Daniel SALLIER THE 2ND KENZA LAW OF DEMAND: A QUALITATIVE AND QUANTITATIVE APPROACH FOR ASSESSING LONG-TERM DEVELOPMENTS OF LEISURE DESTINATIONS MARKETS By Daniel SALLIER Aéroports de Paris ABSTRACT In this contribution, I introduce a set of two non-econometric but very closely related methods for demand modelling and forecasting. These methods are called the Kenza laws of demand. The 1st law is dedicated to the modelling of generic and hardly substitutable perishable goods or services, such as the worldwide set of foreign destinations of the British leisure market, for instance, while the 2nd law is to be used for more specific perishable goods or services, such as the set of Tunisian destinations of the French or German leisure markets. The 1st and 2nd Kenza law of demand have been initially used by Airbus S.A.S. in the late 90s and the 1st law represents, today, the primary demand short, medium and (very) long-term forecasting tool of Aéroports de Paris (CDG & Orly). Tourism represents a major component of many countries economy in the world, some European ones included: local airlines and hotels developments and profitability, real estate, local employment, hard currency reserves, etc... It simply means that airlines, airports, hotel groups, the State and local administrations, if not the financial community, do have very strong methodological requirements which can deliver fully supported qualitative and quantitative understanding of the today’s up to long term developments of the tourist demand. One of the main assets of the 2nd Kenza law of demand is to provide a better assessment and understanding of the different "ages" of any leisure market: from the times of being a fancy destination for very rich people to the times of being a very popular destination; from the times of a highly lucrative niche market to the times of huge volumes and low (unit) profits. To start with the empirical concepts leading to the 1st and 2nd Kenza laws of demand will be exposed and discussed. Then exclusive attention will be paid to © Association for European Transport and contributors 2010 1 Daniel SALLIER the 2nd Kenza law of demand and more specifically to market structural evolution over the time and the demand elasticity properties. To finish with we will have a look at the French, German, Italian, UK and Spanish leisure markets of Tunisian destinations as an illustration of the theoretical concepts. As the second Kenza law of demand is a direct and simple derivative of the first Kenza law it does not deserve specific bibliographic references with the only exception of T. Veblen work. 1. INTRODUCTION I started the research work on the 1st Kenza law of demand1 by the end of the year 1994. Beginning 1995, I was given the opportunity to apply it on a real case within the frame of an Airbus aircraft sales campaign for the renewal and expansion of Tunisair fleet. The issue was that a major component of Tunisair traffic is the leisure market of Western Europeans to Tunisia. The 1st Kenza law of demand had to be adapted to provide a better modelling of a leisure market and that is how we came up with the second law of Kenza which is the very subject of this paper. It was decided together with Tunisair fleet planning team that we will focus exclusively on the global leisure market from a set of individual European countries to Tunisia and that Tunisair market share evolution was no part of the scope of analysis. Of course, Tunisiar team had the academic background and the expertise to use classical econometric models for demand forecasting purposes, but the Kenza approach provided a "physical" understanding of markets behaviour, sometime paradoxical, which was highly praised and appreciated at that time. I take the opportunity of this paper for thanking again Tunisair and Mr. Moncef Ben-Dhahbi for their very friendly, unrestricted and even enthusiastic support over the years which is the reason why the market examples used in this article are all based on the Tunisian leisure market. The Tunisian market is just an example of a leisure market which is an important source of economical development in Tunisia (7% of 2009 GDP2). Beyond the Tunisian case, the tourism economy worldwide represents billions of US dollars or euros every year. Some orders of magnitude to start with. In 2007, the French tourist sector accounted for3: - 81.9 million foreign visitors having 1 overnight stay and more in France; 114 million same day visitors 18.5 million beds 199 million nights