Modelling and forecasting multi-generation, multi-country ... · Web viewIslam and Meade (1997) argue that after the radical innovation of the first generation of a technology, in

Transitions between Technological Generations

of Alternative Fuel Vehicles in Brazil

Abstract

The transportation sector is responsible for nearly a quarter of greenhouse gases emissions (GHG); thus,

incisive policies are necessary to mitigate the sector’s effect on climate change. Promoting alternative fuel

vehicles (AFV) is an essential strategy to reduce GHG emissions in the short term. Here, we study the

effects of governmental incentives on the diffusion of ethanol and flex-fuel vehicle technologies in Brazil.

We use a multi-generation diffusion model which assumes that new technologies introduce fresh market

potential for adopters as well as upgraders from established technologies. Our analysis indicates that tax

rates affected the adoption of both gasoline and ethanol technology, but for flex vehicles, the effect of

taxation is not significant. The effect of fuel price shocks during the 1990s meant that the introduction of

ethanol technology made no significant impact on market potential and a negative word-of-mouth effect

contributed to the technology’s failure. In contrast, the introduction of flex technology led to almost a

doubling of total market potential. As policy suggestions, we emphasise the importance of tax reduction

in addition to promoting versatile technologies, which insulate consumers against price fluctuations.

1

1. Introduction

The diffusion of alternative fuel vehicles (AFV) for transport faces uncertainty inherent of

any innovation. Costs, compatibility issues, and the complexity of new technologies are all risks to be

assessed from the consumer’s perspective (Rogers, 2003). In the case of transport, there is the imperative

of reducing CO2 emissions. In 2010, transport accounted for 23% of energy-related CO2 emissions

according to the Intergovernmental Panel for Climate Change (IPCC, 2014). Their report emphases that,

if countries do not promote more aggressive and continuous policies, emissions of transport greenhouse

gases (GHG) would be the fastest to increase. The Paris Agreement recognises the need to undertake

rapid reductions of GHG to hold the increase of global average temperature below 2º C above pre-

industrial levels (UNFCCC, 2015).

Past experience, where technologies have gone through their full diffusion cycle, show how

these policies may impact positively and negatively. One of the major examples found in the literature is

the Brazilian Alcohol Program (Pró-Álcool). Officially established in 1975, its goal was to supply the

internal market with an alternative liquid fuel to gasoline, thus reducing imports of oil products by

producing ethanol from sugarcane (Goldemberg et al., 2004; Moreira and Goldemberg, 1999). It was

Brazil’s primary strategy to overcome the oil crisis from the 1970s. Despite the program’s shortcomings

during the 1990s, the rapid adoption of flex-fuel technology has enabled a rebirth of ethanol as a

widespread vehicle fuel. Throughout its history, government support was an important variable.

Through diffusion studies, we can understand how a new idea or product spreads among a

social system, such as a country or region, and which communication channels are used to reach each of

its individuals (Rogers, 2003). Nevertheless, social systems are complex entities where change agents (e.

g. government, companies) can push or discourage new technologies among a pool of competing options.

Some modelling frameworks have sought to study the diffusion of renewable energy technologies (RET)

and AFV, but few have investigated the latter’s relationship with government incentives and taken into

account the interaction of successive generations of technologies. Danaher et al. (2001) show that single

2

generation models miss essential components of diffusion processes, especially when considering

marketing factors. Thus, models that incorporate multiple technological generations can bring substantive

insights into strategic and policy decisions. Norton and Bass (1987) argue that applying multi-generation

models is plausible and historically accurate, with plenty of examples (e. g. in marine power, steam

engines replaced ship’s sail, and then it was then substituted by internal combustion engines).

Our objective is firstly to understand the evolution of a successive generation of vehicle

technologies associated with the Brazilian Alcohol Program, namely ethanol and flex-fuel; and secondly

to investigate how government incentives (tax reduction) have affected the diffusion of such

technological generations. The contributions of this paper are the extension of the multi-technology

diffusion model by incorporating the effects of leapfrogging and marketing effects. We also include the

acceleration of diffusion speed (Islam and Meade, 1997) and relative influences of external effects and

social learning (i.e., imitation) on the diffusion of multi-generations of AFV.

