (flats) 74.35% Cluttons (houses) 73.35% Gerald Eve 76% J D Wood 82.5% Charles Boston 76.5% The average was 76%, or 75% if the J D Wood graph was excluded.

[83]The John D Wood working paper showed a number of graphs which plotted the Gerald Eve 1996 graph, the John D Wood settlements graph and the J D Wood Tribunal’s decision graph which were all closely related. The Tribunal’s decisions 2011 update graph plotted relativity at approximately 77% for an unexpired term of 52.5 years. More accurately this can be calculated at 76.7%.

[84]In summary, the three attempts to draw conclusions from comparable evidence were not reliable or conclusive and accordingly in accordance with Arrowdell and Nailrile Mr French had turned to the graphs of relativity. This was the most commonly used method by the vast majority of experts. It was accepted that the graphs were not certain but they had been applied and tested in many claims and tribunals and remained the favoured approach owing to the faults and flaws of using market data. It was arguable that they had themselves become “market evidence” because the market based enfranchisement transactions upon them.

[85]From all of the above Mr French’s opinion of relativity was 76% which led to an enfranchisement premium of £2,763,890.

[86]The last section of Mr French’s expert report dealt with his concerns with the approach taken by Dr Bracke. First Mr French considered that Dr Bracke’s analysis was based upon the incorrect hypothesis that the “no Act world” assumed that the lack of enfranchisement rights extended to comparables as well as the subject property and that no discount had been applied to differentiate between the two. The Tribunal’s preference in Arrowdell was for real world evidence that “can usefully be applied” which did not mean transactions some 20 to 24 years prior to the valuation date. In addition, the data was taken from transactions occurring within a market in which there was a real expectation of enfranchisement legislation. Implicit in the values was an expectation to be able to extend the lease. Accordingly a discount would need to be applied.

[87]Market characteristics in Prime Central London were very different at the valuation date from 1987-1991. At the valuation date there was more international demand, the evolution of an ultra prime market, the role of PCL as a safe haven, and greater value on higher levels of specification. Mr French considered that different economic conditions and fundamental differences in the market would cause a change in relativity, with the effect that greater discounts would need to be applied to short lease values that would not qualify under current enfranchisement legislation. The market for short leases was on any view different in 1987 – 1991 compared with the valuation date. There were significantly higher proportions of leasehold transactions during the earlier period than around the valuation date. That was a fundamental market change.

[88]Mr French had considered data from Lonres, for properties in the same postcode districts used by Dr Bracke, for the periods 1987-1991 and 2009-2013. In the earlier period, there were 3,192 flat sales and in the later period there were 7,153. Approximately half of the flat transactions had to be excluded due to unknown lease length.

[89]There was a substantial change in the composition of the market between the two periods. Around the valuation date a significantly higher proportion of marketed leaseholds had an outstanding term of over 100 years given the effect of lease extension activity. Consequently shorter leases were significantly less common. The tenure data for houses in the earlier period was less complete. Even assuming that the unknown tenure entries in the earlier period were freehold, there was a substantial shift towards freeholds given the progressive impact of enfranchisement and showed the extent to which leasehold houses are a rarity in the current market. A further point was the divergence of flat and house values. Dr Bracke’s research comprised far more data of short leases from flats than houses but the reverse for long leasehold ownership. This was likely to distort results because in the earlier period there was not the same difference in flat values compared to houses. At the valuation date many buildings within PCL were being converted to houses as there was a significant profit to be made. Mr French believed that in the earlier period the same difference in value did not exist. These factors would lead to distortion in the results as would whether or not a property had been refurbished, whether or not there was a garden, whether a flat was in a modern block or a period building.

[90]Mr French had concerns about the sample of data and the number of streets. There was a significantly higher number of streets than that for which the Lonres data had been obtained. From the Lonres data there were 668 streets of which 213 had only one entry – which Dr Bracke had ignored. But of the 484 streets which had five entries or less, Mr French did not consider that to be sufficient to obtain a representation and accurate result for that location.

[91]He was equally concerned about the reliance on asking prices and questioned Dr Bracke’s assertion that the average difference was between 4.01% and 4.85%. Mr French considered that an accurate range would be 12% to 2.5%.

[92]Mr French considered that there had been insufficient analysis of the properties. There was no allowance in Model 1 for condition, floor level, outside space, parking etc. Model 2 adopted some further attributes but these were not extensive. The only allowance for comparing size was on the number of bedrooms, which was inaccurate. A further inaccuracy was created by Dr Bracke including data where it seemed that a long leasehold interest was worth more than the freehold. There were also issues in terms of verification of the data which Mr French considered was insufficient.

[93]He considered there were discrepancies between Dr Bracke’s report for the subject property and the report he had produced for flat 7, 101 Mount Street. He had also seen four other reports of Dr Bracke and the number of transactions analysed changed in each one. Mr French said there was general concern in the community of enfranchisement specialists with Dr Bracke’s general approach. He considered that the Tribunal should give it no weight at all.

[94]Mr French recognised that the calculation of the appropriate relativity to apply to the unexpired 52.45 year lease of the property at the valuation date involved the comparison between two values, namely the value (which has been decided and is therefore a given) of the freehold with vacant possession and the existing lease value. A decision as to the appropriate relativity to apply is a means of discovering what this existing lease value was at the valuation date. However he accepted that a more direct method of discovering the existing lease value on the valuation date was to consider how much a hypothetical purchaser would pay for the existing lease at the valuation date if it was then offered for sale in the open market by a willing seller on the statutory assumptions. If the existing lease had been offered for sale in this manner Mr French accepted that it is likely that a prudent and reasonable hypothetical purchaser, who would on any basis be preparing to spend well over £10 million on the purchase, would take professional advice from an experienced and qualified valuer expert in prime central London house values as to how much it would be appropriate to bid for the existing lease. Mr French said that a valuer who was consulted by such a hypothetical purchaser would inevitably take into account the graphs which had been published by the RICS and which were in existence as at the valuation date and would be likely to advise the hypothetical purchaser that an appropriate amount to bid was an amount which was in line with these graphs. The parties’ submissions

[95]We are grateful to both counsel who produced for us written opening submissions and written closing submissions.