spent out of which 72.5 million by foreign visitors €39.6 billion expenditure by foreign visitor in France 833,683 employments attached to the tourist activity. © Association for European Transport and contributors 2010 2 Daniel SALLIER Tourist sector accounted for 6.2% of the French GDP in 20074, while it accounted for 10.8% of the Spanish GDP5 and 18% of the Greek 2008 GDP6 but only 2.68% of the US GDP in 2007 while being the 1st one in the world in terms of sector turnover (US$ 594.1 billion)7. Tourism shares with civil aviation the settlement of dedicated international organisations in charge of these economical sectors: World Tourism Organisation (UNWTO) for the 1st one and International Civil Aviation Organisation (ICAO) for the second. This is another clear indication of how important the tourism sector is considered worldwide. The second Kenza law of demand is directly derived from the first law with which it shares all the assumptions made among which: 1/ it is a modelling of annual passenger demand. It is not designed for modelling quarterly, monthly, weekly, daily, ... traffic and for taking into consideration traffic seasonality for instance; 2/ it is a modelling origin/destination leisure flows which means, for instance, that modelling leisure traffic flows between the USA and Western Europe would require, at least, two different models to be developed, one for the passengers living in the USA and a second one for those living in Western Europe. To start with, we will quickly remind the empirical concept at the origin of the 1st Kenza law of demand and we will detail those leading to the second Kenza law of consumer demand. We will pay a specific attention to the structural market evolution over the time as it results from the set of Kenza equations. Then we will look at the demand elasticity as it results from the second Kenza law of demand. To finish with we will use the Tunisian air markets to illustrate the ability of the 2nd Kenza law of demand to provide both a good estimate of demand data and the keys for a better understanding of market evolutions. 2. THE FIRST KENZA LAW OF DEMAND Content of this chapter is a synthesis of the second chapter of the paper "LongTerm demand forecasting: the Kenza Approach" by Sallier, D. (2010), ATRS 2010, Porto. The very idea at the origin of the empirical approach is that people are flying because they "feel like" or have to, they can afford it and they are sensitive to the relative ticket or inclusive tour price to their (annual) income. We assume that they are not sensitive to the absolute price. The following chart illustrates the successive steps of demand formation according to the 1st Kenza law of consumer demand: © Association for European Transport and contributors 2010 3 Daniel SALLIER In the former chart the red thick line F * rn represents the 1-complement of the cumulative distribution of normalised individual income – Kenza distribution – it is to say the percentage of the population of which the annual income is greater or equal to rn which is equal to the individual annual revenue r divided by the normalisation variable r which might be equal to the GDP per capita or the average/median annual individual revenue, … The equation of the 1st Kenza law of demand is: D t P t K1 F * K2 pn t Where: p pn r is K 2 pn inclusive-tour sale price, which is actual price divided by the normalisation quantity at a given date t; is the threshold of normalised income. K 2 is constant; is the proportion of elected population relative to the total population which can F * K 2 p n the normalised © Association for European Transport and contributors 2010 average ticket or 4 Daniel SALLIER P K1 D P K1 F * K2 pn afford to buy the considered service or the perishable good; is the total population; is a aggregated constant which aggregates both the proportion of actual consumers out of the elected population and the number of the average number of goods/services bought by each actual consumer. is the total number of service/good units actually bought/sold;. 