Competitiveness of the Dutch Automobile Lease Market

(1)

Competitiveness of the Dutch Automobile Lease Market

An Analysis of the IRRs of Three Leasing Companies

University of Groningen

Faculty of Economics and Business

MSc Business Administration specialization Finance

Amsterdam, September 2008

Bart van den Akker 1333283

Supervisors: dr. ing. N. Brunia

(2)

Preface

This master thesis forms the final part of my MSc Finance programme at the University of Groningen. I started working on my thesis during my internship at Duff & Phelps in Amsterdam. I chose to combine writing a thesis and doing an internship to do research that would have at lease some practical relevance. I thought had found a perfect opportunity to do so at Duff & Phelps, which provided me with a research topic. Towards the end of my

internship, and after extensive discussions with my supervisor, we decided that the topic I was researching would not lead to a acceptable academic thesis in that form. The topic was

restructured in such a way that proper data-analysis could be done, which, as I have learnt the hard way, is essential when writing a thesis in finance.

The process of writing the thesis has, therefore, taken somewhat longer than I had originally expected, but I do feel that I have managed to learn a thing or two along the way. I would like to take the opportunity to thank a number of people for making this outcome possible. First of all, my supervisor, Nanne Brunia, who provided me with critical, but always useful and to-the-point comment on my work. I would also like to thank my colleagues at Duff & Phelps for making my internship truly enjoyable and giving me the opportunity to use my Corporate Valuation knowledge in practice. Lastly, I would like to thank my parents for their support throughout my entire studies.

(3)

Abstract

(4)

1. Introduction

Under current IFRS rulings, a company must perform a purchase price allocation when acquiring another company. Such a valuation allocates the purchase price across tangible assets, identifiable intangible assets and goodwill. When valuing the intangible assets (e.g. an R&D project) a ‘contributory asset charge’ is determined. The principle behind this is that the project ‘rents’ the assets required to produce its cash flows from a hypothetical third party at a fair price. The net cash flows remaining after these charges are the attributable to the project and can be used to determine the value of the project (AICPA, 2001).

The American Institute of CPA’s suggests that for fixed assets, the charge rate could be determined based on rates implied by operating leases for such assets (AICPA, 2001). This implicitly assumes that the rates charged by lessors are fair returns on the assets and,

therefore, the operating lease market is perfectly competitive. In this research I investigate the competitiveness of the Dutch automobile operating lease market. The reason for choosing the automobile lease market is that the availability of data for this segment is relatively high compared to other segments of the Dutch lease market.

Competitive markets are considered to possess three distinct characteristics: (1) prices equal marginal costs, (2) economic profits are zero or negative, (3) absence of inefficiency in production (Baumol, 1982). In this thesis I investigate the competitiveness of the Dutch automobile operating lease market by examining whether the first two characteristics are present. I am unable to examine the presence of the third characteristic due to a lack of available data. The dataset used consists of operating lease prices for 444 individual automobile types from three major Dutch leasing companies. Combined with a number of sources used to estimate the costs incurred, this allows for an estimation of revenues and both marginal and average costs for individual leases. I calculate the marginal IRRs on individual leases to assess whether prices equal marginal costs and the average IRRs to assess whether economic profits are zero.

(6)

lease1 market and found that, over the five-year period investigated, increased competition led to a decrease in economic profits generated by the lessors.

The main question I answer in this thesis is:

Do competitive conditions exist within the Dutch automobile lease market?

In order to answer this question I answer the following sub-questions:

1. Do prices equal marginal costs in the Dutch automobile lease market?

a. What are the marginal IRRs generated by Dutch automobile leasing companies on individual leases?

b. Are the marginal IRRs generated equal to the opportunity cost of capital, indicating that prices equal marginal costs?

c. Do the marginal IRRs calculated vary across leasing companies and across market segments?

2. Do Dutch automobile leasing companies generate zero or negative economic profit? a. What are the average IRRs generated by Dutch automobile leasing companies? b. Are the average IRRs generated equal to the opportunity cost of capital,

indicating that economic profits are zero?

This thesis contributes to existing finance practice by examining whether the assumptions underlying the AICPA guideline (see above) regarding the competitiveness of the lease market are correct. Furthermore, it contributes to the literature on competitiveness by applying methods that have not yet been used in practice to assess the competitiveness of a market.

This thesis is organized as follows. The next section contains the literature review. The third section discusses the methodology and data used. Section four discusses the empirical results. The fifth, and last section, presents the conclusion of this research.

1

(7)

2. Literature Review

This literature review is split into two subsections. The first discusses the most commonly used concepts of market structure and competition. The ways the competitiveness of a market can be assessed is discussed in the second subsection.

2.1 Competition

This subsection discusses the basic concepts of competition and market structures. The two most extreme forms of market structure, perfect competition and monopoly, are discussed first, followed by monopolistic competition and oligopoly. The concept of contestable markets is also discussed in this subsection. The purpose of this subsection is merely to provide a framework for the discussion on measures of competitiveness. I refer to Besanko, Dranove and Shanley (2000) for a more elaborate discussion on competition and market structures.

Perfectly competitive markets consists of many sellers of homogeneous goods and many, well-informed consumers who can costlessly shop around for the best price (Besanko et al., 2000). Perfectly competitive markets have three characteristics relating to the behaviour of firms competing in such markets. The first characteristic is that prices equal the marginal costs of production. In perfectly competitive markets all products are sold at a single market price. A firm trying to sell its products at a price higher than market price sells nothing, since the customer can obtain an identical product from a competitor at a lower price. A firm trying to sell its products at a price lower than the market price needlessly sacrifices revenue. Given that a firm maximizes its profit by producing a volume such that price equals marginal costs, the market price will be driven down to the level of the marginal costs of production in a competitive equilibrium (Besanko et al., 2000). The second characteristic is that firms generate zero or negative economic profit in perfectly competitive markets. Any incumbent generating economic profit will lead to a new entrant replicating the incumbent’s business, slightly undercutting its prices and still making a profit. This is feasible since perfectly

(8)

A monopoly is the form of market structure that is most distant from perfect competition; the position of a monopolist firm is not challenged by actions of competing firms. A monopolist can, therefore, raise prices or reduce quality without facing the risk of customer switching to a competitor. This does not mean that a monopolist is not faced with elasticity of demand; a change in price will affect the monopolist’s revenue. Equal to a firm facing perfect

competition, a monopolist sets its price such that its marginal revenue equals marginal costs, thus maximizing its profit. However, given the lack of competitive pressure, the price is well above the marginal cost level and the output volume is well below the competitive level (Besanko et al, 2000). A monopolist, therefore, is able to (1) charge its customers a price above the level of marginal costs, (2) generate an above zero economic profit and (3) operate with production inefficiency.

