[adapted from the document of March 2010]
ANNEX IISupplementary provisions
ANNEX II [adapted from the document of March 2010] Reply to the submission of 19 October 2009 prepared by LECG The reply of the consulting company LECG to the Commission’s analysis focuses on four issues: (1) Alleged erroneous use of the CAPM model. While LECG accepts that a CAPM-based analysis is the correct avenue to pursue, it questions the parameter values that have been used by DG COMP. More precisely, LECG asserts that: — the equity premium should have been lower — the ‘betas’ should also have been lower (2) The Commission should not have limited itself to Consultia IT’s business plan, but should also have used Arthur Andersen’s 2000 business plan to establish the relevant benchmark. (3) The Commission did not use the appropriate financial ratios to estimate the expected profitability of the CDL project. (4) Furthermore, the Commission did not take into account the expected revenues deriving from the development of hotels and office space. 1. The parameters of the CAPM 1.1. Equity premium LECG questions our choice of using historical data relating to the premiums, based on Fernández (2004), to estimate the market risk premium in Spain for the years 2000 and 2004. Instead, LECG suggests using the information found in a later paper by the same author ( 1 ) . As will be seen below, the LECG report quotes this paper very selectively. The LECG report argues that the required equity premium should be used instead of the historical equity premium. However, Fernández (2009) states the following: ‘It is easy to conclude that there is not a generally accepted equity premium point estimate or a common method to estimate it. […] The recommendations regarding the equity premium, taken from 150 finance and valuation textbooks published between 1979 and 2009, range from 3 % to 10 %. Several books use different equity premiums in different pages and most books do not distinguish among the four different concepts that the phrase ‘equity premium’ designates: historical equity premium, expected equity premium, required equity premium and implied equity premium. […] 129 of the books consider expected and required equity premium to be synonymous terms and 82 do not distinguish between expected and historical equity premium.’ Furthermore, as regards its concrete value, the LECG study proposes using an equity premium recently presented by Fernández (2009). Fernández suggests using a Required Equity Premium (REP) ranging from, 3,8 % to 4,3 %. There are at least two reasons for which this range is not valid for our purposes. First, it suffers from intrinsic limitations, since it is a subjective valuation by a single individual. It should be noted that Fernández also presents an average value obtained from 150 textbooks (see graph below). Secondly, and most importantly, the suggestion to use a value of 3,8 %-4,3 % is an estimation of the current equity premium, not the one prevailing in 2000 and 2004. The values of the parameters to plug in the CAPM should be contemporaneous to the date of the business decision (i.e. ex-ante and not ex-post ). In other words, the relevant equity premium is the one that was prevailing in 2000 (or 2004), not 2009. With respect to that point, Fernández’s paper (2009) does provide consensus estimates of the REP in 2000 and 2004. In particular, the 2004 value of the REP is above 6 %, whereas its value for 2000 is around 7 %. Our estimation before this paper was made public was correct, and was in this range, i.e. 6,8 %. ‘The figure below shows the evolution of the Required Equity Premium (REP) moving average (last 5 years) used or recommended in 150 finance and valuation textbooks.’ Moving average 5 years 1.2. Beta As far as the beta is concerned, the LECG report casts doubts on the use of estimations provided by private companies specialised in valuations. Below, we provide the links to the studies we refer to: Carrere (the information we used can be found on p. 5 of the report below): http://www.carreregroup.com/fr/documents/financials/Etude_Financiere_13062001.pdf Studio Babelsberg (see p. 10): http://www.studiobabelsberg.com/uploads/media/Studio_Babelsberg__14May08.pdf In our view, this is precisely the place where a private investor would look for information to establish a beta ex-ante . These analysts have specific skills to assess ex-ante valuations. The two studies provide betas for certain close competitors of Ciudad de la Luz. Pablo Fernández, the author extensively, but selectively, quoted in the LECG study, has written widely to show that historical betas (as the ones proposed by LECG) are next to meaningless. In his paper ‘Are calculated betas worth for anything?’ (Working paper IESE Business School), Pablo Fernández (2008) presents strong arguments against the use of calculated beta in valuations: ‘We show that, in general , it is an enormous error to use the historical beta as a proxy for the expected beta. First, because it is almost impossible to calculate a meaningful beta as historical betas change dramatically from one day to the next; second, because very often we cannot say with a relevant statistical confidence that the beta of one company is smaller or bigger than the beta of another; third, because historical betas do not make much sense in many cases: high-risk companies very often have smaller historical betas than low-risk companies ; fourth, because historical betas depend very much on which index we use to calculate them. […] ( 2 ) ’ (our emphasis) The third argument above put forward by Fernández (2008) also serves to illustrate that the betas presented by LECG cannot be right. Indeed, Fernández makes the assertion that high-risk companies are high beta. This is very intuitive: the required return for a high-risk project (such as CDL) ought to be above the return of a diversified portfolio (e.g. the IBEX 35). In another paper, focused on Spain (‘On the instability of betas: the case of Spain’, 2008, Working paper IESE Business School) Pablo Fernández shows again that it is a serious error to use betas calculated from historical data to compute the required return to equity. ‘It is a mistake for seven reasons: 1. because betas calculated from historical data change considerably from one day to the next. 2. because calculated betas depend very much on which stock index is used as the market reference. 3. because calculated betas depend very much on which historical period (5 years, 3 years,…) is used to calculate them. 4. because calculated betas depend on what returns (monthly, yearly,…) are used to calculate them. 5. because very often we do not know if the beta of one company is lower or higher than the beta of another. 6. because calculated betas have little correlation with stock returns. 7. because the correlation coefficients (and the R2) of the regressions used to calculate the betas are very small.’ The paper provides numerous examples to support all of the seven statements. 2. The relevant time frame In our previous analysis, we used the year 2004 as a reference year for the calculation of the NPV, as this was the time when the entire investment of EUR 199,4 million had been committed. However, we can easily duplicate our analysis for the year 2000. The resulting WACC would be higher than in 2004 ( 3 ) , due both to a higher market risk premium (which according to the information gathered by Fernández should be around 7 % — see section 1.1) and to a higher risk-free rate (which stood at 5,53 % for Spain in 2000 ( 4 ) ). Hence, an internal rate of return of 8,84 % would still be much lower than the cost of capital (more precisely: ). It is possible to go one step further and compute the NPV of the project in 2000. Bearing in mind all the caveats we stressed in our previous analysis (i.e., the use at face value of the cash flows and the parameters used for the calculation of the terminal value, except, of course, for the WACC) we obtain the following negative NPV: 2000 2001 2002 2003 2004 2005 2006 NPV Cash flow –46 479 –69 719 9 005 12 616 6 930 7 710 68 698 Discount 1 16,66 % 1 0,8572 0,7348 0,6298 0,5399 0,4628 0,3967 DCF –46 479 –59 762,56 6 616,67 7 946,13 3 741,50 3 568,16 27 252,87 –57 116,22 3. Financial ratios The LECG report agrees with our point of view that CAPM is the right instrument to assess the ex-ante cost of capital. Moreover, the report does not provide any argument to rebut the use of the ROCE as a proxy for the WACC. Therefore, it is not necessary to repeat the reasoning already set out in the previous note. However, it may still be worth recalling why the ROCE is the most appropriate ratio. Since the economic profitability of a company is measured ex-post by the difference between the return on capital employed and the cost of capital, then intuitively, if one were to construct a benchmark based on past data of competitors in order to determine the minimum return required by investors, then the appropriate proxy would be the average industry ROCE. 4. The extension of the investment to hotels and office spaces The assertions in this point are based on new evidence, namely six land valuation reports. As a preliminary remark, we note that this evidence is produced ex-post , and therefore was not part of the information available to a potential investor at the time ( 5 ) . Secondly, we note that the Arthur Andersen report only mentions the possibility of developing a hotel complex (in phase III) and certain office space (in phase IV). The report contains an estimate of costs for developing hotels and offices but none of the associated revenues; hence the information to assess the profitability of developing the land is missing. Indeed, no business plan is provided for the development