[96]On behalf of the appellant Mr Gallagher advanced the following submissions.

[97]In closing Mr Gallagher conceded a point which he had originally indicated an intention to argue, namely the point regarding whether the existing lease value was to be assessed(i) on the assumption that the existing lease had no Act rights and that other leaseholds available in the open market at the valuation date also had no Act rights, or

(ii) on the assumption that the existing lease had no Act rights but that other leaseholds available in the open market at the valuation date did in fact have Act rights. We do not need to dwell upon this point however because, in closing, Mr Gallagher conceded (in our view correctly) that the position was as in (ii) above.

[98]Accordingly the point was not argued before us and we do not have to make any decision upon it. It is however appropriate for us to record our provisional view that, for the following reasons which can be briefly expressed, Mr Gallagher was correct in making this concession. The relevant extract from section 9(1A) is set out at paragraph 2 above. As pointed out by Mr Cowen the assumption to be made involves an assumed sale by a willing seller at the valuation date “in the open market”. This reference to the open market would appear to be a reference to the open market as it existed on the valuation date not such market as would have existed if on the valuation date Part 1 of the Leasehold Reform Act 1967 as amended conferred no right to acquire the freehold or an extended lease in respect of any lease which was in existence on the valuation date. Also the phraseology of paragraph (a) of section 9(1A) would seem to make clear that the assumption to be made, namely that “this Part of this Act conferred no right to acquire the freehold or an extended lease” is an assumption directed towards “the tenancy” subject to which the fee simple is being sold rather than an assumption directed towards all tenancies.

[99]As regards the effect of the position being as in (ii) above however, Mr Gallagher argued that the proper analysis was as follows. Dr Bracke’s work was a reliable guide as to what was the proper relativity to apply when valuing the existing lease, because the existing lease must be assumed to have no Act rights and Dr Bracke’s study is concerned with exactly such a property, because all of the leasehold properties in his study were leaseholds without Act rights. Mr Gallagher accepted that if a lease did possess Act rights then this was a feature which increased the value of the lease above the value it would have without Act rights. But this meant that if, for instance, there was a lease of a similar term (52.45 years) of a similar property but with Act rights available for sale at the valuation date then the value of this hypothetical alternative lease would be increased above the value indicated by the relativity given by Dr Bracke’s work. He argued that there was no reason to believe that Dr Bracke’s work gave the relativity to apply to this hypothetical alternative lease with Act rights such that the relativity to apply to the subject lease (without Act rights) had to be diminished below the figure shown in his work. Also, in so far as at the valuation date the mix of properties available in the market had changed such that short leases were less available and long leases/freeholds more available than as at 1987-91 then a short lease would have acquired a scarcity value which would work to increase rather than decrease the value of such a lease.

[100]Mr Gallagher submitted that there were three important questions to ask in relation to Dr Bracke’s work namely:(a) (A) Is hedonic regression an appropriate technique to calculate relativity?

(b) (B) Is Dr Bracke’s analysis of a dataset of over 8000 flats and houses sales from the “no-Act” world robust and reliable?

(c) (C) What is the reliability of the application of Dr Bracke’s work to valuations as at the valuation date both in absolute terms and by comparison to the graphs published by the RICS and relied upon by Mr French? Mr Gallagher expanded upon each of these points in summary as follows.

[101]As regards point (A) Mr Gallagher pointed out that both Prof Holly and Prof Lizieri agree that hedonic regression is an appropriate technique to use and that other well-recognised bodies use such methods.

[102]As regards point (B) Mr Gallagher pointed out that Prof Holly and Prof Lizieri had replicated Dr Bracke’s results; that Dr Bracke had attempted to “kill the result” but that the result was robust and lived on; that Dr Bracke’s Model 1 and Model 2 had around 8,000 data points and that Model 3 (which excluded unverified prices) had over 2,600 data points and, although throwing up larger confidence bands, was consistent with the results of the earlier models. Mr Gallagher compared the number of data points examined by Dr Bracke with the far fewer data points used by the various compilers of the graphs published by the RICS. Dr Bracke’s work had been the subject of detailed study by Prof Holly and Prof Lizieri and if there was some significant defect in the work such a defect could be expected to have emerged from this level of scrutiny. Dr Bracke had justified his choice of a fourth order polynomial and had shown the robustness of this choice by also examining a fifth order polynomial curve and a local polynomial curve. Dr Bracke had performed various robustness checks. It is important to have firmly in mind that Dr Bracke’s objective was to estimate the ratio of the value of leasehold properties to freehold/long leasehold values, rather than to obtain a graph from which to value individual properties. Various criticisms could be made of certain detailed inputs or choice of variables, but there was nothing to indicate that any such criticism applied disproportionately to short leaseholds as compared with freeholds/long leaseholds or vice versa. As regards criticisms of the dataset used by Dr Bracke Mr Gallagher reminded the tribunal of Prof Holly’s observation, namely that when carrying out statistical analysis of data it is invariably possible to conclude that it would be nice to have more or more accurate data, but the statistician had to examine the data that was available and that one must not let the best be the enemy of the good.