3. THE SECOND KENZA LAW OF DEMAND While the 1st Kenza law of demand can provide a good modelling tool of the leisure demand of a given population for any destinations such as, for instance, the German leisure market globally considered, the 1st Kenza law cannot be used for modelling the demand for a given set of destinations such as the German leisure market to Tunisia for instance. It has to do with the fact that it is very likely that, generally speaking, the richer among the German population will tend to be attracted by more fancy destinations such as the Seychelles Islands for instance even if the service they are ready to pay for is not that different from the one offered in Tunisia. It has to do with self-consciousness of one's social status8. Of course this "conspicuous consumption" behaviour as named by T. Veblen is not at all a specificity of the German population and it can be generalised worldwide. It means that a second threshold of income should be considered in the 1 st Kenza law of consumer demand. While the first leftmost threshold of income keeps determining what the elected population is, the second rightmost threshold of income determines which part of the population will not go anymore to this set of destinations for not being "classy" enough. The following chart, illustrates the different steps of the leisure demand formation. © Association for European Transport and contributors 2010 5 Daniel SALLIER This chart translates into 2 different versions of the equation of the 2 nd Kenza law of consumer demand. 3.1 Equations of the 2nd Kenza law of consumer demand 1st version D pn P K1, L F * K 2, L pn K1, H F * K 2, H pn In this equation P K1, L F * K 2, L pn represents the demand which would result from the leftmost lower threshold of normalised income K2,L pn while P K1, H F * K 2, H pn represents this part of the demand defined by the rightmost upper threshold of normalised income K 2,H pn which is no more attracted by the considered destination. In this equation we should have: - K 2, H K 2, L being constant which means that the upper threshold is proportional to the lower one; © Association for European Transport and contributors 2010 6 Daniel SALLIER - K1, H K1, L being constant which means that only part of the richer customers are considering a more fancy destination K1, H K1, L or that all the richer customers consider a more fancy one K1, H K1, L This version of the 2nd Kenza law of demand is difference between the 2 normalised thresholds characterised by the fact that the pn K 2, H K 2, L is proportional to the price and narrowing over the time as the normalised price is decreasing. 2nd version D pn P K1, L F * K 2 pn K1, H F * K 2 pn pn ,min In this equation P K1, L F * K2 pn represents the demand which would result from the leftmost lower threshold of income K2 pn while P K1, H F * K 2 pn pn ,min represents this part of the demand defined by the rightmost upper threshold of income K 2 pn pn,min which is no more attracted by the destination. In this equation we should have: - K1, H K1, L being constant which means that only part of the richer customers are considering a more fancy destination K1, H K1, L or that all the richer customers consider a more fancy one K1, H K1, L This version is characterised by the fact that the difference between the 2 normalised thresholds of income K2 pn,min is constant. It is very likely that in fact consumer behaviour moves from the 1st version to the 2nd one as the relative price is decreasing. 3.2 Model calibration 1st version Kenza calibration process requires that K1,L, K1,H, K2,L and K2,H to be determines out of the sets of historical data. One way, a very classical one, is to minimise the sum of quadratic errors between actual data and their estimate. To start with, let us assume that we already know K2,L and K2,H values. The sum of quadratic errors e² is equal to: © Association for European Transport and contributors 2010 7 Daniel SALLIER e2 Di Pi K1, L F * K 2, L pn,i K1, H F * K 2, H pn,i i 2 where i index, refers to the date i in the set of data Di actual demand measured at the date i * * Pi K1, L F K 2, L pn,i K1, H F K 2, H pn,i Kenza demand estimate at the date i For a given value of K2,L and K2,H, e² is minimised for D P F K K1, L i pn ,i Pi F * K 2, H pn ,i Di Pi F * K 2, H pn ,i Pi 2 F * K 2, L pn ,i F * K 2, H pn ,i 2 * i 2, L i i i D P F K * K1, H i i Pi F K 2,L pn,i i Pi F K 2,H pn,i i Pi 2 F * K 2,L pn,i F * K 2,H pn,i 2 * i i 2, L i i * i 2 2 pn ,i Pi 2 F * K 2, L pn ,i F * K 2, H pn ,i Di Pi F * K 2, H pn ,i Pi F * K 2, L pn ,i P F K i * 2, H i 2 i 2 2 pn ,i Pi F * K 2, L pn ,i Pi 2 F * K 2, L pn ,i F * K 2, H pn ,i i i 2 K2,L and K2,H can be determined by using iterative approach to minimise e² value. 