of phases III and IV. In the Consultia IT business plan, the reference to the hotels is made only indirectly through the mention of an area of complementary services, while offices are not even mentioned. Furthermore, these activities are not considered at all in the section on the economic and financial forecasts. In other words, at no stage would the private investors have been presented with a business plan concerning the development of the land. As such, this implies that the ex-ante assessment of the project should solely be based on the business plan concerning the film studios; indeed, all the relevant evidence indicates that land development was not part of the original business plan. In fact, the LECG submission of April 2008 explicitly stated, in section 5 ( The operational value of a hotel complex and commercial area in Ciudad de la Luz ) that the details of the investment in the services complex (hotels and commercial area) were not analysed in either of the two business plans because (i) the above mentioned services would in any event be built after the studio and (ii) the profitability of the studio was high enough and therefore it was not necessary to analyse at that moment the additional cash flows Ciudad de la Luz might generate. The only pertinent information relevant for the assessment of the land is represented by the bids received by Ciudad de la Luz in 2005 for the construction and operation of hotels and a commercial area, which were presented in section 5 mentioned earlier. This represents an evaluation made by the market in a period contemporaneous with the investment decision, a period moreover characterised by the real estate boom. As will be explained in the next section, the Euroval valuations do not offer proper guidance regarding the value of the land plots. For the purpose of calculating the NPV including this additional land development, it is possible to use the price set by the market (the fact that it is relatively low, as recognised by the Spanish, is indeed evidence that this property development did not seem to be very attractive for private investors). The following table provides NPV calculations that include the property development on the basis of the bids that were actually made in 2005 (EUR 560 000 per year, adjusted for inflation, as stated by LECG). As can be readily seen, the NPV remains negative. 2002 2003 2004 2005 2006 2007 2008 Cash flow – 884 –1 792 –2 601 – 118 177 –55 813 –25 088 4 462 IPC hotels 1,000 1,021 1,042 1,064 Hotels 560 572 584 596 Total – 884 –1 792 –2 601 – 117 617 –55 241 –24 504 5 058 Discount 1 14,91 % 1,3204 1,1491 1,0000 0,8702 0,7573 0,6591 0,5735 DCF –1 167,26 –2 059,19 –2 601,00 – 102 355,76 –41 835,77 –16 149,82 2 901,01 2009 2010 2011 2012 2013 2014 NPV 5 526 6 561 6 842 8 057 8 307 80 447 1,087 1,110 1,133 1,157 1,181 1,206 609 621 634 648 661 9 401 6 135 7 182 7 476 8 705 8 968 89 848 0,4991 0,4344 0,3780 0,3290 0,2863 0,2491 3 061,91 3 119,74 2 826,09 2 863,46 2 567,38 22 383,73 – 126 445,48 A series of land valuation reports were prepared by Euroval. These reports were commissioned by the Sociedad Proyectos Temáticos de la Comunidad Valenciana, the investor in Ciudad de la Luz (CDL). The reports purport to provide an objective valuation of the land that would be used for the construction of hotels and offices in the CDL complex. They provide a valuation for the years 2000, 2002, and 2009. For each of these years, two plots are valued: one destined for the development hotel complex, and the other for office blocks, making for a total of 6 valuation reports. Since the reports broadly follow the same methodologies, some of our comments are generic; whenever reference is made to specific assertions, the corresponding report is identified. This memo describes the limitations of the aforementioned reports and concludes that the valuations presented therein are untrustworthy. In addition, the memo notes that the valuation report is not compatible with some of the claims found in the documents previously provided by Spanish authorities. Last, the memo indicates the kind of evidence that could have been used to obtain an estimate of the value of the land. 4.1. The reports As can be seen, Euroval has not signed nor stamped any of these reports, i.e. there is no evidence at all that the entity takes responsibility for the valuations. The Spanish authorities state that the valuation reports on the land drawn up in 2000 and 2002 do not fulfil the requirements of Ministerial Order ECO 805/2003 regarding the valuation of buildings, ‘since their purpose is other than the scope of application of that Order’ (e.g. p. 3 of the report providing the 2000 valuation for office space). It is not explained why the aforementioned ministerial order should not apply. The aforementioned ministerial order is also mentioned in conjunction with Law 6/1998 on land valuation (p. 13 of the aforementioned report). It is argued that, since it is a retrospective valuation for the year 2000 made in 2002, ‘the acting technical team is free to apply the calculation adjustments and hypotheses it considers most appropriate for the objective pursued.’ However, the ‘adjustments’ and ‘hypotheses’ are not spelled out. Furthermore, it is not clear why existing law can not be applied, particularly with regard to the valuation of the land. Nor is a justification provided for not applying the laws that were in force at the time. The report providing the current (2009) valuation of the plot destined for office blocks contains an even more surprising statement. It is claimed that the valuation methods applicable according to the current legislation in force can be adapted because: ‘since the purpose of the valuation is a non-monetary contribution at the date in question, the acting technical team is free to apply the calculation adjustments and hypotheses it deems most appropriate for the objective pursued.’ In other words, since the valuation will not lead to a monetary transaction, the valuation methods that have to be applied by law can be modified. To be more precise, since no real money/transaction is involved, the technical team admits that they consider that they can make adjustments that they deem opportune given the ‘objective pursued’. It is therefore no wonder that Euroval has not officially endorsed any of the six reports. 4.1(a) Reports pertaining to the building of office space (NNT plot) The reports should have also explained more clearly what the qualification of ‘tertiary use’ for the land entails. The most glaring limitation of these reports is the way they establish benchmark prices for office space in the Ciudad de la Luz complex. On p. 18 of the 2000 report, nine transactions are listed, including the price paid per square meter. These transactions took place between December 1999 and September 2000, and the average price paid was of EUR 1 111,7 per square meter. This is the value taken to represent the benchmark for office space in the Ciudad de la Luz complex. The Ciudad de la Luz complex is outside the city of Alicante and next to an industrial complex ( polígono industrial ). By contrast, the nine transactions mentioned above involve offices (average size: 115,26 square meters) located in the most prestigious business streets in Alicante. Using these transactions to establish a benchmark for the 66 576,54 m 2 that would be built in Ciudad de la Luz simply does not make sense. The report uses the Spanish Consumer Price Index (CPI — Indice de Precios al Consumo, IPC) to deflate current values. In particular, the IPC is used to deflate a current valuation of the land back to the year 2000 (p. 24). We note that, as its name indicates, the CPI/IPC does not encompass asset price inflation; the index is computed on the basis of a basket of consumer goods. Using the CPI/IPC to deflate land values and/or real estate values is far from being innocuous, for the reasons explained below. As is well known, Spain experienced a housing boom that lasted more than 10 years (to a notable extent, its current macroeconomic woes are related to that bubble). As a consequence land prices increased much faster than the general CPI. According to statistics gathered by the Ministry of Housing of the Spanish Government (Ministerio de Vivienda, http://www.mviv.es/es/), housing prices in the Alicante province increased by 122,2 % during the period spanning the first quarter of 2000 and the third quarter of 2009 (this time period has been chosen as it corresponds to the one used in the 2000 Euroval report: January 2000 to August 2009). A cursory glance at the time series indicates that house prices increased fastest at the start of the time period under consideration. The same statistical source also provides data on land prices; unfortunately, the series only begins in 2004. The average price per square metre of urban land in the Alicante province stood at EUR 234,3/m 2 during the first quarter of 2004. The corresponding figure for municipalities of more than 50 000 inhabitants in the province of Alicante stood at EUR 464,9m 2 during the first quarter of 2004. Since the price of land that can be used for real development of urban land is closely correlated to housing prices, it is possible to obtain an approximation to the deflator for land prices during the period Q1 2000 – Q4 2003 by looking at house price inflation during the same period. According to the Spanish Housing Ministry, house prices in the Alicante province increased by 80,1 % over the period between the first quarter of 2000 and the third quarter of 2003. Thus, irrespective of whether one takes the average land price for the entire province (EUR 234/m 2 in Q1 2004), or that for large municipalities (EUR 464,9/m 2 in Q1), applying