The term monopolistic competition is used to distinguish markets that differ from perfectly competitive markets in one specific way: each firm sells a (slightly) differentiated product rather than identical products. This means that customers make choices between different sellers based on aspects other than just price. Differentiation can be vertical (one product undoubtedly better than the other) or horizontal (one product appeals to different preferences than the other). Therefore, contrary to seller operating in a perfectly competitive market, a seller will not lose all its customers when raising prices. Since, contrary to under perfect competition, demand elasticity is thus not infinite, sellers are able to maintain a price higher than the marginal costs of production. As in perfectly competitive markets, economic profits attract new entrants. This erodes economic profits since the incumbent’s market share erodes and / or the increase in competitive pressure leads to lower prices. In equilibrium, therefore, economic profits will be zero (Besanko et al., 2000). Analogue to the existence of economic profit, the existence of production inefficiency constitutes an opportunity for profitable entry. Therefore, inefficiency in production cannot exist in equilibrium.

Oligopolies, contrary to perfectly competitive and monopolistically competitive markets, are characterized by the presence of only a few sellers. Therefore, it is expected that the actions of individual firms do affect the choices made by its competitors. The two most commonly used models of oligopolies are Cournot quantity competition and Bertrand price competition. Both models make different assumptions about the way firms respond to their competitors’

(9)

the price such that it will sell its production volume, in the Betrand model a firm chooses the price at which to sell, and then produces to match the resulting demand (Besanko et al., 2000).

In the Cournot model, all firms produce identical goods, forcing these firms to charge equal prices. Therefore, the only choice a firm can make is the amount it produces. Each firm sets its price such that it will sell all its output, resulting in the market price being such that the total volume produced by all firms is sold. Thus, if any firm decides to increase its

production, the market price will fall. Therefore, each firm sets its production level such that it will maximize its profits based on its expectation of the production levels of all competitors combined. Since a firm is only interested in maximizing its own profit, it may increase its production level to do so, thereby reducing the market price and reducing overall industry profit. This is called the revenue destruction effect. As a result, the market price in a Cournot equilibrium is lower that it would be in market served by a monopolist. The divergence between the private gain and the revenue destroyed increases as the market share of an

individual firm drops. Therefore, as the number of firms increases, the market price will move towards the marginal costs of production and the average economic profit per firm approaches zero. However, since these levels are only reached with an infinitely large number of sellers and the number of sellers in an oligopoly is, by definition, limited, these levels will not be reached if the oligopoly is maintained. Prices above marginal costs and positive economic profit can exist because all firms know that they have nothing to gain from lowering prices since the competition will immediately match a price cut in order to continue to sell its production volume (Besanko et al., 2000).

In the Bertrand model, each firm sets a price in order to maximize its own profits, given the prices it expects its competitors to set. Each firm believes its price setting will not affect the pricing of its competitors. When the products produced by each firm are identical, all

customers will purchase the product from the firm with the lowest price2. Therefore, each firm will always undercut its competitors’ price until it reaches the level of marginal costs. In the Bertrand model, therefore, only two firms are required to produce the result of perfectly competitive markets; price equals marginal costs. Since each firm believes it can steal business by slightly undercutting its competitor’s price, and will continue to do so until no

2

(10)

economic profit remains. In equilibrium, therefore, economic profits equal zero (Besanko et al., 2000).

The concept of ‘perfectly contestable markets’ introduced by Baumol (1982) forms an extension to the basic concepts of market structures. It shows that firms will behave in a welfare maximizing way (equal to under perfect competition) even in markets that, by structure, would not seem to be competitive. This is caused by the existence of, or even the threat of, new entry (Ashton, 2000). The characteristics of contestable markets, therefore, apply to a much wider range of industry structures than just perfect competition. This makes perfect contestability, more so than perfect competitiveness, a widely applicable benchmark for welfare-maximizing behaviour.

Contrary to under perfect competitiveness, the presence (or potential presence) of two competing firms in a market can be enough to guarantee welfare maximizing behaviour. A contestable market is characterised by freedom of entry and absolutely costless exit. Freedom of entry means that entrants do not suffer cost disadvantages relative to incumbents. Freedom of exit implies that a firm leaving the market can do so recouping any costs incurred in the entry process (other than normal cost of use and depreciation of assets). Freedom of entry and costless exit ensure that a new entrant can instantaneously take advantage of an opportunity for profitable entry and then depart from the market without further cost, this is referred to as a hit-and-run entry (Baumol, 1982).

(11)

perfectly competitive (Baumol, 1982). Thus, in general, a market can be considered competitive when firms behave in such a way that the three characteristics of welfare maximizing behaviour are present. In this thesis I focus on the first two characteristics only since the available data does not allow for investigation of the presence of production inefficiency.

2.2 Measures of Competitiveness

A number of different measures have been used in prior research to assess the

(12)

Table I

Methods for assessing competitiveness of a market

Method Characteristic /

Concept

Data required Conclusion w.r.t.

Competitiveness Percentage contribution margin (PCM) Price equals marginal cost

Price, marginal costs Any value above zero indicate non-competitiveness

Lerner index Zero economic profit

Economic profit, Sales All values lower than 1 indicate non-competitiveness

ARR Zero economic profit

EBIT, Capital Invested Return above cost of capital indicates some

non-competitiveness IRR Zero economic

profit

Investment’s cash flows Return above cost of capital indicates some

non-competitiveness RoS Zero economic

profit

EBIT, Sales Return above benchmark value indicates some

non-competitiveness NPV Zero economic

profit

Investment’s cash flows NPV above zero indicates some non-competitiveness

Cost level comparison Absence of inefficiency

Time series of total cost data for firms in an industry

Presence of unused scale economies indicates non-competitiveness

N-firm concentration Concentrated market is less competitive

Market shares of n largest firms

High concentration indicates non-competitiveness

Herfindahl (HHI) Concentrated market is less competitive

Market shares of all firms with a share larger than 1%

Arbitrary ranges of HHI values

Marginal cost based measures

(13)

Economic profit based measures

A large number of commonly used measures of competitiveness are based on the

characteristic of perfect contestability that economic profit must be zero. The main difference between these methods and the marginal costs based measure discussed above is that these measures are based on average rather than marginal costs. The Lerner index (li), or price cost margin, is a well known measure in this category, it is defined as economic profit divided by sales. To assess the competitiveness of a market one calculates the weighted (by sales) average of the Lerner indices across the firms within an industry and deducts this figure from 1. Perfect competition is thus indicated by the value of one since the average economic profit is then zero. A value below one indicates some degree of non-competitiveness (Aghion et al., 2005). Examples of application of the Lerner index in empirical research are widespread, two can be found in Agnelini and Cetorelli (2003) and Parker and Kim (1997)

Boone, Van Ours and Van der Wiel (2007) argue that an increase in competition can lead to a higher weighted average li (usually associated with a decrease in competitiveness) because of an increase in market share of efficient firms (with higher li’s). Another issue is that as firm’s costs fall over time because of an increase in efficiency, the li will rise. According to Boone et al. (2007) this should not be interpreted as a fall in competition, but rather as an increase in efficiency. They state that conditional of a firm’s costs, a high li indicates market power, but conditional on price, a high li indicates efficiency. Following Baumol (1982), I argue that these deviations from the characteristics of behaviour in competitive markets are not

sustainable in equilibrium. After all, even the presence of just two firms should be enough to ensure welfare maximizing behaviour. Therefore, the criticism may apply when assessing changes in the competitive situation, but not when assessing whether competitive conditions are present in a market.