2nd version Kenza calibration process requires that K1,L, K1,H, K2 and pn ,min to be determines out of the sets of historical data. As for the 1st version, one way, a very classical one, is to minimise the sum of quadratic errors between actual data and their estimate. To start with, again, let us assume that we already know K2,L and K2,H values. The sum of quadratic errors e² is equal to: e2 Di Pi K1, L F * K 2 pn K1, H F * K 2 pn pn ,min i 2 where i Di Pi K1, L F * K 2 pn K1, H F * K 2 pn pn ,min index, refers to the date i in the set of data actual demand measured at the date i Kenza demand estimate at the date i For a given value of K2 and pn ,min , e² is minimised for © Association for European Transport and contributors 2010 8 Daniel SALLIER D P F K * K1, L i i 2, L i i P F K * i 2, L i D P F K * K1, H i i pn,i Pi F * K 2 pn pn,min 2, L 2 i pn ,i Pi F * i * i 2 i 2 i * K 2, L pn,i F * K p 2 n i P F K p i 2 pn,i Pi F * K 2 pn pn,min 2 i D P F K p n pn ,min 2 n pn,min Di Pi F 2 * i i 2, L i Pi 2 F * K 2, L pn,i F * K 2 pn pn,min i * i P F K pn ,min Pi 2 F * K 2, L pn ,i F * K 2 pn pn ,min K p 2 n pn ,min Pi F * K 2, L pn , i 2 i 2 pn ,i Pi 2 F * K 2, L pn,i F * K 2 pn pn,min i 2 2 And once again an iterative approach can be used to determine K2 and pn ,min values. 4 The 3 ages of ANY leisure market Whatever the formula used for the 2nd Kenza law of demand, we will get the same qualitative evolution of any leisure market. 1/ age of market development The 2 thresholds of income are located right enough on the Kenza distribution of income and are moving leftward not necessarily because of fares decreasing but of the population standard of life improving. This translates into a rather dynamic leisure destination which keeps developing, attracting visitors and making profit. 2/ age of market 1st stagnation The 2 thresholds of income are now located on Kenza distribution of income so that additional visitors/passengers induced by a higher standard of life, if not decreasing fares are balanced by the number of former visitors which do not feel like going to such a "popular" destination anymore. Most of the time the reaction of the tourism sector (airlines, tour operators, hotel, etc…) is to enter a fare competition which does not translate into a significant higher number of visitors while significantly altering the overall profitability. Last but not least the experts of the sector are very likely to misunderstand the very structural reason at work in this stagnation and are more than likely ready to consider it results from the competition of other sets of destinations. 3/ age of market decline & final stagnation Further to fare competition and increase of the population standard of life, the 2 thresholds of income have moved left enough on Kenza distribution of income for the additional visitors/passengers induced by a higher standard of life and decreasing fares are out balanced by the number of former visitors which do not feel like going to such a "popular" destination anymore. The following chart illustrates those 3 ages of any leisure market: © Association for European Transport and contributors 2010 9 Daniel SALLIER A quite interesting point here is that the "conspicuous consumption" as named by T Veblen is at work all the time but does not necessarily translate into the leisure destination behaving like a Veblen good. In fact depending on the 2nd Kenza law parameters, mostly the K 1, L , K1, H couple, the 3 ages of any leisure market may be more or less "contrasted". This is illustrated in the following chart. © Association for European Transport and contributors 2010 10 Daniel SALLIER The issue is that whatever the 2nd Kenza law parameters are, the demand will demonstrate a maturing process resulting in a structural decline (in absolute value) of demand elasticity to GDP and fares which will characterise a market segment which is less and less profitable. 5. DEMAND ELASTICITIES 1st Version Assumed that the Kenza distribution is steady over the time, derivation of the 1st version of the 2nd Kenza law leads to the following set of demand elasticities: dD pn dP D pn P * * dp K1, L K pn K 2, L F K 2, L pn K1, H K pn K 2, H F K 2, H pn p K1, L F * K 2, L pn K1, H F * K 2, H pn * * drn K1, L K pn K 2, L F K 2, L pn K1, H K pn K 2, H F K 2, H pn rn K1, L F * K 2, L pn K1, H F * K 2, H pn Where K xn is the intrinsic Kenza elasticity. Alike the 1st Kenza law of demand we have: 1/ demand elasticity to the population is equal to 1; © Association for European Transport and contributors 2010 11 Daniel SALLIER 2/ A "symmetrical" value of the demand elasticity to the price and the one to the normalisation quantity: the average revenue per capita or the GDP per capita most of