[103]As regards point (C) above Mr Gallagher pointed out that, in so far as Dr Bracke’s work was criticised for being based upon historic market transactions between 1987 and 1991, it should be observed that certain of the graphs published by the RICS also used such historic data but Mr French nonetheless relied upon them. Also Mr French did not appear to prefer such graphs as included contemporary observations nor was Mr French clear as to whether (and if so how) these published graphs had been revised over time to incorporate more contemporary material. Mr Gallagher drew attention to observations of the Lands Tribunal and the Upper Tribunal in various cases, in particular Arrowdell Ltd v Coniston Court (North) Hove Ltd LRA/72/2005, Nailrile Ltd v Cadogan

[2009] 2 EGLR 151

, Re 38 Cadogan Square

[2011] UKUT 154 (LC)

, and Re Coolrace Ltd

[2012] 2 EGLR 69

. From these cases Mr Gallagher extracted the approach of the Lands Tribunal and the Upper Tribunal namely that for the purpose of assessing relativity the Tribunal must do the best it can with any evidence of transactions which may be available and that graphs of relativity should not be used in preference to any reliable open market evidence, but that in the absence of such evidence then reliance can properly be placed upon the graphs. He drew attention to the invitation by the Lands Tribunal to the RICS expressing the hope that the RICS might be able to reach conclusions upon definitive graphs of relativity, to the inability of the RICS to reach such conclusions, and to the fact that Dr Bracke’s work could be said to be an important response, based upon the study of real transactions in the open market, to the Lands Tribunal’s invitation. Mr Gallagher emphasised that Dr Bracke’s work was based upon open market transactions, which was not the case for most of the graphs published by the RICS. As regards these graphs Mr Gallagher drew attention to the fact that several of them included settlement evidence, which should be treated with caution. As regards these graphs he advanced in summary the following criticisms:a. WA Ellis:- the graph was based upon transactions involving only 200 houses, the data was from mid-1980’s, and the methodology was not clear;

b. Knight Frank:- the graph was based upon settlements and upon decisions of LVTs and the Lands Tribunal from 2002 onwards;

c. Cluttons:- the graph was based upon settlement evidence from 1996/1997 onwards;

d. John D Wood:- the graph was based upon settlement evidence during 1978 to 2004 (including John D Wood opinion where explicit agreement on components was not possible) and upon LVT and Lands Tribunal decisions but did not include transaction evidence;

e. Charles Boston: - the graph was based upon 1993 to 2009 settlements and transactions and LVT decisions.

[104]The existing graphs have not been endorsed by the RICS – they have merely been published by the RICS together with a statement that the working party had not been able to reach a conclusion on definitive graphs.

[105]As regards Mr French’s evidence that relativity may have changed over time, Mr Gallagher submitted that nothing in the published graphs seemed to recognise that. As regards Prof Lizieri’s evidence regarding the change in the profile of potential purchasers in the market at the valuation date as compared with 1987/91 Mr Gallagher submitted that it would be wrong to assume that the person bidding the highest would be someone looking at the property as an investment – the successful purchaser may well be a person who actually wants to live in the property. As regards Prof Lizieri’s point regarding deferment rates being lower as at the valuation date than in 1987/91, that was once again a “property portfolio” approach to the valuation of the existing lease – it was the approach of an investor rather than someone actually wanting to live in the property.

[106]Mr Gallagher submitted that Dr Bracke’s work was amazingly detailed and careful and expert and presented a rare opportunity at least to improve upon if not provide the answer to the seemingly intractable problem of assessing relativities and giving effect to the no Act world assumption. The fact that his work was a work in progress was indicative of a proper piece of scientific research. He submitted that the alternative to adopting Dr Bracke’s work was a depressing continuation of the present unsatisfactory approach of attempting to find contemporary market evidence or attempting to make ultimately arbitrary adjustments from real world transactions and then, having failed in these attempts, reaching for the graphs. He submitted that care must be taken to ensure that the graphs did not become self-fulfilling and perpetuating. On this latter point he argued that it would be wrong to approach the assessment of the existing lease value by proceeding on the assumption that, as at the valuation date, the hypothetical purchaser would have taken valuation advice and would have been advised broadly in accordance with the graphs, because this would indeed be to make the graphs self-perpetuating. Also the successful hypothetical purchaser may well not feel constrained to limit their bid to the average of the graphs.

[107]On behalf of the appellant he invited the Tribunal to accept Dr Bracke’s work and to apply the relativity derived from that work in preference to applying the graphs, alternatively to give substantial weight to Dr Bracke’s work in combination with a consideration of the graphs. He drew attention to the fact that, while Dr Bracke invited acceptance of his work and rejection of the graphs, Dr Bracke recognised (as a potential fallback position) that his work could also be used alongside the graphs, whereas Mr French’s evidence was that no weight whatever could be given to Dr Bracke’s work. Mr Gallagher submitted that this inflexible response from Mr French did not add to the weight that should be given to Mr French’s evidence.

[108]On behalf of the respondents Mr Cowen advanced the following submissions.

[109]He accepted that, technically, Dr Bracke had carried out an interesting and scientifically competent statistical analysis. However this does not mean that his work gave a realistic estimate of the appropriate relativity in respect of the property at the valuation date.

[110]He drew attention to Dr Bracke’s financial interest in the outcome of the case (we have already dealt with this aspect earlier in this decision). He also drew attention to the responsibility which the Upper Tribunal has to promote consistent practice in land valuation matters as recognised by Carnwath LJ in Sportelli in the Court of Appeal. He pointed out that the vast majority of cases where relativity is an issue ultimately settle on the basis of relativities as indicated by the existing graphs which were well understood and used by experienced valuers who practice in the field. He invited the Tribunal to be very slow potentially to reverse the present well understood and accepted position by adding Dr Bracke’s graph to the existing graphs.