Below I discuss a number of measures that use a rate of return figure to assess the competitiveness of a market. The rates of return calculated need to be compared to a

benchmark figure (the required rate of return) to assess whether a company is generating an above zero economic profit.

(14)

deprival value), which may lead to significant biases on the return figure (Grout and

Zalewska, 2006). Conceptually, of course, there is a direct relationship between the ARR, the cost of capital and net present value. In practice, however, the ARR is hardly ever calculated in such a way to ensure this relationship (Grout and Zalewska, 2006). Kay (1976) shows that the average ARR, when measured over a period of years, closely matches the IRR. This is, however, not the case when the ARR is measured over a single year, since the ARR varies over time. This makes it more difficult to compare the ARR to an objective benchmark figure, especially when it is calculated over a shorter period of time, making it less suitable for use in the assessment of competitiveness. While I have not come across any applications of the ARR in competition analyses, Feenstra, Huijgen and Wang (2000) do point out that high ARRs tend to fall over time as a result of high profitability attracting competition (and vice-versa for low ARRs).

The IRR is directly linked to the concepts of NPV and cost of capital since it is defined as the discount rate that gives a net present value of zero when applied to a series of cash flows. It can therefore be directly compared to the investor’s required rate of return to determine whether economic profit is being generated. When used in capital budgeting, the IRR has two specific flaws. Firstly, it does not reflect the differences in scale between mutually exclusive investment opportunities. Secondly, an investment may have multiple IRRs when consisting of both inflows and outflows after the initial investment (Koller, Goedhart and Wessels, 2005). The issue of scale is irrelevant in this context since I am not faced with the question of choosing between different projects but comparing individual returns to a benchmark. The issue of multiple IRRs does not apply to automobile leases, since, as I show later, the only net cash outflow is the initial purchase of the car and all subsequent periods show positive net cash flows.

(15)

The net present value is the present value of the cash flows an investment generates minus the cost of the investment. By definition, the NPV is equal to the present value of the economic profits generated by an investment. Therefore, a firm with zero economic profit (welfare maximizing behaviour) also generates a NPV of zero in its investments. While the

measurement of the NPV is very useful from the viewpoint of an investor, who is interested in total wealth gain, it is less useful in assessing competitiveness of a market since it provides an absolute rather than a percentage return. Therefore, the only statement one can make based on the observation of an above zero NPV is that above zero economic profits are being

generated. It is, however, more difficult to compare these economic profits across companies since one should take differences in scale into account (Grout and Zalewska, 2006).

Efficiency based measures

A measure based on the efficiency of a firm by comparison of cost levels is applied by Bikker and Gorter (2008) in their analysis of the competition in the Dutch non-life insurance market. By regressing the total costs of individual insurance firms against a number of variables they are able to estimate the level of economies of scale of different firms. Their measure is based on the concept that strong competition would force firms to exploit any unused scale

economies since no inefficiency may exist under competitive conditions.

Other methods

(16)

Both measures are based on the concept that the larger the number of firms, the more the situation of perfect competitiveness (and, thus, welfare maximizing behaviour) is approached. This concept, however, has three significant flaws, the first, and most obvious one being that a large number of firms is not required to achieve welfare maximizing behaviour. As

(17)

3. Data and Methodology

The previous section has discussed the main measures available for assessing the strength of competitive pressure facing a firm. This section discusses the methodology used in this research. The basic equation used to perform the calculations on the lease data is discussed in the second subsection. The raw data available for the analysis of the Dutch automobile lease market and the samples selected are discussed in the third subsection. The fourth subsection presents the models used to construct the necessary inputs for the IRR calculation and the selection of the samples. Subsections five and six present the calculations of the WACC and credit loss respectively.

3.1 Methodology

Most competitiveness measures discussed in the previous section require either aggregate financial data of the companies in the industry under investigation or price and cost data on individual products. Since most Dutch automobile leasing companies are part of larger financial institutions, aggregate financial data for the leasing business is generally not available. Prices, however, are available and, as I show below, the cost elements can be estimated based on other data sources. Therefore, a measure based on the prices and associated (marginal) costs of individual products must be used. One could thus imagine using the percentage contribution margin (PCM) method to assess the competitiveness of the lease market. There is, however, a drawback of using this method. Since the only benchmark available for the PCM is the value of zero (perfectly competitive or contestable) and it is not a rate of return figure, it is impossible to make a statement about the competitive pressure in a market when a value other than zero is found. Based on PCM it is only possible to assess whether or not a market is perfectly competitive.

The IRR is better suited to handle the specifics of the available data. Rather than just

calculating the average IRR to determine the presence of above zero economic profits, I focus mainly on calculating the incremental IRR of a single lease. Based on these calculations, it is possible to determine whether lease prices are equal to marginal costs (including the

(18)

Given the lack of relevant data, it is impossible to make an estimate of any inefficiency in the production. Therefore, I do not further investigate this characteristic of welfare maximizing behaviour in this thesis.

The expected marginal IRRs on leases issued by Dutch leasing companies can be used in a number of ways in order to assess the competitiveness of the Dutch automobile lease market. The first method involves comparing the average IRRs of the different leasing companies to a benchmark rate of return. The first benchmark to be used is the opportunity cost of capital3 (WACC) of European automobile leasing companies, adjusted for the duration of a lease, which is shorter than the time-horizon of a stock investor. This represents the return required by an investor investing in a lease. Should the marginal IRR of one or more of the leasing companies lie significantly above the WACC, this is a sign that leases are priced above their marginal cost level. The same holds for the average IRR with respect to the economic profits being generated. The second benchmark is based on the issue raised by Grout and Zalewska (2008) that a margin above the cost of capital is required before a rate of return is considered excessive. Following the British Competition Commission (Grout and Zalewska, 2008), I set this margin at 20 percentage points. I use single sample t-tests to investigate whether the group of calculated IRRs differs significantly from both benchmarks. These tests are carried out using Eviews.

The second method consists of comparing the marginal IRRs across the different leasing companies. A lack of competitive pressure could result in significant differences in returns generated by Dutch automobile leasing companies. After all, in a competitive market, one firm would not be able to charge a significantly higher price than any of its competitors for an identical product. Two tests are performed, the first to investigate whether the average

marginal IRR of all leasing companies are equal. I use the ANOVA test for equality of means for this purpose. The second test investigates differences in mean returns between pairs of companies; I use independent samples t-test here. The tests are carried out using Eviews. This approach can, of course, only be used to find differences in market power between companies. Since it does not use an objective benchmark, statements about the absolute market power of the leasing firms cannot be made.

3

(19)

3.2 Lease IRR Model

Equation (1) shows the formula for calculating the present value of an automobile lease to the lessor. It is based on the cost components as presented in Smith and Wakeman (1985).