the time. Empirical study9 of the Kenza distribution shows that the intrinsic elasticity is an almost linear function of the rate of elected population: K Tn A F * Tn B where A 0 and B A (demand elasticity equal zero for 100% of the population. Assumed that the two normalised thresholds of income are located on the rather linear section of the intrinsic elasticity of Kenza, the demand elasticity to price is equal to: K F* K p 2 K F* K p 1, L 2, L n 1, H 2, H n p A * * K1, L F K 2, L pn K1, H F K 2, H pn 2 1 In the very extreme case of K1, H K1, L which is a market which cannot secure the loyalty of the richest visitors, the demand elasticity to price is equal to p A F * K 2, L pn F * K 2, H pn 1 . For the highest normalised price we have F * K2, L pn 0.5 and F * K 2, H pn 0.5 which results in p 0 : a negative demand elasticity to price which translates into a "normal" behaviour of the market. On the other hand there is a normalised price threshold below which F * K 2, L pn F * K 2, H pn 1 which results in a positive elasticity of the demand to price and the leisure destination turns into a Veblen good. 2nd Version The following equation gives the different components of the demand elasticity. © Association for European Transport and contributors 2010 12 Daniel SALLIER dD pn D pn dP P * * dp K1, L K pn K 2 F K 2 pn K1, H K K 2 pn pn ,min F K 2 pn pn ,min p K1, L F * K 2 pn K1, H F * K 2 pn pn ,min K1, H K 2 pn ,min dF * K 2 pn pn ,min d K 2 pn pn ,min dp * p K1, L F K 2 pn K1, H F * K 2 pn pn ,min * * drn K1, L K pn K 2 F K 2 pn K1, H K K 2 pn pn ,min F K 2 pn pn ,min rn K1, L F * K 2 pn K1, H F * K 2 pn pn ,min K1, H K 2 pn ,min dF * K 2 pn pn ,min d K 2 pn pn ,min drn * rn K1, L F K 2 pn K1, H F * K 2 pn pn ,min Same comments as before can be done on the demand elasticity to population and the symmetrical characteristics of the demand elasticity to price and the one to the normalisation quantity. Assumed the same almost linear shape of the intrinsic elasticity function of the elected population, K1, H K1, L and a small enough value of pn ,min for using 1st order development, it can be demonstrated that: p A F * K 2 pn F * K 2 pn pn,min 1 1 which is positive for F * K 2 pn F * K 2 pn pn,min 1 1 . Once again a leisure market will A progressively evolve from a "normal" price stimulated behaviour towards the Veblen good category. 6. EXAMPLES OF UTILISATION OF THE 2nd KENZA LAW OF DEMAND We are using the Tunisian leisure markets of 5 different European countries as an illustration of the 2nd Kenza law of demand: France, Germany, UK, Italy and Spain. The choice of these markets results from the availability of the Kenza distributions of income for those countries. 6.1 Dataset Visitors, GDP, Population, average passenger income. We have been given the average airline revenue per passenger by Tunisair which are non-public figures which is the reason why the related data are shaded in the table hereafter. © Association for European Transport and contributors 2010 13 Daniel SALLIER © Association for European Transport and contributors 2010 14 Daniel SALLIER Kenza distributions The Coste estimator10 of a Kenza distribution is 1 F * rn min 1, a 1 d 1 ebcrn We have been using the following set of parameters Demand forecast beyond 2009 The forecast is based on an extrapolation of the recent GDP, population and average fares growth over 2000-2007. The purpose is not so much to produce a forecast as to try to identify if the related market has already, is about or will enter is Veblen life cycle. 6.1 French originated leisure market to Tunisia Kenza models are best fitted to historical data with the following set of parameters: © Association for European Transport and contributors 2010 15 Daniel SALLIER Actual traffic figures before 1990 are a bit paradoxical for showing a slightly decreasing trend while at the very same time the French GDP is increasing, average fares are decreasing and the market is still far from entering its Veblen cycle. Close analysis of the French market as modelled by the Kenza law tells us that: 1/ the 2 models are best fitted when considering 1 year shift of the GDP per capita which means customers are sensitive to their former year wealth but to today’s prices. 2/ the 2 models are best fitted considering an equal value of K 1, L and K 1, H . So it is a market which still lacks the ability to secure the loyalty of its “upper class segment”. 