-This video is the second of two videos on the analysis of demand and potential demand for electric vehicles using the concepts and tools of economics. The goal of this video is twofold, firstly, knowing the tools for evaluating the potential demand for electric vehicles, and secondly, providing a critical perspective on the analyses and forecasts of the deployment of electric vehicles. Assessing the potential demand for electric vehicles for a given territory or population, i.e. the level of demand that a product may expect to reach in the future on this territory or in this population, proves very difficult. This is notably due to the complexity of interactions within mobility systems in general, and electric ones in particular, to the uncertainty about the main parameters affecting the conditions of deployment of electric mobility, energy prices for example, and to the complex and uncertain game of the parties involved in this deployment. The results of assessing the potential demand are therefore uncertain, all the more so as the considered geographic scope is large, and the projected time horizon is remote. In 2009, estimates of the potential demand for electric vehicles ranged from 8 to 50% of new global vehicle sales by 2030, and from 10 to 90% of new global vehicle sales by 2050. Two types of approaches may be used for assessing the potential demand for electric vehicles, each type corresponding to different goals. The first type, aggregate approaches to potential demand, is essentially used to identify long-term trends in vehicle purchase behavior, therefore in the market shares of innovative technology such as electric vehicles, generally on broad scopes. However, it does not inform about the characteristics of potential buyers, or about their preferences. Other methods, that we will refer to as disaggregate approaches to potential demand, are used to identify the characteristics of potential buyers, their location if necessary, their preferences, and so on. Aggregate analyses of potential demand for electric vehicles consider demand from a group as a whole, for example all French households, and study what part of this whole is compatible with the characteristics of the new product or service considered. Diffusion models fall into this category. The Bass diffusion model, developed in the 1960s, and widely used in marketing and management sciences, links the diffusion of a new product or service firstly to an inclination of potential buyers to innovation, and secondly to an inclination of the same buyers to imitate those having already purchased the product or service. Analyses of compatibility of practices generally provide an assessment of potential demand for electric vehicles based on the segmentation of daily journeys into distance classes. If two thirds of company vehicles drive less than 100 km a day and the range of the electric vehicle is 100 km, then the electric vehicle may be a practical alternative for two thirds of company vehicles. Finally, cost-effectiveness analyses provide an assessment of potential demand for electric vehicles based on the comparison of purchase and usage costs of these vehicles compared with those of a reference vehicle, often thermal. Here, for a given annual mileage, we understand that the higher purchase cost of the electric vehicle may be paid off after a number of years of ownership given the lower usage costs. The same reasoning could apply for a fixed number of years of ownership and be used to determine beyond which annual mileage the electric vehicle becomes cost-effective compared to its thermal rival. We recognize the aggregate analyses that we discussed to the fact that they do not inform on the individual characteristics of the potential buyers of electric vehicles. They cannot be used to identify the socio-professional category, income group, age, or place of residence of these potential buyers, nor can they be used to distinguish different preferences among these buyers. The target groups that are identified by aggregate analyses are identified at a very aggregate level. Explicitly or implicitly, aggregate analyses are based on assumptions, for example on future energy prices, electric vehicle ranges, and so on. And the results of these analyses may prove very sensitive to the levels chosen by the analyst for some of these parameters. Any result of aggregate analyses of compatibility or cost-effectiveness that does not come with a sensitivity analysis to the main parameters should be considered at least with caution. Some aggregate analyses of potential demand, if they are based on averaged or partial data, may also lead the analyst to underestimate the exceptional uses of the vehicle, such as holiday journeys, and the weight that such journeys may have in the choice of vehicle at the time of purchase. Finally, aggregate approaches, usually relying on a limited number of choice parameters, for example average daily journeys, or total ownership costs, or a combination of both, leave out many choice factors, notably those that may reflect the diversity of preferences within the members of the considered group, in terms of access to parking and charging infrastructure, but also in terms of self-image, of comfort, of technical performance, of environmental performance, and so on. Disaggregate approaches to potential demand are generally used in scientific literature to identify the individual characteristics of potential buyers of electric vehicles, their location, and so on, rather than to predict long-term diffusion trends. Unlike aggregate approaches, they take into account individual characteristics of potential buyers, whether they are individuals, households, or companies, and may reflect different preferences and purchase motivations among these buyers. The first category of disaggregate approaches to potential demand that we will discuss here is that of models of vehicle choice. These models may be based either on detailed data on potential buyers, their current vehicle, their current travel practices, usually collected through revealed preference surveys, notably household travel surveys, or on data from surveys placing potential buyers in a hypothetical situation of choice of a new vehicle, then called stated preference surveys. Models of vehicle choice may provide an assessment of potential demand for electric vehicles based on the various choices made by various types of buyers, more or less young, more or less rich, more or less urban, and so on, between targeted characteristics of the vehicles, such as their price, range, comfort, and so on. Models of vehicle choice as we saw, are used to identify the individual characteristics of potential buyers, and to locate potential demand. They are also used to quantify the choices that they make between different attributes of the vehicles presented to them. A choice between purchase and usage cost, between total ownership costs and range, between cost and environmental performance, and so on. Note however that these models rely on data costly to produce. Furthermore, for models based on revealed preference surveys, the issue arises of the transferability of the observed choices to choices in the presence of vehicles with all new characteristics. And for models based on stated preference surveys, the issue arises of the transferability to real life of the choice made in a hypothetical situation, maybe in a situation of partial information. A second category of disaggregate analyses of potential demand that we will discuss here is analysis by constraint. Here we will apply it to the particular case of households. The principle of analysis disaggregated by constraint is as follows: with a detailed database of households, their vehicles, and their mobility, for example a household travel survey, first is considered the population of the households as a whole, then is applied a series of selection filters, corresponding to as many criteria of compatibility with the electric vehicle. For example, we may consider that only motorized households are likely to buy an electric vehicle. We may then consider that among them, only those with access to a parking space where they will be able to install a charging station are likely to take the leap. We will then study the compatibility of their journeys with the range of the electric vehicle, and finally we will check if the electric vehicle is economically beneficial to them. Like models of vehicle choice, analyses disaggregated by constraint may be used to identify the individual characteristics of potential buyers, and to locate potential demand. These analyses may also be used to assess the relative weight of the various constraints, and may help public authorities or vehicle manufacturers to identify the principal barriers to the deployment of electric vehicles, parking, range, cost, and so on. However, like models of vehicle choice, analyses disaggregated by constraint are based on costly data, notably data from large-scale surveys. Analyses by constraint are sensitive to the assumptions of the analyst, for example on the way to process households with one vehicle and households with several vehicles. Finally, analyses by constraint only take into account a limited number of dimensions in the choice of vehicle, and generally assume the substitution, all other things being equal, of a thermal vehicle by an electric vehicle. In conclusion, you understood that there is no magic formula for assessing potential demand for electric vehicles. There is a variety of approaches, more or less adapted depending on the objectives, and on the available data. Regardless of the approach, assumptions are made, explicitly or implicitly, about the parameters of the analysis, and these assumptions influence, sometimes significantly, the estimated potential demand. There is no objective assessment of potential demand, at best these assessments are transparent and discussed, for example through a sensitivity analysis.