∑

= = + + + + + − − − + − = T t t t T T t t t t t r DEP r SALV r MVTINS MAINT LEASE INV PVCONT 1 1 (1 ) ) ( ) 1 ( ) 1 ( ) ( ) 1 ( τ τ (1)

PVCONT = Present value of a lease contract INV = Purchase price of the car

τ = Corporate tax rate

LEASEt = Monthly lease payment MAINTt = Monthly maintenance costs

MVTINSt = Monthly motor vehicle tax and insurance costs SALV = Re-sale value of the car after the duration of the lease DEPt = Monthly depreciation of the car

T = Duration of the lease (in months)

r = Discount rate

(20)

As I explain in subsection 3.3 on raw data, motor vehicle tax (MVT) and insurance costs (INS) are combined into one component (MVTINS) in order to make use of the available data. Since I am investigating marginal costs of a lease contract, I assume overhead costs to be zero. The last adjustments are related to inflation. Maintenance costs are incurred by the lessor and increases in these costs cannot be passed on to the lessee. Therefore, maintenance costs (MAINT) are assumed to increase by the expected inflation (see below) on a monthly basis. The lessor is contractually allowed to pass on increases in insurance costs and taxes to the lessee, therefore any expected increases in these costs are not included in the IRR

calculation. Once the lease has been entered into, the lessor cannot adjust the lease price to cover any cost increases other than these.

The IRR of the lease is calculated by substituting the IRR for the discount rate such that Equation (1) equals zero. This substitution is shown in Equation (2).

0 ) 1 ( ) ( ) 1 ( ) 1 ( ) ( ) 1 ( 1 1 = + + + + + − − − + −

∑

= = T t t t T T t t t t t IRR DEP IRR SALV IRR MVTINS MAINT LEASE INV τ τ ₍₂₎

3.3 Raw Data and Sample

In order to evaluate all revenues and costs associated with a single automobile lease, all the components shown in Equation (2) must be collected or estimated. This subsection discusses the raw data and samples used to estimate these costs. A brief overview of the data acquired, the units of measurement and the sources is presented in Table II.

Table II

Overview of Data Sources and Units

Data Unit Source

Lease price € / month Leasing companies Retail price € ‘000 Carbase

(21)

Automobiles

Since this research involves investigating individual lease contracts issued by a number of Dutch leasing companies the population consists of all makes and models cars for which a lease price is available from these leasing companies. In order to avoid confusion surrounding the terminology used, a brief overview is presented below:

Make – The brand of the car (e.g. Volkswagen)

Model – The distinguishable product (e.g. Passat station wagon)4 Type – The individual specification within the model range

(e.g. Passat station wagon 1.6 Easyline)

In order to select a representative sample from the population a number of choices need to be made. Given the wide variety of makes and models available in the Dutch market I choose to include at least one type of each model in the analysis. Including all available models rather than a larger number of types but a smaller variety in models allows for segmentation of the sample into different ‘categories’ of cars (e.g. geographical origin, market segment). Given that most models are available in both petrol and diesel powered types and that one could expect differences in costs associated with operating differently powered types, I include both types in the sample. This results in a sample of 444 individual types (244 petrol, 200 diesel).

All leasing companies consistently include the cheapest available type of each model and fuel type in their on-line price overviews. Not all leasing companies include all the other types within a model range in their price overviews, nor is the selection of types the same across the various leasing companies. Therefore, I include the cheapest available type (both petrol and diesel powered) in the sample. There is, of course a risk that required returns on these types are different that those of more expensive types within model ranges. However, given the large variety of models included in the sample, I expect larger differences to exist across the various models than within model ranges.

The retail price, horsepower, weight and fuel consumption of the cars are obtained from Autoweek / Carbase5, which publishes the information provided by the car manufacturers.

4

Note that I distinguish between e.g. the ‘Passat sedan’ and the ‘Passat station wagon’ as being different models.

5

(22)

The data was collected between April 22nd and 25th. Descriptive statistics of the automobiles included in the sample are presented in Table III.

Table III

Descriptive Statistics of Sample of Automobiles

Sample Variable Average Median

Standard

Deviation Minimum Maximum

Total sample Price 30453 28348 13966 7495 77650 Horsepower 92 87 32 38 213 Weight 1379 1382 281 740 2394 Fuel consumption 7.0 6.7 1.8 3.8 16.7 Petrol Price 27265 24600 14117 7495 71995 Horsepower 91 81 36 38 213 Weight 1304 1293 273 740 2195 Fuel consumption 7.7 7.3 1.8 4.4 16.7 Diesel Price 34343 31660 12775 14995 77650 Horsepower 93 91 26 48 173 Weight 1471 1467 263 910 2394 Fuel consumption 6.1 5.9 1.3 3.8 10.8 Lease prices

The contractual lease prices for each car included in the sample are obtained from three different leasing companies: Leaseplan6, ING Car Lease (Toplease)7 and Kroymans (Directlease)8. These companies are chosen selected because they are all established lease vendors in the Dutch market and make their lease prices available on-line. The companies hold positions 1 (17%), 2 (11.2%) and 8 (unknown) respectively (market share in brackets) (VNA, 2008; Leaseplan website; Toplease website). Lease prices for all 444 car types are obtained from all three leasing companies. The lease prices (and other data on the cars, see above) were collected between April 28th and May 2nd, 2008. Descriptive statistics of the lease prices included in the sample are presented in Table IV.

6

Leaseplan Direct (www.leaseplandirect.nl)

7

ING Car Lease (Toplease) (www.toplease.nl)

8

(23)

Table IV

Descriptive Statistics of Sample of Lease Prices

Sample Company Average Median

Standard

Deviation Minimum Maximum

Total sample Leaseplan 663 616 286 174 1704 Toplease 650 609 284 174 1620 Directlease 659 617 289 195 1633 Petrol Leaseplan 554 512 260 174 1471 Toplease 540 490 258 174 1415 Directlease 550 495 264 195 1418 Diesel Leaseplan 796 764 257 346 1704 Toplease 784 749 256 375 1620 Directlease 791 740 263 392 1633

The average contractual duration of an automobile operating lease in The Netherlands is 44.6 months (VNA, 2008). Given that individual lease contracts generally have a duration of a multiple of 6 months, I collect the lease prices for leases with a duration of 48 months. This, of course, does limit the scope of this research to the market for 48-month leases. Given that all leasing companies offer leases of various durations, I do not see why the findings of this research in terms of the competitiveness of the market should not apply to the market for automobile leases of other durations as well. Another reason for choosing this lease duration is that the data source I use for estimating the maintenance expenses, insurance costs and motor vehicle tax (see below) contains estimates based on a 48 month operating period. In order to ensure consistency with these estimates the contractually agreed upon mileages for the lease prices I collect are 15,000 km / yr for petrol cars and 30,000 km / yr for diesel cars.

In order to ensure that I do compare returns on homogenous products, it is necessary to ensure that the contents of the lease contract are consistent across the leasing companies. Therefore, I select the following contractual agreements.

- The lessor can order the car from any dealership it selects. Leasing companies usually have specific agreements with a few dealerships of each make to obtain additional discounts.

- The lessor does not provide a set of winter tyres.

(24)

- In case a temporary replacement vehicle is required (e.g. due to damage or extended maintenance), a small car is provided only after 24 hours.

Leasing companies state lease prices including and excluding the expected fuel costs. In practice, the lessee pays the leasing company an advance for the expected fuel costs and the fuel bills are paid directly by the leasing company. The balance of the advances and the actual costs is settled periodically. Since in the end the lessee always pays the actual fuel costs, the prices excluding the advance for fuel expenses are used.