3/ the market moved from the 1st to the 2nd version of the 2nd Kenza law of demand between the years 1995 and 2000. According to this model p n ,m in 0.0009 represented a value of €18.03 in 1995 and €28.19 in 2009 (one way). 4/ It is more than likely that the French leisure market to Tunisia will reach is stagnation phase in the incoming years before starting its decline. 6.2 German originated leisure market to Tunisia The German market reflects the turmoil the country went through along its recent history: - 1989: fall of the Berlin Wall en German reunification. Before 1990, visitors, GDP, population and average fares are those of the former West Germany; © Association for European Transport and contributors 2010 16 Daniel SALLIER - 2000 & 2001: Tunisia enjoyed strong arrivals of German tourists further to terrorist attacks in Egypt and Turkey; - 2001 and aftermaths: 9/11 terrorist attack in the USA, war in Afghanistan and Irak. Kenza models are best fitted to the 1987-1994 historical data with the following set of parameters: Both versions of the Kenza model seem to deliver very similar accuracy. Kenza models are best fitted to the 2000-2009 historical data with the following set of parameters: © Association for European Transport and contributors 2010 17 Daniel SALLIER While the lower threshold of income constant, K 2 , L keeps being the same for the 2 periods of time, it is the upper threshold of income constants K 2, H and p n ,m in which change significantly and translate a move for a more “choosy” behaviour with a set of lower and upper threshold constants very close to those of the French market. Together with a more choosy behaviour market penetration of the elected population is almost doubled between the 2 periods of time. As for the French market, both models are fitted considering a 1 year shift of the GDP per capita: 1999 GDP per capita for 2000 demand. A conclusion left to be confirmed is that the German market has already entered its Veblen life cycle. 6.3 Italian originated leisure market to Tunisia Kenza models are best fitted to historical data with the following set of parameters: © Association for European Transport and contributors 2010 18 Daniel SALLIER Close analysis of the Italian market as modelled by the Kenza law tells us that: 1/ the 2 models are best fitted when considering 1 year shift of the GDP per capita which means customers are sensitive to their former year wealth but to today’s prices; 2/ the 2 models are best fitted considering an equal value of K 1, L and K 1, H . So it is a market which still lacks the ability to secure the loyalty of its “upper class customers”; 3/ the market is best modelled with the 1st version of the 2nd Kenza law of demand; 4/ It is more than likely that the Italian leisure market to Tunisia has reached is declining phase and already started behaving like a Veblen good; 5/ While the constant for the lower threshold of income K 2, H is very close to the French and German ones, the upper one K 2, H is far greater 300 instead of 195/200 which can but to mean that the Italian market is less “choosy”. In fact it is the very nature of the market which is different, Tunisia being a long weekend destination for the Italians while it is definitely a vacation destination for French and German people. 6.4 UK and Spanish originated leisure market to Tunisia Both markets prove to be rather badly modelled by a Kenza law, but for 2 different reasons which are: - UK market: It is traditionally a very low yield market mostly served by British charter (low-cost?) airlines which means that Tunisair average passenger © Association for European Transport and contributors 2010 19 Daniel SALLIER income data are not representative of the market and those are the only data we have access to; - Spanish market: it is a rather narrow market which accounted 46,000 visitors in 1987 and about 100,000 visitors today; the sample size to work with is too small for having a chance to fit any accurate Kenza model. 