a deflator of that order of magnitude (+/- 80 %) would result in a land price in Q1 2000 much lower than that reported in the Euroval report. It is also worth mentioning that, regarding the evolution of prices, the reports contradict each other. As mentioned above, on p. 18 of the 2000 report, nine transactions are listed, including the price paid per square meter. These transactions took place between December 1999 and September 2000, and the average price paid was of EUR 1 111,7 per square meter. This is the value taken to represent the benchmark for office space in the Ciudad de la Luz complex. The 2009 report follows the same methodology. On p. 18, seven transactions that took place between April 2007 and December 2008 are listed. The average value is EUR 2 736,22 per square meter. Thus, the 2009 report provides glaring evidence that real estate inflation was far above the 33,5 % used in the 2000 report. By comparing the two reports, we obtain inflation of 146,1 %. This is in line with the Ministry of Housing reports: between the first quarter of 2000 and the second quarter of 2008 (the period of most of the transactions reported in the 2009 report), housing prices in the Alicante province increased by 152,1 %. The fact that the figures found in the 2009 report are deflated by 15 % to establish current prices does not change the overall picture in view of the magnitudes involved. As a by the way, we would also reiterate that using prices for small offices in the centre of Alicante to establish a benchmark for the Ciudad de la Luz complex does not make sense. Two further comments should be borne in mind. Firstly, during property boom periods, the price of land usually increases faster that final house prices. In any case, the house price index understates land price inflation. Secondly, the average land price for large municipalities is probably not representative of land prices in a location such as that of CDL. Indeed, most of the underlying data used to construct this average price stems from land located in cities or their immediate vicinity (e.g. suburban ‘urbanizaciones’); CDL is located outside the city in a location with little (if any) residential housing. Lastly, house price inflation is illustrated by the material used to teach second-year undergraduates at Unversidad Carlos III de Madrid that can be found at: http://www.eco.uc3m.es/~ricmora/ee/, materiales, Tema I, slide 67, which we reproduce below. This graphical evidence further confirms that house prices have increased much faster than the CPI/ICP. LA VIVIENDA EN ESPAÑA: DESGRAVACIONES, BURBUJAS Y OTRAS HISTORIAS Original source: LA VIVIENDA EN ESPAÑA: DESGRAVACIONES, BURBUJAS Y OTRAS HISTORIAS , José García Montalvo. Gráfico 20. Crecimiento nominal de los precios de la vivienda e inflación. Fuente: Ministerio de Fomento e INE. P. vivienda Inflación In all the reports, the arguments used to show that the costs of land development are nil are rather thin, given that some additional costs would have to be borne. Finally, in the case of the office complex (NNT) a difference of 35 % (2002) or more than 50 % (2000) between the values obtained from the two valuation methods (dynamic residual method and increase in CPI) is indicative of the highly speculative nature of the exercise. 4.1(b) Reports on the valuation of the plot destined for a hotel complex (NNH plot) The 2000 and 2002 reports on the second plot (‘NNH’) suffer from the same limitations (e.g. use of an inadequate deflator, lack of justification for not applying valuation methods established in Spanish law, etc.). As mentioned above, none of these reports are signed or stamped. Some additional comments are nevertheless in order. The 2000 and 2002 reports establish benchmark prices for hotel rooms with the help of average prices. However, the source is not provided, nor is it clear whether it refers to Alicante province (2002 report), or the Comunidad Valenciana (2000 report). The resulting average prices are EUR 44,95/night in the 2000 report, and EUR 47,65/night in the 2002 report. Although the source underpinning these values has not been provided, they are not out of line with previous figures. By contrast, the 2009 report provides a list of 6 well located hotels, all of which are 4- or 5-star, to establish a benchmark price. The latter is established EUR 104/night. There is no explanation as to why the methodology to establish benchmark prices is different from the one used in the 2000 and 2002 reports. More importantly, it is hard to understand why the focus is on 4-5-star hotels. Is there any indication that demand would exist for 4-5-star hotels outside Alicante, located next to a highway and a ‘poligono industrial’? In the same line, it is quite telling that the only hotel close to the CDL complex has been omitted. The latter is an IBIS, whose prices range from EUR 49 per night for