Maintenance costs

The ANWB (the Dutch motorists’ association) publishes an overview of the expected costs a car owner incurs for owning and driving a specific car. This list contains a total of 591

different types of cars (365 petrol, 226 diesel). The cost estimates are monthly expenses based on the assumptions of purchasing a new car, selling it after a period of 4 years and driving 15,000 km (petrol) or 30,000 km (diesel) per year. One of the costs components included in the ANWB analysis is ‘repair, maintenance and tyres’. These ANWB estimates are used to estimate the maintenance and repair costs incurred by lessors. Two issues arise when using this data. The first is that, in general, lease car drivers are known to be less careful with their car than people driving their own cars. This could lead to an underestimation of the

maintenance and repair costs since lessors may incur higher costs than private owners. The second issue is that lessors may be able to negotiate lower maintenance prices with repair shops than consumers. This could lead to an overestimation of the maintenance costs. For this research, I assume that these two effects offset one another. I thus assume that the ANWB estimates are a good proxy for the actual maintenance costs incurred by lessors. Additionally, these are simply the only reliable estimates available. Since the ANWB estimates are based on the price levels as of March 1st, 2006, the costs are adjusted for inflation based on the inflation rates obtained from the Dutch Central Statistics Bureau9 (4.7% increase).

The sample of cars included in the ANWB sample obviously differs from the sample for which lease prices are obtained. The sample was collected on a different date, which means than some models have been discontinued and others have been introduced since then.

9

(25)

Therefore, regression analysis on the data is performed to create a model to estimate the maintenance expenses for the cars included in the lease price sample.

The ANWB sample includes also high-performance sports cars and other exclusive models which are, in general, not offered by leasing companies. In order to obtain a reliable estimate of the maintenance costs, these need to be removed from the sample. This is done by

removing all types that exceed the maximum values of the lease car sample (see above) in terms of price, horse power or weight from the sample. Considering the differences in values of these variables between petrol and diesel powered cars (see Table III), I do distinguish between the two fuel types. This results in a sample of 525 individual types (313 petrol, 212 diesel). Descriptive statistics for this data is shown in Table V, the motor vehicle tax and insurance costs (discussed below) are also included since the same sample is used for both cost components.

Table V

Descriptive Statistics of Sample of Maintenance Costs, Taxes and Insurance

Sample Variable Average Median

Standard

Total Sample Price 31774 28860 14981 6999 75800 Horsepower 97 92 37 30 213 Weight 1374 1350 311 720 2315 Maintenance 82 61 44 29 241 Tax and Insurance 225 212 87 67 506

Petrol Price 29531 26250 15215 6999 69953 Horsepower 100 92 40 37 213 Weight 1313 1293 298 720 2170 Maintenance 50 48 13 29 111 Tax and Insurance 179 175 60 67 349

Diesel Price 35085 33148 14021 14095 75800 Horsepower 93 90 30 30 171 Weight 1464 1448 308 730 2315 Maintenance 129 127 27 75 241 Tax and Insurance 291 286 78 127 506

(26)

Motor vehicle tax and insurance costs

Another cost components incurred by the lessor consists of motor vehicle tax (NL: MRB – motorrijtuigenbelasting) and insurance costs. These are included in the component ‘fixed costs’ in the ANWB estimates and are therefore combined into one component10.

There is one issue with using this dataset that may influence the results of the analysis. Some leasing companies decide to carry the insurance risk themselves, others insure the cars with an external insurer. In either case, one might expect that the actual costs to the leasing companies differ from the insurance premium paid by a consumer. This depends on the competitiveness of the automobile insurance market and the bargaining power of leasing companies over insurance companies. While prior research does not prove that the Dutch automobile insurance market is perfectly competitive, it does indicate that some degree of competitive pressure exists (Bikker and Gorter, 2008; Hardwick and Dou, 1998). Since all three firms are large, established leasing companies, I do not expect that they have significantly different bargaining positions over insurance companies. Based on the above, I assume that the ANWB cost estimates can be used as a reliable proxy for the actual costs incurred by lessor. The adjustments in terms of sample selection and inflation described above are, of course, also applied here.

Expected Salvage Values

Since the lease contracts used in the sample have a duration of 48 months, the residual value at the end of a lease contract is the market value of a 48 month old car. Therefore, the economic depreciation of automobiles over a 48 month period is of interest. I calculate the economic depreciation based on current second hand prices for cars that were sold new 48 month ago. A dataset of average market prices for all cars, specific for age and total mileage is obtained from Autoweek / Carbase. This is generally considered a very reliable source for second hand prices. As with the data sources mentioned above, it is impossible to include every type included in the lease price sample in this sample too, since the available model ranges of all makes change over time. Therefore, the sample consists of, again, the cheapest available type of each make and model as offered 48 months ago. As above, I exclude high-performance, exclusive cars. I collect a sample of current second hand prices for 48 month old

10

(27)

cars with mileages of 60,000 km or 120,000 km for petrol and diesel respectively. The database quotes two prices for each car, one is the price a consumer should expect to pay for such a car at a dealer, and the other is the price a consumer should expect to receive when selling such a car to a dealer. For my analysis, I use the average of these prices since automobile leasing companies are known to be able to sell their cars a higher price than consumers, but I do not expect them to receive the price paid by a consumer to a dealer. I perform a regression analysis on this sample to estimate the residual value of cars currently leased. The sample consists of a total of 417 individual car types (241 petrol, 176 diesel). The data was collected between May 5th and May 7th, 2008. Table VI presents the descriptive statistics for this sample.

Table VI

Descriptive Statistics of Sample of Depreciation

Sample Variable Average Median

Standard

Total sample List price (new) 28415 26100 12620 7895 80300 Second-hand market value 14187 13000 6635 4200 47150 Depreciation (% total) 50 49 7 33 71

Petrol List price (new) 26661 23950 13452 7895 80300 Second-hand market value 13968 12750 7331 4200 47150 Depreciation (% total) 47 46 7 33 69

Diesel List price (new) 30818 28893 10972 13995 63900 Second-hand market value 14488 13475 5549 6150 32350 Depreciation (% total) 53 52 6 41 71

3.4 Data Construction and Adjustment

(28)

Since lessors can reclaim VAT (NL: BTW), this needs to be deducted from a number of cost components to arrive at the actual costs incurred by lessors:

- The purchase prices of cars are stated including VAT, the appropriate correction is made to arrive at the initial investments. This same goes for the second-hand prices of cars.

- Lease prices are always stated excluding VAT, therefore no adjustment is required. - Maintenance costs as stated in the ANWB overview do include VAT, therefore an

adjustment is made.

- Both motor vehicle tax and insurance premiums are not subjected to VAT, therefore, no adjustment is necessary.