7. CONCLUSION The second Kenza law of demand looks like delivering what it has been designed for: - an ability to deliver a better understanding of the structural forces at work in any leisure market which allows to differentiate between conjectural effects and structural ones: Is my record figure of 1 million German visitors in 2000 and 2001 something likely to repeat? Is it worth spending millions of Euros of advertisement in Germany to drain back German visitors? Have we to strongly suggest our flag carrier – Tunisair – to have a more aggressive fare policy on the German market? Is the German market still profitable? Isn’t it better to reallocate our marketing expenses and efforts on blossoming East European markets such as the Polish or the Russian ones? Is my continuously declining Italian market since 2005 despite 5% fare decrease over the period a consequence of tougher competition from new destinations such as Croatia? It is a question which leads to exactly the same set of additional questions as the ones of the German market. It looks like that my French market, my bread and butter, is likely to start being rather stagnant in the next future whatever the aggressiveness of my fare policy. It looks like it has to do with the inability to keep the upper segment of this market. Is there any way to revert this process? - A rather robust model which is not that sensitive to the value of its parameters and of which parameters have a physical meaning which make them comparable between different markets; - An unexpected output of the second Kenza law of demand is that it tell us that any leisure market will potentially turn into a Veblen good at some time of its product life cycle: the higher the price, the higher the demand. The tourism sector and its actors may find in the luxury industry – a typical sector of Veblen goods – the keys for keeping their business profitable; - 200,000 visitors, 400,000 passengers a year seem to be a very minimum demand/traffic volume below which it is pretty hard to fit any Kenza model; © Association for European Transport and contributors 2010 20 Daniel SALLIER As for the 1st Kenza law of demand, the very "disturbing" issue with this approach is the very demanding assumption made on K 1s and K2s of being constant over the time. This is the reason why we have conducted additional research works, yet to be published, which seem to prove the case; Ks property would result from the aggregation to all a population of a probabilistic model of individual consumption behaviour. To finish with, we have developed and used the second Kenza law of demand for modelling leisure air traffic and tourist visitors demand, but this approach far exceed the only topic of leisure air transportation and can address all sort of consumer demand issues. 8. REFERENCES Sallier, D. (2010) Long-term demand forecast: the Kenza approach, ATRS 2010, Porto Veblen, T. (1898), The Theory of Leisure Class, Prometheus Books (New York) French Ministry of Economy, Industry and Employment (2008) Les Comptes du Tourisme – Compte 2007, Ministère de l'Economie, de l'Industrie et de l'Emploi, Paris Le Garrec, M.-A. (2008), Le Tourisme : Un Secteur Economique Porteur, Ministère de l'Economie, de l'Industrie et de l'Emploi – Direction du Tourisme, Paris Instituto Nacional de Estadística (2010), Cuenta satélite del turismo de España. Base 2000. Serie contable 2000-2008, Madrid Griffith E., Zemanek S. (2009), U.S. Travel and Tourism Satellite Accounts for 2005–2008, US Bureau of Economic Analysis, Washington © Association for European Transport and contributors 2010 21 Daniel SALLIER NOTES 1 2 3 4 5 6 7 8 9 10 Sallier D. (2010) Long-term demand forecast: the Kenza approach, ATRS 2010, Porto Source Tunisair French Ministry of Economy, Industry and Employment, Direction du Tourisme (2008) Les Comptes du Tourisme – Compte 2007, Ministère de l'Economie, de l'Industrie et de l'Emploi, Paris Le Garrec, M.-A. (2008) Le Tourisme : Un Secteur Economique Porteur, Ministère de l'Economie, de l'Industrie et de l'Emploi – Direction du Tourisme, Paris Instituto Nacional de Estadística (2010), Cuenta satélite del turismo de España. Base 2000. Serie contable 2000-2008, Madrid Hellenic Statistical Authority (2010) Athens S. Griffith E., Zemanek S. (2009) U.S. Travel and Tourism Satellite Accounts for 2005–2008, US Bureau of Economic Analysis, Washington This behaviour is at the origin of what it's called the Veblen good category which was initially identified and described by the economist Thorstein Bunde Veblen (1857-1929): Veblen T. (1898), "The Theory of Leisure Class", Prometheus Books (New York) Unpublished work based on the study of the Kenza distributions of Brazil, Canada, Denmark, France, Germany, India, Italy, Japan, Mexico, San Salvador, Spain, South Africa, UK, USA. Unpublished work based on the study of the Kenza distributions. The Coste estimator is 1 order of magnitude more precise that the classical log-normal estimator which proves to be significantly more precise than the Pareto law. © Association for European Transport and contributors 2010 22
© Copyright 2026 Paperzz