[111]Quite apart from the potential problems of using Dr Bracke’s work, which is based upon market transactions in 1987-91, for the purpose of assessing relativity as at the valuation date (as to which see below), there were problems with Dr Bracke’s work even so far as concerns the assessment of relativities as at 1987-91. Mr Cowen drew attention to the large number of variables, to the colossal number of potential combinations of the variables, to the fact that some of the variables are interrelated rather than independent, and to the fact that Models 1 and 2 rely upon data from Lonres which was unverified and unreliable. As regards Model 3, which concerned only verified transactions, here the confidence bands were much wider and, if one looked at the data points before the fitting of a curve, these showed practically no trend such that the shape of the curve depended upon the choice of polynomial equation used. He also drew attention to two particular variables in Dr Bracke’s work where, he submitted, it was clear that the allowance for such variable should be different as between short leaseholds on the one hand and long leaseholds/freeholds on the other hand. One was the variable for onerous ground rent (which would not affect the freehold). The other was the variable for “in need of refurbishment”, because if a property was in need of refurbishment this would operate to affect the value of a short leasehold disproportionately more than the value of a long leasehold/freehold. Also it is too broad brush an approach to divide all of the 8,000 odd transactions into either refurbished or in need of refurbishment. It is true that Dr Bracke relies upon a sophisticated averaging exercise, but it cannot (or at least has not) reached a sufficient level of sophistication – for instance a flat on the fourth floor with a lift is averaged with a flat on the fourth floor without a lift. The residuals in Dr Bracke’s work indicate that a substantial amount of the model (between 15% and 18%) it is not explained by the results. There is a large margin of error.

[112]While it may be a cliché it is still true to say that valuation is an art and not a science. It is necessary to stand back and see whether a provisional result which has been reached accords with a valuer’s knowledge and experience of how the market behaves. Dr Bracke is not a valuer nor an estate agent and has no experience of buying and selling property in the market. Dr Bracke has not been able to stand back and to consider as a matter of valuation judgement whether the results can be correct. Mr French does have that experience and expertise and is able, on standing back, to say that Dr Bracke’s conclusions cannot be correct. There are obvious problems with certain of the data points, for instance on Model 1 at 75 years unexpired the point estimate is 105%, at 65 years it is 96% but at 68 years it is 79% and at 50 years it is 96.6%. For Model 3, where only verified data is used, the issue is exacerbated. For the period 49-51 years the point estimate is 101.9% and for 52-54 years it is 98.7%. No valuation thought has gone into these results and they are obviously wrong. If the points themselves do not make sense then no significant weight can be placed upon a curve which is fitted to the points. This is especially so when there is reason, as a matter of valuation judgement, to reject the curves as being illogical – the Model 1 curve shows relativity peaking at around 80 years and then dipping and the Model 3 curve seems to be effectively horizontal from about 50 years to 80 years.

[113]Quite apart from these difficulties with Dr Bracke’s work even if such work is being used solely for the purpose of assessing relativity during the period 1987-91, there are further difficulties which make it inappropriate to seek to use the work for the purpose of assessing relativity as at the valuation date in 2011. The RICS working paper on relativity noted that “changes in economic conditions may impact on freehold or leasehold values to a greater or lesser extent, thereby affecting relativity”. In the present case there is evidence that in the 1987-91 market there were relatively more sales of short leases and fewer sales of long leases/freeholds as compared with the position as at the valuation date; that the type of buyer in the market was increasingly as at the valuation date looking to acquire a freehold or long leasehold and less likely to want a short leasehold as compared with the situation that existed in the 1987-91 market; and that there is a significant difference in the economic climate in general and deferment rate in particular between the 1987-91 market and the market as at the valuation date. There is also the separate point that the statutory assumptions under section 9(1A) require the existing lease value to be assessed on the assumption that the existing lease is offered for sale without any Act rights in a market where other leaseholds do carry Act rights.

[114]The existing graphs are well-established, they have been prepared by those with experience of the market, and they have been used in countless decisions in the Upper Tribunal and First-Tier Tribunals (and their predecessors). They are not used as a last resort. They can properly be used in conjunction with market evidence (if market evidence is available), see for instance 38 Cadogan Square especially at paragraphs 79-80. If valuers considered that the published graphs generally understated relativity levels we would have heard about it.

[115]What the appellant, through Dr Bracke, is suggesting in the present case is that all of the learning in the graphs as published by RICS is wrong, that all of the valuers with their wealth of experience in practice and as members of LVT panels have been applying the wrong relativity by a factor of about 10% and that the existing graphs should in effect be laid on one side and Dr Bracke’s work should be used instead. It is notable that there is not a single scrap of corroborative valuation evidence produced by the appellant to support Dr Bracke’s work notwithstanding the criticism by the LVT on this very point. It is proper to infer from this that such corroborative evidence is not available. Conclusions