Maintenance Costs

No prior research has investigated the determinants of the maintenance costs of a car. However, one may assume that the wear of many components (e.g. tyres and brakes) is positively related to the weight and horsepower of a car. Mileage, of course, is also an

important determinant, but it is not relevant in this analysis since both the source data and the lease contracts use the same annual mileage. Furthermore, the prices of replacement parts for more expensive cars are generally higher than those for cheaper cars. Therefore, I expect that the maintenance costs to be positively related to the weight, horsepower and purchase price of a car. The ANWB data is used to estimate the values of the coefficients in Equation (3). I perform the regression analysis both for the sample as a whole and for the petrol and diesel sub-samples. A dummy variable (DUMDIE) is included to distinguish between the two fuel types. The coefficients found in the regression analysis are then used to calculate the expected maintenance costs incurred by the lessors for each specific car type.

ε + + + + + = i i i i

i b b PRICE b WEIGHT b HP b DUMDIE

MAINT ₀ ₁* ₂ * ₃ * ₄* (3)

PRICE = Retail price of the car (in € ‘000) WEIGHT = Weight of the car (in kg)

HP = Horsepower of the car (in kW)

(29)

Since the results obtained when using Equation (3) in the regression analysis show a far from normal distribution of the residuals (see Appendix I), two alternatives are considered. The first is scaling the maintenance costs against the retail price of the car (PRICE), the second is using the logarithm of the maintenance costs as the dependent variable. I find that combining both alternatives, thus using the logarithm of the scaled maintenance costs (LOGMAINTPR) as the dependent variable leads to the most normally distributed residuals. Outliers where identified by examining the residuals, a total of seven individual cars (4 petrol, 3 diesel) were removed from the sample since their outliers were disproportionately large (more than three times the standard deviation in the residuals). The regression results are shown in Table VII. Since the residuals are not homoskedastic, I estimate the regression results using White’s heteroskedasticity-consistent standard errors and covariance.

Table VII

Regression Results of Maintenance Costs

Dependent Variable: LOGMAINTPR

Total sample Petrol Diesel

Variable Coefficient Prob. Coefficient Prob. Coefficient Prob. C 0.5245 0.0000 0.5863 0.0000 0.7938 0.0000 PRICE -0.0069 0.0000 -0.0076 0.0000 -0.0072 0.0000 WEIGHT -0.0367 0.0210 -0.0872 0.0003 0.0016 0.9017 HP 0.0000 0.8320 0.0002 0.3408 0.0005 0.0030 DUMDIE 0.3657 0.0000 - - - - Adjusted R² 0.9426 0.8605 0.8715 Durbin-Watson 1.895 1.985 1.983 Bera-Jarque 65.44 0.000 26.40 0.000 8.47 0.014 F Statistic 2,121.83 0.000 634.08 0.000 471.08 0.000

None of the regressions result in normally distributed residuals, therefore the results are not completely reliable. However, non-normality is less of a problem when sample sizes are large (Brooks, 2002). Additionally, since this is the only data on maintenance costs available, there is no feasible alternative but to use these results. The values for the Durbin-Watson statistics are well within the limits to reject the presence of autocorrelation in the residuals11. Given that the highest fit of the model to the data (highest adjusted R-squared)12 is obtained when the sample is analyzed as a whole, I choose to use the results for the entire sample to estimate the maintenance expenses. Since horsepower (HP) is clearly not significant, it is not included in

11

The Durbin-Watson critical values were obtained from Standford (www.stanford.edu/~clint/bench/dwcrit.htm)

12

(30)

the maintenance cost estimation, all other independent variables are included since they are significant at the 5% level.

Since lease prices are not adjusted for inflation, the risk of rising maintenance expense remains entirely with the lessor. As stated above, the source data has been adjusted to reflect the price level as of June 1st, 2008. Based on the expected annual inflation of 1.75% over the duration of the lease, the estimated maintenance expenses are increased by 0.145% monthly to reflect inflation.

Motor Vehicle Tax and Insurance cost

The motor vehicle tax is positively related to the weight of the car. Since repair costs in case of damage are higher for more expensive cars I expect the insurance costs to be positively related to the purchase price of the car. Additionally, the risk of an accident (both in terms of probability of occurrence and severity of damage) increases with the weight and horsepower of the car. Therefore, the weight, horsepower and purchase price of the car are included in the analysis. Equation (4) is used in the regression analysis.

ε + + + + = _i _i _i i b b PRICE b WEIGHT b HP MVTINS ₀ ₁* ₂ * ₃* (4)

The sample that was used in the analysis of the maintenance expenses is also used here. Since the residuals are not homoskedastic, I estimate the regression results using White’s

heteroskedasticity-consistent standard errors and covariance. The results of the regression analysis are shown in Table VIII.

Table VIII

Regression Results of Motor Vehicle Tax and Insurance Costs

Dependent Variable: MVTINS

(31)

It is obvious from Table VIII that the most reliable results are obtained when analyzing the sample as a whole since this does result in normally distributed residuals. The residuals for this analysis are normally distributed and the R2 statistic shows a good fit of the model with the data. The Durbin-Watson statistic is within the critical values, showing no autocorrelation within the residuals. Splitting the sample into separate groups for petrol and diesel powered cars does lead to an improved fit, but the results are not reliable since the residuals are far from normally distributed. This does not significantly improve when adjustments to the variables are made as is done in the analysis of the maintenance costs. Since all regressors show significant coefficients, they are all included in the motor vehicle tax and insurance costs estimate.

Expected Salvage Value

The car’s market value at the end of the lease partly determines the pay-off to the lessor. Unfortunately, very little research has been performed in this area. Pratt and Hoffer (1990) perform an analysis on the determinants of depreciation in the US market using second-hand prices of domestic, European and Japanese cars. As could be expected, they find that total depreciation is positively related to the age of a vehicle. They also find that total depreciation is significantly positively related to maintenance expenses and fuel consumption.

Additionally, technological obsolescence (i.e. introduction of a new generation of a model) led to a significant decline in second-hand value. However, the fit of the model of the Pratt and Hoffer (1990) model with the data was low (R2: domestic: 0.469; European: 0.696; Japanese: 0.670). Given, however, that this method is the only one proposed in prior literature, I use this method for this research.

Based on Pratt and Hoffer (1990), I presume that the total depreciation is positively related to the maintenance expenses and fuel consumption. The maintenance expense is using the results shown in Table VII. The fuel consumption as stated by the car manufacturer is used. The replacement of a model by a new generation is difficult, if not impossible to predict, and is not included in the analysis13. Given the differences Pratt and Hoffer (1990) find between

automobiles from different geographic origins, it may be necessary to split the sample into different geographic origins in order to increase the accuracy of the estimates. In addition to

13

(32)

these variables, I also add the original purchase price of the car to the equation. It is generally assumed that more expensive cars depreciate faster than cheaper cars. The model shown in Equation (5) is used for this analysis.

ε + + + + = i i i o

i b b MAINT b CONS b PRICE

DEP ₁* ₂ * ₃* (5)

DEPi = Economic depreciation of car i over the first 48 months after purchase CONSi = Fuel consumption of car I (L / 100 km)

Since the results obtained using the model above show a far from normal distribution of the residuals (see Appendix II), the logarithm of the total depreciation (LOGDEP) is used as dependent variable. The results of the regression analysis are shown in Table IX. Since the residuals are not homoskedastic, I estimate the regression results using White’s

heteroskedasticity-consistent standard errors and covariance.