For many authors, vehicle innovations are less likely to have a wide diffusion if supporting

refuelling stations are absent; however, building costs are prohibitive if the scale is small (Brito et al.,

2017; Gnann and Plötz, 2015; Kloess and Müller, 2011; Leibowicz, 2018; Meyer and Winebrake, 2009;

Mouette et al., 2019). They consent that, to break this so-called chicken-and-egg dilemma, infrastructure

must come first, and that is where government incentives are important. We did not address this issue for

ethanol in Brazil, because it follows a similar distribution chain to gasoline – however, we acknowledge

that this issue is relevant for diffusion studies about non-liquid fuel technologies such as hybrid/plug-in

electric, hydrogen, and natural gas.

In Section 2, we review some studies on the effect of incentives designed to enhance the

diffusion of renewable technologies and alternative fuels for vehicles. Further, we present a historical

overview of ethanol and flex technology in Brazil, along with policy and economic factors that

contributed to such development. Section 3 presents our data and their sources and how they have been

processed for use in our model. In Section 4, we describe the multi-generation diffusion model developed

for this study explaining our assumptions, limitations, and hypotheses. In Section 5, we give the results of

3

our analysis and discuss their implications. In Section 6, we give our conclusions and policy

recommendations appropriate for the diffusion of future generations of AFV in Brazil, or for other

countries wishing to introduce similar programs.

2. Review of Literature

Innovation diffusion models have been used since the 1960s to study the diffusion of:

consumer goods, such as electronics and house appliances; farming techniques; drug consumption

(Rogers, 2003), information and communications technologies (Islam and Meade, 1996; Meade and

Islam, 2015). Traditional models generally assume that consumer innovativeness and social learning, i.e.,

imitation are the main drivers of the diffusion process (Bass, 1969). Nevertheless, governments,

companies or other change agents can also stimulate diffusion. The following subsections discuss how

literature has approached the diffusion of alternative fuel vehicles and how government incentives can

promote them. We then present a brief history of the ethanol program in Brazil, along with technological

innovations that contributed to its development to clarify the context of our study.

2.1. Innovation Diffusion of Energy Technologies

Single generation diffusion models have been used to study natural gas (Zhu et al., 2015),

electric cars (Wansart and Schnieder, 2010), and fuel cell (Collantes, 2007), solar PV systems (Masini

and Frankl, 2002; Peter, 2002), and wind energy (Ibenholt, 2002; Rao and Kishore, 2009). These

applications are usually one of three types: general spread, pricing a diffusion variable or forecasting.

However, they to focus on commercial products and few paid attention to policy impacts and interactions

with successive generations of technologies (Al-Alawi and Bradley, 2013; Rao and Kishore, 2010).

Islam and Meade (1997) argue that after the radical innovation of the first generation of a

technology, in most cases, successive generations are less radical but can bring significant performance

improvements. They do not require the user to acquire significant new skills, as observed in computers

and electronics products. In this context, later adopters may prefer to acquire a more modern version of a

4

product, without buying the earlier product, while early adopters will subsequently upgrade. Part of the

pool of potential adopters of the earlier technology shifts to swell the pool of potential adopters as the

whole increase (Norton and Bass, 1987; Stremersch et al., 2010). Some authors used multiple

generational approaches to study high technology products, such as computers and video games (Shi et

al., 2014), cellular phones (Kim and Shon, 2011), memory chips (Jun and Park, 1999; Versluis, 2002),

milk containers (Speece and Maclachlan, 1992).

In the context of RET, Guidolin and Guseo (2016) employed a model incorporating

competition and regime change to investigate the substitution of fossil and nuclear energy in Germany.

They identified a high word of mouth influence on the diffusion of wind and solar power and confirmed

that this model was the best option for the German case, where the predominant energy transition is

directed to electricity production. Meade and Islam (2015a) developed a model to study the impacts of 10

covariates on the diffusion of wind, solar and bioenergy in 14 European countries. The model provided

relevant insights regarding such covariates, accurate forecasting densities and divided the countries into

four groups concerning their rate of RET adoption (slow, normal, fast and very fast).