[116]We remind ourselves that we must decide this appeal only on the basis of the evidence and arguments that have been presented to us. It is a notable feature of the case that no valuation evidence from any witness with valuation expertise has been called before us on behalf of the appellant - this is so despite the LVT in paragraph 54 of its decision observing that Mr Ingram-Hill (who before the LVT was relied upon as an expert witness in contrast to the position before us where he was only a witness of fact) had provided no evidence save for reliance on the research carried out by Dr Bracke and further observing that the LVT would have expected him to have provided evidence of graphs even by way of a cross-check on Dr Bracke's research. During this case there was much reference to the various graphs published by RICS including the graph produced by John D Wood (which is the graph which indicates a relativity significantly higher than the other graphs). Mr Ingram-Hill is a chartered surveyor in the valuation and surveying department of John D Wood, but he did not give any evidence explaining or supporting his own firm’s graph. Thus the appellant presents us with Dr Bracke's work (as supported by Professor Holly) and then asks us to apply the relativity suggested by that work for a 52.45 year lease when calculating the existing lease value of the property at the valuation date. It is also notable that the respondent's evidence comprises evidence from Professor Lizieri (who critically appraises Dr Bracke's work but who does not give any evidence as an expert in valuation matters) and from Mr French, who is a valuer but who relies only upon the graphs published by RICS because he has tried, but failed, to find any helpful comparable evidence from market transactions. These graphs relied upon by Mr French have not been formally approved by RICS as providing definitive guidance on calculating relativity. Further the respondents have not called any evidence from any person with knowledge of the construction of any of these graphs who could explain for instance the nature of each input point on the graph and the spreads between points, i.e. whether the line on the graph followed closely all the points or whether it is some form of best fit in relation to points which are widely spread (and if so, how widely). In the present case the Upper Tribunal observed, when granting permission to appeal, that:

"In any event the issue of "relativity" is of considerable importance and Dr Bracke's approach is likely to be relied on in other high value cases. It is appropriate that it be considered by the Tribunal."

[117]In the light of this it is unfortunate that no valuation evidence whatever from a valuer has been called from the appellant and that no evidence explaining and justifying any of the graphs published by the RICS has been called by the respondents. We must decide this case on the material before us. However the absence of this more extensive valuation evidence will lessen the assistance that the present decision may provide in other cases and may result in a similar dispute about relativity being contested in a future case on more extensive evidence.

[118]The question has been posed before us in summary as a question of deciding on the appropriate relativity. However the question which is before us is this: What is the existing lease value or, stated more fully, what is the amount which at the valuation date the existing lease (52.45 years unexpired) if sold in the open market by a willing seller on the statutory assumptions in section 9(1A) might be expected to realise.

[119]The merit of Dr Bracke's work in assisting to find the existing lease value depends upon the following:(1) Upon the merit of Dr Bracke's work in assessing the appropriate relativity to apply to a 52.45 year lease of the property supposing that the task was concerned with the appropriate relativity in the market as it existed between 1987 and 1991 rather than as at the valuation date.

(2) Upon whether Dr Bracke's work, supposing it had merit in helping calculate the appropriate relativity for the 1987/91 market, can reliably be used at the valuation date to find the appropriate relativity for calculating the existing lease value of the property at the valuation date.

[120]We will consider these two points in turn.

[121]As regards point (1) this depends upon the force of various strands of criticism of Dr Bracke's work which can for convenience be described (using a phraseology adopted during the hearing) as micro criticisms and macro criticisms. As regards micro criticisms it must be remembered that the experts agree that: "1.1 hedonic regression is an appropriate technique for estimating the impact of property attributes or characteristics on the price of those properties; 1.2 The output of a hedonic regression is determined by the data employed and the model specified; 1.3 Dr Bracke's report provides an accurate representation of the outcome of the models he has run on the Lonres/John D Wood data used in his analysis; 1.4 Professor Lizieri was able to replicate Dr Bracke's results." (See the agreement at page 542 of the bundle).

[122]It is also right to remember that Professor Lizieri paid respect to Dr Bracke for the amount of time and diligence he had devoted to his work and observed that there were many strengths in Dr Bracke's work and that he had no reason to dispute that Dr Bracke had adopted a technically very competent way to analyse the time period he had chosen. Also in closing submissions Mr Cowen accepted that what Dr Bracke had done was technically perfectly valid and he did not seek to dispute that it was a serious minded analysis. We also have in mind Professor Holly's observation that one should guard against letting the best be the enemy of the good - i.e. one should avoid rejecting a piece of work merely because (in perhaps an ideal world) more extensive or more accurate data might be available to be the subject of the statistical analysis. Having said that, we accept that there are a significant number of micro criticisms and that some of them bear weight. The standard answer given by Dr Bracke to most of the micro criticisms was to the effect that, yes the point had some validity but that there was no reason to believe that the point impinged disproportionately on short leasehold values as compared with long leasehold/freehold values such that it affected equally both the numerator and the denominator of the fraction obtained by comparing leasehold values at a certain lease length with freehold values. We mention two micro criticisms here, being criticisms which in our view cannot be satisfactorily dealt with through this standard answer, namely: (1) Dr Bracke includes onerous ground rent as a variable. However he has directed himself that a ground rent of more than 0.1% of the freehold value should be treated as an onerous ground rent (which is contrary to what is contemplated by the RICS at page 490 of the bundle). Also by definition a variable for onerous ground rent will affect leaseholds but will not affect freeholds. The value given by the hedonic regressions for onerous ground rent was 11.4% in Model 2 and 17.1% in Model 3, so the factor was one of some substance although only, of course, affecting a limited number of properties. (2). There is a variable included for whether the property is refurbished or in need of refurbishment. We see some force in the criticism that it is too broad brush an approach to categorise all of the 8,000 plus properties into either refurbished or in need of refurbishment, but more importantly we see force in the argument that the effect of a need for refurbishment would depress the value of a short leasehold proportionately more than the value of a long leasehold/freehold.

[123]However if one were to consider solely the micro criticisms and one were solely concerned with whether Dr Bracke's work could be reliably used to assist in calculating relativity for a 52.45 year lease in the 1987-91 market, we would conclude that these micro criticisms are not of such a substance as seriously to undermine the value of his work.