the summer of 2010 (http://www.ibishotel.com/gb/reservation/multirates.jshtml consulted on 24 February 2010). As further evidence, luxury apartments next to a golf course (and relatively close to CDL) charge EUR 80 per night in August 2010 (peak season, see http://www.hotelesoasis.com/WebOasis/EN/ficha_hotel/plantio_golf/descripcion.jsp, consulted on 24 February 2010). The 2009 report for the NNH plot also contain some surprising statements. One is initially pleasantly surprised to find out that the method to undertake the valuation is in conformity with the law (p. 13 reads ‘The methodological references are the criteria laid down in Ministerial Order ECO 805/2003 on real estate valuation and Law 6/1998 of 13 April 1998 on land valuation, the aim of the valuation being to establish the value at the present date.’ However, on p. 25, one discovers that, after all, the valuation report has not complied with the requirements set out in the Ministerial Order: ‘In accordance with Article 61(1)(b) of Order ECO 805/2003 of March 2003 on rules for real estate valuation, we would point out that this Valuation Report does not formally comply with that Order since its purpose is other than the scope of application of the Order’ (‘p or ser su finalidad distinta al objeto de aplicación de la misma ’). It is also interesting to note that there is a huge difference between the estimates per square meter for the two plots (NNH and NNT). While not denying that the projected uses (offices vs. hotels) may explain part of the difference, the gap is still striking. For hotels, the values range from EUR 144,12/m 2 (2000) to EUR 188,35/m 2 in 2009. For office space, the respective magnitudes range from EUR 491,64/m 2 (2000) to EUR 800/m 2 in 2009. What can explain the significant difference in the price evolution (inflation) of these two plots situated next to each other? How is it possible to account for this difference of over 300 %? Should we conclude that this is due to the fact that the methodology used to value the plot destined for hotels is closer to the standards established by the law? Or should it be put down to the greater transparency of the price of a hotel night? It is also claimed that variable costs represent 50 % of revenues (e.g. on pp 18-19 of the 2009 report). Where does the estimate of such large margins come from? Building costs are estimated on the basis of a benchmark price module. For the 2000 and 2002 reports, the correct figures are chosen. However, for the 2009 report, the 2006 figure is used. Why is this? 4.2. Risk premiums We also note that the reports recognise that investment in hotel facilities is riskier than investment in office space (e.g. on p. 12 and p. 26 of the 2000 report). At all times, the Spanish authorities have consistently envisaged the construction of a ‘hotel complex’. Similarly, the report indicates that the discount rate ( tipo de actualización ) is given by the following formula (p. 16 of the 2000 report): On the same page, a table provides the minimum risk premiums to be applied. The relevant premiums are above the ex-ante expected return on investment in Ciudad de la Luz. Lastly, the risk premium used in the calculations appearing on p. 21 of the 2000 report (10 %) is lower than the one appearing in the corresponding table on p. 16 (11 %). 4.3. Credible valuations In order to obtain a credible valuation of the land, valuations for around 2000 of various plots of similar characteristics in terms of size and location (next to the Aguas Amargas industrial complex), and intended for ‘tertiary use’ and ‘hotel use’, should have been provided. ( 1 ) Pablo Fernández (2009): The Equity Premium in 150 Textbooks . Working paper IESE Business School. ( 2 ) Damodaran (Damodaran, Aswath (1994), Damodaran on Valuation. New York: John Wiley and Sons) also makes this point by calculating the beta of Disney. With daily data, he gets 1,33; 1,38 with weekly data; 1,13 with monthly data; 0,44 with quarterly data; and 0,77 with annual data. With a 3-year period, he gets 1,04; 1,13 with 5 years; and 1,18 with 10 years. Also, the beta depends on the index taken as the benchmark; thus, the beta with respect to the Dow 30 is 0,99; with respect to the S&P 500, it is 1,13, and with respect to the Wilshire 5000, it is 1,05. ( 3 ) According to our calculations the WACC in 2004 was 14,91, using a risk-free rate of 4,1 % and a market risk premium of 6,8. ( 4 ) See OECD Statistics. ( 5 ) The LECG report of 25 April 2008 follows the same line, as it states at the outset (beginning of section 2.1.) that the relevant question for the application of the MEIP is ‘With the information available at the time the investment was made, did the project offer enough profitability to induce a private investor also to decide to undertake this investment? […] The investment decision must be analysed from an ex-ante perspective.’