While studying substitution of internal combustion engines (ICE) vehicles by hydrogen and

hybrid-electric cars, Struben and Sterman (2008) applied a dynamic behavioural model by introducing the

willingness to consider the adoption of those vehicles. They point out that awareness and adoptions must

pass given thresholds to allow a self-sustaining diffusion, and that a positive word of mouth effect might

have a significant impact on lowering this threshold. Jeon (2010) modelled successive technologies for

hybrid, plug-in hybrid and electric vehicles in the United States by applying a multi-generation diffusion

model(Norton and Bass, 1987) (Norton and Bass, 1987) to capture the impacts of vehicle and gasoline

price. The study projected that these technologies could achieve 8 million sales annually by 2030. These

studies are examples of the need to identify how AFV will fare amid the competition with conventional

energy sources.

2.2. The Impact of Tax Incentives on Diffusion Rates

5

Despite the availability of financial and fiscal incentives, renewable energy technologies

spread rates are still low, which is concerning in the face of the urgency posed by climate change and

other environmental issues (Rao and Kishore, 2010). According to Pfeiffer and Mulder (2013), diffusion

of RET (except hydro) across 108 developing countries, from 1980 to 2010, has accelerated in stable and

democratic regimes with high per capita income and schooling. Conversely, the probability of adopting

RET has reduced if the country is a large producer of hydro or fossil energy sources.

Freitas et al. (2012) indicated that Kyoto Protocol mechanisms incentivised the increase of

energy generation from renewables rather than promoted efficient and sustainable energy usage in BRICS

countries (Brazil, Russia, India, China, and South Africa). Aguirre and Ibikunle (2014) suggest that some

public policies fail to promote renewable technologies in BRICS countries because they are driven by

environmental concerns, but commitment to RET decreases under energy supply constraints.

Diffusion models have been used to study the effectiveness of government incentives in the

photovoltaic (PV) sector in several countries. These studies suggest that incentives must be timed to

coincide with the technology's diffusion stage as renewable technologies have a low innovation tendency

(Guidolin and Mortarino, 2010). Studies further suggest that several policies such as feed-in tariffs, trust

funds, specific lines of credit and the regulation of distribution tariffs for final consumers are necessary to

effectively diffuse PV energy (Ferreira et al., 2018; Muñoz et al., 2007; Pinto et al., 2016). In a review of

50 case studies throughout developed countries, Negro et al. (2012) indicate the necessity of stable long

term, but flexible, incentives.

Governmental policy is regarded as essential for the successful diffusion of vehicle

technologies, regardless of a country’s development level. Jenn et al. (2018) concluded that in the US, for

each $1000 incentive, electric vehicle sales rise by 2.6%. Lieven and Rietmann (2018) found that the

number of charging stations magnifies the effect of monetary incentives on the diffusion of electric

vehicles. Other studies also identified that economic stimulus has a significant and positive impact on

electric vehicles diffusion rates (Langbroek et al., 2016; Sierzchula et al., 2014). Egbue and Long (2012)

find that the use of tax credits to subsidise the cost of EVs increases consumer confidence. Benvenutti et

6

al. (2016) suggest that tax reductions are more effective when given to manufacturers than directly to car

owners.

For the diffusion of hybrid electric vehicles in Brazil, Benvenutti et al. (2017) modelled a

range of policy scenarios such as importation tax exemption, local tax reduction, ten-year federal tax

exemption and a more radical policy of banning conventional combustion engine cars. Their results

reinforced the importance of incentives, especially when combined. Similarly, Baran and Legey (2013)

simulated the diffusion of electric vehicles in Brazil and its impact on energy consumption. They used

exogenous parameters under different penetration scenarios and concluded that the Brazilian market is

open to new technologies. The following historical overview corroborates these findings.

2.3. The Alcohol Program and Ethanol and Flex Vehicles

The choice of ethanol as an alternative fuel for gasoline has foundations deep in Brazilian

history. The country had built up a sugar-cane industry over 400 years since its colonisation period, due to

its favourable tropical climate, with abundant water and land availability (Moutinho dos Santos and

Parente, 2006). Since 1931, the federal government has authorised a 5% blend of anhydrous ethanol with

imported gasoline, but until 1970, the industry used 70% of its production (Puglieri, 2013).