[124]As regards the macro criticisms these principally consist of the argument that one must stand back and look at the results of Dr Bracke's work and must ask whether they make sense as a matter of economics and valuation. We refer to the points made by Mr Cowen as summarised in paragraph 112 above. The point estimates of relativity given for certain periods of unexpired lease are unrealistically (and in some cases impossibly - more than 100%) high. Dr Bracke says that this may happen from a statistical model but what is important is the trend as shown by the curve which he has fitted through the various data points. However the curve brings with it its own problems. The Model 1 curve shows relativity peaking at around 80 years and then dipping, which is contrary to expectations. The Model 2 and 3 curves seem to be almost horizontal between about 50 years and 80 years. Dr Bracke suggested that the reason some of his results were contrary to expectations was that there was a general misconception as to what the results should be – in other words our expectations were wrong and his results were right rather than vice versa. Mr French as a valuer says that these results cannot be right. We agree with Mr French that these point results cannot be right and we further agree that an effectively flat graph between 50 and 80 years (suggesting that the value of a 50 year lease is equal to the value of an 80 year lease) also cannot be right. It does not follow from this that no weight can be placed on Dr Bracke's work, but it is clearly a matter of substantial concern.

[125]Pausing at this stage in our examination of Dr Bracke’s work we consider that, so far as concerns the market in 1987/91, Dr Bracke’s work does satisfy us on the balance of probabilities that the value of a 52.45 year lease in that market would have reflected a relativity somewhat higher than the average relativity shown for such a lease by the graphs subsequently published by the RICS. We deliberately use the broad expression “somewhat higher” because, while we accept Dr Bracke’s work indicates this somewhat higher relativity as a generalised qualitative feature, we do not accept the quantification of that feature as presented in his work, because of the matters mentioned in the previous paragraph.

[126]We now turn to the question raised in paragraph 119(2) above namely whether, on the assumption that Dr Bracke's work has merit in helping to calculate relativity for the period 1987/91, the work can be reliably used for the purpose of calculating relativity as at the valuation date. The points raised by the respondents in relation to this can be summarised as follows:(1) The point regarding alleged anticipation in 1987/91 of an extension of enfranchisement rights (thereby inflating the value of short leaseholds).

(2) The point concerning the fact that section 9(1A) requires there to be assessed the amount which the property, if sold on the open market by a willing seller at the valuation date, might be expected to realise assuming that the existing lease carried with it no Act rights but that the market in which it was being offered for sale included other leaseholds which would carry with them Act rights (in Dr Bracke’s study all the relativities derived were for a lease without Act rights offered for sale in a market where all other leases also lacked Act rights).

(3) The point concerning the difference in the market, as between 1987/91 on the one hand and the valuation date on the other hand, of the mix of properties available with short leases forming a much higher proportion of properties on the market at the earlier dates as compared with the valuation date.

(4) The point concerning whether the make-up of the pool of potential purchasers in the market, and the aspirations of such purchasers, are different at the valuation date as compared with the earlier dates.

(5) The point concerning changes in the economic climate in general and in discount rates in particular as between 1987/91 and the valuation date and whether these may impact upon the existing lease value at the valuation date.

[127]We will take these points in turn.

[128]Whilst it was raised in evidence, point (1) above was not greatly pressed. Also Dr Bracke did a robustness check by adding in the 1992 transactions (when anticipation of an extension of enfranchisement rights would in particular have existed, as compared with earlier years) and found no significant difference in his analysis. No quantification has been placed by Mr French upon this alleged effect nor have we received any valuation evidence from a valuer practising at that time that at some particular time during the period 1987/91 purchasers of leaseholds became more enthusiastic, and prepared to increase their bids, because of the prospect of an extension of enfranchisement rights. We do not find this a point of any significant weight.

[129]As regards point (2) above, we consider that it would have a significantly negative impact upon the value of the existing lease of the property that the lease (52.45 year lease) must be assumed to be offered for sale without any Act rights in circumstances where other leasehold properties were being offered for sale with Act rights. It was suggested by Mr Gallagher to Mr French(i) that Dr Bracke's work gave a reliable relativity for the existing lease of the property at the valuation date with no Act rights and

(ii) that the (admittedly more valuable) similar lease of a hypothetical property with Act rights would command a higher relativity than that shown in Dr Bracke's work, rather than the existing lease of the property commanding a lower relativity than that shown in Dr Bracke’s work. Mr French answered that by pointing out that in so far as a hypothetical 52.45 year lease of a similar property but with Act rights was to command a significantly higher relativity than that shown by Dr Bracke's work, then this 52.45 lease would have a relativity of over 90% which could not be right. We see force in Mr French's evidence.

[130]As regards points (3) and (4) above they can in our view be taken together. We accept that as at the valuation date there was a higher proportion of long leaseholds/freeholds on the market and a lower proportion of short leaseholds on the market as compared with the market mix as at 1987/91. We do not accept Mr Gallagher's argument, on the basis that if something is in short supply its price should go up because of scarcity value, that the value of the existing lease is actually enhanced by reason of being one of comparatively few short leases on the market as at the valuation date. We consider that the reason the mix of properties on the market has changed is because a freehold or long leasehold is what purchasers want if they can get it at an acceptable price. We also see force in Professor Lizieri's evidence that the make up of at least one element of those buying properties in prime central London, namely those wishing to lock up in a safe haven substantial capital sums with a view to growth, is a further reason why a short lease (52.45 years) would be somewhat less attractive in the market as at the valuation date as compared with its level of attraction in the market as at 1987/91. This lessening of attractiveness would be compounded by the feature already discussed in the previous paragraph, namely that this 52.45 year lease would be offered for sale without Act rights in a market where other leases did carry with them Act rights. We accept that this category of investor in property is not necessarily the person who will become the successful hypothetical purchaser, but we do see this feature as identified by Professor Lizieri as being a significant distinction between the market as at 1987/91 and the market as at the valuation date.