Table IX

Regression Results of Expected depreciation

Dependent Variable: LOGDEP

Variable Coefficient Prob. Coefficient Prob. Coefficient Prob. C 0.4331 0.0000 0.1592 0.0000 0.2962 0.0000 MAINT 0.0047 0.0000 0.0143 0.0000 0.0058 0.0000 CONS 0.0223 0.0000 0.0180 0.0000 0.0118 0.0003 PRICE 0.0077 0.0000 0.0021 0.0194 0.0034 0.0002 DUMDIE -0.2065 0.0000 - - - - Adjusted R² 0.8877 0.9276 0.9240 Durbin-Watson 2.094 1.983 2.010 Bera-Jarque 4.61 0.100 1.67 0.434 2.79 0.248 F Statistic 823 0.000 1,026 0.000 710 0.000

(33)

The expected depreciation of the cars in the lease price-sample is calculated by using the coefficients as shown in Table IX and the characteristics (fuel consumption, price) of the cars. The maintenance costs are estimated using the results shown in Table VII. This method does not correct for possible differences in inflation between the past 48 months and the next 48 months.

Credit risk

The lease payment compensates the lessor for both the ‘foregone use of the asset’ and the ‘potential consequences of default’ (Grenadier, 1996). Even though Smith and Wakeman (1985) do not distinguish between the promised and actual lease payment in their model, there is a difference in practice. A lessor is dependent upon the creditworthiness of the lessee for obtaining the lease payment. The difference between the contractual lease price and the expected lease payment is formed by the expected loss due to credit risk.

(34)

Table X

Expected Credit Losses on Automobile Leases Age (months) Probability of Default (PD) Cumulative PD Loss Given Default (LGD) Expected Loss (PD x LGD) Cumulative E(Loss) 0-11 1.99% 1.99% 31% 0.62% 0.62% 12-23 3.13% 5.18% 15% 0.47% 1.09% 24-35 2.92% 8.25% 1% 0.03% 1.12% 36-47 2.09% 10.52% 1% 0.02% 1.14% Average 0.28% (Source: Schmit, 2004)

In order to integrate the credit loss figures into the calculations of the IRRs, I perform the following adjustments. LGD is defined as a percentage of the amount outstanding which is equal to the value of all promised future payments at default plus the estimated re-sale value of the automobile. Therefore, the cumulative product of the LGD and PD can be directly deducted from the contractual lease payments to arrive at the expected lease payment. The average expected loss is deducted from the expected re-sale value to arrive at the expected cash flow from re-sale. When a lease defaults (and is terminated), the lessor no longer incurs any expenses related to the car and cannot further depreciate it for tax purposes. These amounts are reduced by the cumulative probability of default to reflect this effect in the IRR calculations. Table XI shows a overview of the adjustments made.

Table XI

Cash Flow Adjustments for Credit Loss Period

(month) Lease payment

Re-sale at

maturity Maintenance costs

Motor vehicle tax and Insurance Depreciation tax shield 0-11 -0.62% -1.99% -1.99% -1.99% 12-23 -1.09% -5.18% -5.18% -5.18% 24-35 -1.12% -8.25% -8.25% -8.25% 36-47 -1.14% -10.52% -10.52% -10.52% 48 (end) -0.28%

3.5 WACC

Since one of the benchmarks for the IRRs on leases is the cost of capital, I need to calculate the cost of capital of a Dutch leasing company. I follow the methods presented in Koller et al. (2005) to calculate the Weighted Average Cost of Capital for a firm. The complete

(35)

The peer group is composed using Amadeus, it consists of all European publicly listed companies of which the primary BIK code is 71102 (Operational leasing of automobiles). Of this peer group seven companies were removed due to lack of financial data or stock

information. Additionally, two of these companies are located in former Eastern Block countries, which cannot be considered representative for the Dutch market. Of the five

remaining companies, four are located in the UK and one is located in Belgium. The financial data and stock information on the companies was also obtained from Amadeus. The company β’s are obtained from Bloomberg.

The four-year German Government bond yield (obtained from the Deutsche Bundesbank) is used as the risk-free rate, German government bonds are generally considered the safest investment category in the Euro area. The duration was chosen as this matches the duration of the leases in the sample. The BBB credit spread for Financial Services firms is used since this is most representative of the average Dutch leasing company. A summary of the calculations is shown in Table IX.

Table XII

Weighted Average Cost of Capital Calculation

WACC parameters Value Explanation

Risk-free rate 3.90% 4-yr Deutsche Bund yield (1-5-2008) Market premium 5.00% Koller et al. (2005) average

Credit spread 2.50% Damodaran BBB credit spread (Financials) Marginal tax rate 25.50% Dutch corporate tax rate

Required rate of return on debt 6.40%

Target Debt to Equity 5.48 Peer average Unlevered Beta 0.17 Peer average

Levered Beta 1.11

Required rate of return on levered beta 9.44%

Weighted average cost of capital 5.49%

The obtained WACC seems to be rather low compared to other industries. Benchmarking the parameters against the data from Damodaran (2008) shows an interesting difference. For European leasing companies he obtains a levered beta of 1.07 (against 1.11), but a

Debt/Equity ratio of 10.43. Entering these parameters into the calculation leads to a WACC of 5.16% (ceteris paribus). The average forward-looking BARRA beta14 for the peer group is

14

(36)

0.24 (Barra, 2008), using this input leads to a WACC of 5.83%. Both comparisons indicate that the WACC obtained is quite robust to changing the assumptions. There are two issues concerning the WACC calculation that must be mentioned. The first is that the WACC calculation is error-prone, especially when a high leverage is concerned. The second that one could disagree on whether all parameters chosen correctly reflect the required return on a lease, especially since it has a relatively short duration. Therefore, sensitivity analyses on the comparison of the IRRs with the WACC are performed in the next section.

3.6 Fixed Costs

In order to determine the competitiveness of the Dutch automobile lease market I also

investigate the average IRR generated by leasing companies rather than just the marginal IRR. I use the financial data for 2006 and 2007 for peer companies to determine average fixed costs (‘other operating expenses’) as a percentage of revenue. I use the same peer group as in the WACC calculation, the data is obtained from Amadeus. The calculations are shown in Table XIX.

Table XIII

Revenue and Operating Expenses of Peer Group

2007 2006

Revenue Other OpEx OpEx /

Revenue Revenue Other OpEx OpEx / Revenue Avis Europe Plc 1,332 477 35.8% 1,344 513 38.2% Northgate Plc 772 108 14.0% 537 75 13.9% Helphire Group Plc. 437 110 25.1% 339 97 28.8% Autohellas (Hertz) S.A. 124 10 8.4% 117 10 8.2% Axis Intermodal Plc 16 5 30.9% 14 4 29.9%

Average 22.9% 23.8%

(37)

4. Empirical Results

This section first presents the results of the calculation of the IRRs on individual leases. The results of the statistical tests performed on the IRRs are then presented.

4.1 Lease IRRs

The following model was used to calculate the IRRs of the leases investigated.