By the end of the 1970s, as a result of the oil shocks (Brito et al., 2012) petroleum prices had

two consecutive sudden rises (in 1973 and 1979), Brazil started its Alcohol Program to substitute

gasoline-fueled passenger cars for ethanol ones (Moreira and Goldemberg, 1999). The proportion of

ethanol in the blend with gasoline increased to about 20% during this period and remained at this level. In

addition to the oil shock, the national sugar industry was facing a crisis due to the collapse of the sugar

price in the international market. The Alcohol Program was created primarily to give sugarcane producers

a new commercial use for their crops, promoting the country’s energy security was secondary (Moutinho

dos Santos and Parente, 2006).

The Brazilian automobile industry was developed in the 1950s and afterwards by

multinational companies such as Volkswagen, Ford, GM, and Fiat, who imported their oil-based

7

technologies. By the start of the Alcohol Program, these gasoline engines converted to run on pure

ethanol were gravely deficient due to inadequate fuel-air mixture control and corrosion when (Moreira

and Goldemberg, 1999). The development of electronic fuel injection technology during the 1970s was

the main innovation that allowed the use of ethanol as a complete substitute to gasoline in Otto-cycle

engines, in conjunction with higher compression ratio engines using corrosion-resistant materials (Yu et

al., 2010). The first passenger ethanol vehicle registrations were in 1979 and increased during the 1980s

(ANFAVEA, 2018). By that time, the government had started to give incentives to ethanol producers and

dedicated vehicle manufacturers (Coelho et al., 2006; Moreira and Goldemberg, 1999). Taxes on ethanol

vehicles were about 5% lower than those on gasoline vehicles (ANFAVEA, 2018), and ethanol fuel was

tax-free for final consumers. Confidence in ethanol technology also played an essential role in the

Program's success as it was seen as patriotic (Moutinho dos Santos and Parente, 2006). However, from

1989 to 1992, Brazil experienced an ethanol shortage as sugarcane farmers shifted their production to

sugar, as the sugar price became more competitive than ethanol (Puglieri, 2013). This period is called the

first ethanol shock, which coincided with the oil ‘counter-shock’ when oil prices returned to competitive

levels. The fuel price ratio (ethanol's price compared to gasoline) went, on average, from 62%, during the

1980s to 77% during the 1990s. This situation frustrated consumers who had purchased ethanol vehicles

to insulate themselves from a price rise in oil products price, such as that in the 1970s (Moutinho dos

Santos and Parente, 2006).

Moreover, the government, targeting the low-income and middle-class population, had

started to give incentives to vehicles with small capacity engines (popular cars) (Bastin et al., 2010).

Ethanol vehicle sales systematically decreased during this period, clearly demonstrating that ten years

after the launch of the ethanol vehicle, it was still dependent on government incentives (Moreira and

Goldemberg, 1999). Regardless of a decrease in sales, the government has kept high levels of ethanol

blending with gasoline, as it is an excellent additive, mixtures of up to 20% ethanol were allowed by the

government. This policy sustained the Program since it maintained the producer's security against

8

international sugar prices volatility and encouraged long-term stable ethanol production (Goldemberg et

al., 2008; Moutinho dos Santos and Parente, 2006).

Environmental concerns expressed by the Kyoto Protocol brought a new perspective to

ethanol as a mean of reducing carbon dioxide (CO2) emissions (Moreira et al., 2005). This new phase of

the Brazilian Alcohol Program, marked by the Petroleum Law in 1997, started a series of liberalisation

policies and the expansion of cogeneration and electricity commercialisation with sugarcane bagasse (a

biofuel), which increased ethanol’s competitiveness (Goldemberg et al., 2004; Puglieri, 2013). By the

early 2000s, a sudden rise in oil prices, along with Kyoto Protocol’s rising number of ratifications, have

turned the world’s attention to biomass as an important source of renewable energy (Moutinho dos Santos

and Parente, 2006). Brazil used its expertise to become a leader in the international clean fuels

discussion and a lead exporter by 2004, as a new technological development gave a fresh breath to use of

ethanol as a vehicle fuel.