[131]As regards point (5) above, Professor Lizieri points out the difference in discount rates at the valuation date as compared with 1987/91. We do not see this feature as a major contributor to the ultimate price that a hypothetical purchaser would bid for the existing lease, but some weight must be given to the point because a hypothetical purchaser may well have in mind the possibility of coming to a voluntary agreement with the freeholder for the purchase of the freehold and the lower the discount rate the higher the likely price that the hypothetical purchaser would be advised he might have to pay in order to do so.

[132]Bearing in mind the matters discussed above, it is necessary to consider what if any weight can be accorded to Dr Bracke's work when seeking to assess the appropriate relativity to apply in order to calculate the existing lease value at the valuation date. This raises the question as to whether, on the balance of probabilities, the generalised qualitative feature which we have identified in paragraph 125 above (namely that in the market of 1987/91 the value of a 52.45 year lease would have reflected a relativity somewhat higher than the average relativity as shown for such a lease by the graphs subsequently published by RICS) is a feature which would have persisted into the market as at the valuation date so as to be applicable to a 52.45 year lease offered for sale without Act rights in a market where other leases did carry Act rights. The matters discussed above (and in paragraphs 129 and 130 in particular) lead us to conclude that we are not satisfied on the balance of probabilities that this generalised qualitative feature would have persisted so as to be applicable in this manner. In our view it would not have done so. Accordingly we conclude that Dr Bracke’s work is unable to provide any reliable assistance to us in relation to the existing lease value as at the valuation date. We now turn to consider the valuation evidence presented on behalf of the respondents which comprises the evidence of Mr French and his reliance upon the graphs published by RICS.

[133]In the case of Arrowdell the Lands Tribunal said at paragraph 57:

“It ought, we believe, to be possible to produce standard graphs, distinguishing between mortgage-dependent markets and those that are not so dependent, on the basis of a survey of assessments made by experienced valuers addressing themselves properly to the hypothetical no-Act world. We express the hope that the Royal Institution of Chartered Surveyors may find itself able to carry out such an exercise and to produce guidance in the form of standard graphs that can readily be applied by valuers in carrying out enfranchisement valuations. Such graphs could be used as evidence by LVTs, with the relativities shown being applied by them in the absence of evidence compelling the adoption of other figures.”

[134]The RICS established a working group (comprising various experienced valuers and chaired by a Queen’s Counsel) to consider that request. The RICS reported:

“1.4 It has not been possible for the members of the Working Group to agree upon definitive graphs that could be used for this purpose. Therefore, this report draws together the various graphs of relativity that are in existence together with details of the data that lies behind each. It is hoped that this will provide useful guidance to practitioners considering them.”

[135]If RICS had been able to agree upon definitive graphs this would plainly have been a highly important consideration. However RICS was unable to do so. Mr French, as we understood him at one time in his evidence, appeared to contend that RICS was giving its approval in some way to the graphs which it published – this approval was said to flow from the expression of hope that their publication would provide useful guidance to practitioners considering them. We do not consider that the RICS report can be read as giving some imprimatur of approval, carrying the authority of RICS, upon the graphs which it published. The question remains however as to what weight should be accorded to these graphs.

[136]The point made by Mr French, namely that these graphs have been in existence for a substantial time and are widely used by those practising in the field and by leasehold valuation tribunals and indeed by the Lands Tribunal and now the Upper Tribunal, is a point which we accept carries some weight. We accept that these graphs do provide evidence of some value towards deciding the point before us. However the RICS itself in its report recognises various difficulties with the graphs including:a. that there are a number of graphs in existence compiled by various organisations which may give quite disparate readings at a given lease length;

b. that the supporting data from which the graphs are generated may be scattered quite widely on either side of the line;

c. that there is not necessarily any data to support the position of the line at every lease length;

d. that some of the lines are computer generated best-fit lines and some are hand drawn—and as regards those which are hand drawn these may incorporate the valuer’s view as to how the market would behave between the points for which data exists;

e. some of the graphs rely upon settlement evidence – however settlement evidence is open to criticism where the analysis is not agreed between the parties and in addition settlements may be influenced by the Delaforce effect and may become self-perpetuating;

f. some of the data going to make up the graphs may be criticised for being historic;

g. some of the graphs incorporate valuers’ views, or the views of those consulted, as to the level at which the transactions would have taken place as opposed to the level of actual sales.

[137]We consider that these potential weaknesses individually, and especially cumulatively, are of concern. Of particular concern is a point stressed by Mr Gallagher that we simply do not know how widely spread the data points were on each graph and how good a fit the curve is.

[138]It has been clear from an early stage that in the present case the appellant was mounting an attack upon the use of these graphs and was advancing an argument that a new piece of work, namely Dr Bracke’s hedonic regression analysis, should be preferred or should at least be given substantial weight. The respondents have directed much evidence and argument towards giving detailed analysis to and criticism of Dr Bracke’s work and the graphs which he ultimately obtains. However when it comes to the graphs upon which the respondents rely there is a conspicuous absence of any substantial evidence to allay in respect of any graph the concerns mentioned in subparagraphs (a) to (g) above. We do not overlook the explanatory text published with each graph, but that is notably limited and goes no significant way in dealing with the concerns at (a) to (g). In substance the respondents have done no more than say: market transactions and comparables cannot help so here are the graphs as published by RICS - please accept those.