0 ) 1 ( ) ( ) 1 ( ) 1 ( ) ( ) 1 ( 1 1 = + + + + + − − − + −

∑

= = T t t t T T t t t t t IRR DEP IRR SALV IRR MVTINS MAINT LEASE INV τ τ (6)

A summary of the calculated IRRs is shown in Table XIV. This table shows the descriptive statistics for the sample as a whole as well as for both the petrol and diesel subsample.

Table XIV

Descriptive Statistics of Leasing IRRs

Sample Company Average Median

Standard

Total sample All companies 5.39% 5.19% 2.26% -6.16% 15.49% Directlease 5.39% 5.09% 2.18% -1.73% 15.49% Leaseplan 5.74% 5.54% 2.16% -6.16% 12.62% Toplease 5.05% 4.88% 2.38% -3.22% 11.40%

Petrol All companies 5.18% 4.94% 2.29% -0.66% 19.80% Directlease 5.21% 4.82% 2.25% -0.61% 15.49% Leaseplan 5.58% 5.30% 2.17% 1.40% 19.80% Toplease 4.73% 4.60% 2.38% -0.66% 11.40%

Diesel All companies 5.68% 5.51% 2.25% -6.16% 12.62% Directlease 5.60% 5.36% 2.07% -1.73% 11.30% Leaseplan 6.01% 5.76% 2.32% -6.16% 12.62% Toplease 5.43% 5.34% 2.33% -3.22% 11.17%

(38)

4.2 Benchmark Tests

In order to determine whether the prices charged by leasing companies are equal to the marginal costs I compare the average marginal IRRs to two benchmarks. The first is the WACC for Dutch leasing companies of 5.49%, the second is the WACC plus 20 percentage points (25.49%). The single sample t-test in Eviews (‘simple mean hypothesis test’) is used to perform these tests. The null hypothesis of this test is that the mean of the sample is equal to the benchmark value against the two-sided alternative that it is not. The results of these tests are presented in Table XV.

Table XV

Results of Single Sample T-Test of IRRs against Benchmark Returns

Benchmark: WACC (5.49%) WACC + 20% (25.49%)

Sample Company Average T-Statistic Prob. T-Statistic Prob.

Total sample All companies 5.40% -1.36 0.17 ** -320.57 0.00 Directlease 5.39% -0.98 0.33 ** -194.57 0.00 Leaseplan 5.78% 2.70 0.01 * -184.81 0.00 Toplease 5.05% -3.92 0.00 -181.18 0.00

Petrol All companies 5.18% -3.71 0.00 -240.01 0.00 Directlease 5.21% -1.93 0.05 ** -140.92 0.00 Leaseplan 5.58% 0.68 0.50 ** -143.38 0.00 Toplease 4.73% -4.98 0.00 -136.39 0.00

Diesel All companies 5.68% 2.10 0.04 * -215.29 0.00 Directlease 5.60% 0.78 0.44 ** -135.71 0.00 Leaseplan 6.01% 3.19 0.00 -118.52 0.00 Toplease 5.43% -0.36 0.72 ** -121.95 0.00 * significant at 1% level

** significant at 5% level

(39)

on a marginal level, leaving little or no room to cover fixed costs (see Section 4.5). It may also indicate that one of the cost components in the model has been incorrectly estimated. Additionally, all observations discussed above are very sensitive to the assumptions regarding the WACC. As can be observed in Appendix IV, an increase of decrease of the WACC by 0.5% has a significant impact on the findings.

4.3 Cross-company Differences

I also compare the average marginal IRRs across the three leasing companies in order to investigate whether there are significant differences in the IRRs generated by these

companies. For comparison across the three companies, I use the ANOVA test for equality of means. The null hypothesis is that means across the subsamples are equal against the

alternative that they are not. I also test for the equality of IRRs between each pair of

companies. For this purpose I use the independent samples t-test. Again, the null hypothesis is that the means are equal against the alternative that they are not. The results of these tests are shown in Table XVI.

Table XVI

Results of ANOVA and T-Test Tests across Leasing Companies

Sample Companies

ANOVA

F-stat. T-stat.

Degrees of

Freedom Prob.

Total sample All companies 11.54 (2, 1329) 0.00 Directlease - Leaseplan 6.88 886 0.01 * Directlease - Toplease 2.23 886 0.03 * Leaseplan - Toplease 4.71 886 0.00

Petrol All companies 8.67 (2, 729) 0.00 Directlease - Leaseplan 1.86 486 0.06 ** Directlease - Toplease 2.29 486 0.02 * Leaseplan - Toplease 4.14 486 0.00

Diesel All companies 3.56 (2, 597) 0.03 * Directlease - Leaseplan 1.86 398 0.06 ** Directlease - Toplease 0.78 398 0.43 ** Leaseplan - Toplease 2.51 398 0.01 * * significant at 1% level

** significant at 5% level

(40)

investigating the equality across pairs of leasing companies one observes that Directlease and Leaseplan charge approximately the same in the petrol segment and that the greatest

differences in the diesel segment exist between Leaseplan and Toplease (the null hypothesis is accepted at 1% level only). On the whole, there are significant differences in the average marginal IRRs generated by the three leasing companies. The largest differences are found within the petrol segment and between Leaseplan and Toplease.

4.4 Cross-segment Differences

In order to investigate whether the differences in IRRs calculated are attributable to difference between specific segments I split the sample into sub-segments based on the geographic origin of the car. I distinguish four regions which I use to segment the sample: Northern Europe (NE: UK, Sweden, Germany), Rest of Europe (RE: France, Italy, Spain, Czech Republic, Romania), Asia (AS: Korea, Japan) and USA. The descriptive statistics of the average marginal IRRs per geographic segment are shown in Table XVII. I also present the results for the single sample t-test in this table. I perform the same test as in Section 4.2 here, using the WACC as the benchmark value. The figures for the whole petrol and diesel samples are also included in the table.

Table XVII

Descriptive Statistics and Results Single Sample T-Tests of Leasing IRRs per Geography

Descriptive Statistics WACC (5.49%)

Sample Region

No. of

Cars Average Median

Standard

Deviation Min. Max. T-Statistic Prob. Petrol All 244 5.18% 4.94% 2.29% -0.66% 19.80% -3.71 0.00 NE 84 4.66% 4.27% 2.22% -0.66% 15.49% -5.91 0.00 RE 125 4.91% 4.88% 2.33% -6.16% 19.80% -4.80 0.00 AS 78 5.73% 5.66% 1.97% 0.24% 11.00% 1.83 0.07 ** US 16 7.78% 7.74% 1.85% 4.64% 11.69% 8.6 0.00 Diesel All 200 5.68% 5.51% 2.25% -6.16% 12.62% 2.10 0.04 * NE 76 5.10% 4.81% 1.97% 1.72% 11.16% -2.99 0.00 RE 59 5.32% 5.33% 2.38% -6.16% 10.32% -4.80 0.00 AS 52 6.35% 5.96% 2.06% 2.16% 12.62% -5.59 0.00 US 13 8.08% 8.65% 1.74% 5.01% 11.30% 9.32 0.00 * significant at 1% level ** significant at 5% level

Competitiveness of the Dutch Automobile Lease Market