By the 1980s, researchers from the US and Brazil were testing an engine that could operate

on all blends of gasoline and ethanol (Pefley et al., 1980). The technological breakthrough was the

development of a software sensor installed into the electronic injection module that could measure fuel

concentration and adjust the engine accordingly (Teixeira, 2005; Yu et al., 2010). In the US, these engines

can use only up to 85% ethanol due to possible problems with a cold start (Bastian-Pinto et al., 2010).

However, the Brazilian system memorises the latest composition; thus allowing it to burn any

ethanol/gasoline blend or either separately (Coelho et al., 2006; Mesquita et al., 2013; Teixeira, 2005).

The first flex vehicles were registered in the Brazilian market by 2003 under the same taxation applied to

ethanol. Sales took off quickly, and in only three years, flex cars corresponded to more than 90% of the

market share. Unlike ethanol, flex vehicle sales have been dominant since 2006, and are now the

mainstream vehicle technology in the country. At the time of its launch, there was some concern about

fuel efficiency. However, current figures indicate that flex technology consumption is equivalent to a

single fuel vehicle when using gasoline, and has increased efficiency by roughly 40% when using ethanol

(CETESB, 2017). Both ethanol and flex vehicles presented both economic and technological

9

improvements over gasoline vehicles. From the consumer’s perspective, the flexibility of choosing which

fuel is a relative advantage over older technologies. The total flex technology available in Brazil was

only possible due to previous knowledge obtained through the experience with dedicated ethanol

vehicles. We consider that ethanol and flex technologies represent successive technological generations

after gasoline. These generations occurred under the long term policy of the Brazilian Alcohol Program

which promoted technologies by reducing taxation over them, among other policy instruments.

2.4 Institutional and Cultural Impact on AFV Diffusion in Brazil

According to institutional theory, resources originating from the state, the public sector or

multinational firms exert an external influence over the innovation diffusion process by increasing both

network ties and the adopter base (see DiMaggio and Powell, 1983; Guler et al., 2002; King et al., 1994).

The presence of coercive organisations and/or regulatory bodies or foreign multinationals in a country

will tend to introduce and enforce standardisation and accreditation. These actions may affect a

technology (or generation of technology) either positively or negatively (see Guler et al., 2002;

Rosenzweig and Singh, 1991). Due to its geography and its history, the diffusion of AFVs in Brazil is

firmly embedded in the national culture and the structures of institutions (Goldemberg, 2007). Brazil’s

innovation system for sugarcane ethanol results from an incremental technological learning process and

the interaction among different institutions, universities, research centres and firms. The coordination

between the productive process and the technological trajectory was essential for the competitiveness of

the ethanol industry (see Furtado et al., 2011).

National culture is a multi-dimensional construct which plays an essential role in cross-

country product diffusion. The influence of different cultural dimensions on product adoption/diffusion

varies depending on the nature of the product. Hofstede (2001) cultural dimensions are the most relevant

tools to study the impact of culture across different countries. Hofstede’s dimensions show that Brazil’s

collective society is high on uncertainty avoidance and power distance, while intermediate along other

dimensions. Brazil’s institutional structure, such as the need for complex regulatory systems, is linked to

10

high uncertainty avoidance (Hofstede’s Insights, 2019). These cultural aspects are also observed in

institutional operations (government, ethanol producers and vehicle manufacturers) throughout the

program’s phases. The creation of the Alcohol Program, launched in 1975, when the country was under a

period of military government, reflects the top-down push of the ethanol technology, strongly supported

by this regime’s policy to substitute imported oil products by domestic produced ethanol (Duarte and

Rodrigues, 2017; Nastari, 1983). This reflects the cultural dimension of high power distance where

inequality in power structure is widely accepted in the society. Later, during the 1990s, when the country

had resumed democratic government, a new regulatory system, the Petroleum Law, gave strong support

to ethanol producers (Goldemberg et al., 2014). More recently, a new government policy (Renovabio)

still shows the program’s dependence on regulation in order to sustain itself (Cordellini, 2018). The

inclusion of ethanol as a vehicle fuel has also caused changes to the population’s fuelling and driving

habits. Isabella et al. (2017) surveyed consumer’s preferences in Brazil from the perspective of diffusion

theory and supply change management and found that the environmental and convenience concerns of

flex-fuel vehicle owners meant that they do not invariably make economically rational decisions.