[139]There is however a point which leads us to conclude that the published graphs would constitute an important ingredient in the decision of potential hypothetical purchasers of the existing lease at the valuation date as to how much to bid for the existing lease. At the valuation date these graphs were in existence and, as Mr French says, were widely referred to and relied upon by valuers. A hypothetical purchaser considering bidding in the open market for the existing lease at the valuation date on the statutory assumptions would be aware that they were likely to have to pay substantially more than £10 million for the prospective purchase and would be likely to conclude (unless rash – which we should assume they were not) that they should obtain some valuation advice as to how much would be appropriate to pay before ultimately deciding upon their final bid. The potential purchaser seeking such advice would inevitably go to a valuer experienced in valuing prime central London houses. Such a valuer in giving the advice which was sought would in our judgement inevitably have in mind, as one of the ingredients which informed the advice, these published graphs.

[140]We conclude that the advice which the well-informed hypothetical purchaser would be likely to obtain would include advice that these graphs existed, that they had some strengths but also some weaknesses, and that some suggested prices higher than others.

[141]We remind ourselves that the successful purchaser is the purchaser who, consistent with being prudent and well-informed rather than rash, placed the highest bid. But that is equally the case for the transactions, and to an extent the assumptions behind settlements, reflected in the graphs. Mr French confirmed that the graphs are showing relativity points after an allowance for no Act rights has been made, and no further adjustment was required.

[142]We note that the average relativity as shown across the various graphs is approximately the 76% adopted by the LVT and supported by Mr French. We observe that one graph, namely the John D Wood graph, shows a relativity of about 82.5% which is substantially higher than that shown by all the other graphs. We conclude that the successful hypothetical purchaser would not be prepared to base their bid upon this graph, because they would see that it was out of step with all these other graphs. However we conclude that the successful hypothetical purchaser would be the purchaser who was prepared to base their bid upon the average of all the graphs referred to by Mr French, including the John D Wood graph and who would in consequence outbid the more cautious potential purchaser who was only prepared to bid to the level of the average of the graphs excluding the John D Wood graph. We therefore determine that 76% is the relativity as decided upon by the LVT. Accordingly the appeal is dismissed.

[143]We would conclude by saying that this Tribunal, its predecessor, the LVTs and indeed the profession at large has, unsuccessfully thus far, been seeking to find a settled position on relativities for leasehold properties. Dr Bracke is to be commended for his efforts in this regard. We found him to be an impressive and (despite suggestions to the contrary) unbiased witness and his work is, by common consent, a formidable piece of research.

[144]However we have been unable to place weight upon his evidence in the present case for the reasons set out above. It is of particular significance that no valuation evidence was submitted on behalf of the appellant to explain and support either of the following two propositions namely that(a) Dr Bracke's work produced reliable quantitative guidance for relativities in respect of the 1987/91 market despite the substantial concern referred to at paragraph 124 above and

(b) that his work, which studied relativities in the market of 1987/91, could give reliable guidance upon the relativity relevant for calculating the existing lease value of the property at the valuation date in October 2011 on the statutory assumptions. Dated 7 August 2014 His Honour Judge Huskinson P D McCrea FRICS P D McCrea FRICS SCHEDULE 1 LRA/108/2013 TRIBUNALS, COURTS AND ENFORCEMENT ACT 2007 IN THE MATTER OF AN APPEAL AGAINST A DECISION OF THE FIRST-TIER TRIBUNAL (PROPERTY CHAMBER) His Honour Judge Huskinson and Mr Peter McCrea FRICS 25 June 2014 BETWEEN: MRS LATIFA KOSTA Appellant and (1) Mr F.A.A. CARNWATH (2) Mr MT Chamberlayne (3) The Honourable JHT RuSsell (as trustees of the Phillimore Estate) Respondents and Parthenia Research Limited Intervening Applicant ORDER UPON the application by Parthenia Research Limited dated 23 June 2014 for orders to preserve the confidentiality of certain confidential information consisting of the following documents and matters, each disclosed to the Respondents, (and their experts and legal advisors) in the course of this appeal (collectively, “the Confidential Information”):

a. A. the confidential John D Wood database,

b. B. the cleansed datasets (as described in the report of Dr Bracke dated 30 April 2014),

c. C. the exact details of the steps taken to cleanse the datasets (but not the fact of nor the idea of that cleansing, nor in general terms what the purpose of that cleansing was),

d. D. the precise details of Dr Bracke’s methodology (but not the idea of hedonic regression, or in general terms the use of that technique in the way it is used in that methodology),

e. E. the Stata programming code that runs Dr Bracke’s methodology, and

f. F. the general results of Dr Bracke’s analysis (for a wide range of lease terms) as shown in table 4, figure 9, figure 10 and figure 13 of his report, which reveal information about the bulk dataset or reveal the results of Dr Bracke’s analysis. AND UPON hearing counsel for Parthenia Research Limited, counsel for the Appellant and counsel for the Respondents. IT IS ORDERED THAT

[1]The Parties, their legal advisors and the Respondents’ experts shall not publish or disclose the Confidential Information other than for the purposes of this appeal, provided that this prohibition shall not apply in relation to any details of the Confidential Information that the Upper Tribunal refers to in its decision.

[2]The Parties, their legal advisors and the Respondents’ experts shall not use the Confidential Information other than for the purposes of this appeal provided that this prohibition shall not apply in relation to any details of the Confidential Information that the Upper Tribunal refers to in its decision.

[3]The preceding two paragraphs of this Order shall apply notwithstanding the fact that the Confidential Information or any part of it may have been read to or by the court. His Honour Judge Huskinson P D McCrea FRICS Dated: 25 June 2014

Relativity v 95% Confidence Band [2014] UKUT 319 (LC)

(b) to (f) ……………”

Cited in 7 later judgments