Subsequently, we will review our results in the context of these cultural dimensions.

3. Data Description

Before we discuss our multi-generation model of the Brazilian vehicle market, we will first

describe the data available for analysis. The main covariate in this study is the Tax over Industrialized

Products (IPI), a Brazilian federal tax applied to any manufactured national or imported product and can

only be modified by governmental decrees. Each product has a different rate based on the incentive or

restriction the government intends to promote. Since IPI taxpayers are wholesalers and retailers, the tax

rate has a direct impact on the final product price. Current rates are available at the Brazilian Ministry of

Finance (MF, 2018); however, a historical series is not available. The Brazilian Automotive Industry

Association (ANFAVEA, 2018, 1994) has collated data for the tax rate applied to gasoline and

ethanol/flex passenger vehicles of 1 to 2-litre engine capacity. These figures are available from 1987 for

11

both gasoline and ethanol vehicles. For years without a tabulated rate, we extrapolate the rate from the

last available observation, see Figure 1.

Figure 1. Federal Taxes Over Passenger Vehicles (ANFAVEA, 2018)

Vehicle sales data are taken from the Brazilian Automotive Industry Association

(ANFAVEA, 2018). However, it is important to note that cars are durable goods with a finite life span,

we must consider the fleet size and its composition to distinguish real year-by-year new adoptions from

replacement purchases. We apply a scrapping curve, a statistical estimation of vehicles removed from the

fleet due to total loss accidents, theft without recovery, dismantling and abandonment. Studies showed

that vehicle scrapping can be expressed by a Gompertz function (see Andersen et al., 2007; CETESB,

2017; Zachariadis et al., 1995). The Gompertz function used is shown in (1)

St=1−exp (−exp (a+bt ) ) (1)

12

where St is the surviving fraction of vehicles, t years after purchase. Both a and b values were

parameterised for cars (1.798 and 0.137, respectively) using the Brazilian National Household Sample

Survey (IBGE, 1988; MMA, 2014). By applying this scrapping curve to the vehicle sales data, we

estimate each year's operating fleet, its growth and composition. In Figure 2, we see the estimated fleet

sizes of gasoline, ethanol and flex vehicles plotted over time, the two oil shocks and the ethanol shock are

also shown.

Figure 2. Cumulative Adoptions (Estimated with data from ANFAVEA, 2018)

We see that the growth in the size of the Brazilian gas vehicle fleet slows in the mid-1970s as

a consequence of the first oil crisis, followed by a recovery. In 1979, ethanol vehicles were introduced,

in 1985 their sales represented more than 95% of total vehicle sales, in 1986 their sales peaked at almost

700 thousand units. After the ethanol shock, at the beginning of the 1990s, growth in gasoline vehicle

sales resumes, and ethanol vehicles sales drop to less than 1% of new cars. Flex-fuel vehicle adoptions in

Brazil started in 2003 and overtook gasoline vehicle sales in only two years. Since 2007, flex-fuel has a

yearly market share of about 90%.

13

4. The Multi-Generation, Multi-Country Model and its Estimation

In order to capture the changing dynamics of the Brazilian vehicle market and the effects of the

changing fiscal environment, we propose a multi-generation model in discrete time which draws on the

generalised Bass model (Bass et al., 1994). There are several alternative models for single generation

diffusion processes (e.g., Meade and Islam 2006; 2015); however, unlike the Bass model, these do not

typically generalise to multiple generations. Massiani and Gohs (2015) point out the difficulty of

choosing adequate parameters for a Bass Model of the diffusion of automotive technologies. They find

that most studies present a wide range of estimated parameters that show no pattern and thus cannot be

adopted reliably in future studies. We address this issue by applying a multi-generational model that can

estimate different parameters for newer technologies substituting for earlier ones. The model development

into three generations follows Norton and Bass (1987) and uses further developments by Danaher et al.

(2001), Jiang and Jain (2012) and Stremersch et al. (2010). We draw on the approach of Stremersch et al.

(2010) with necessary revisions and the inclusion of a marketing factor.

4.1. The multi-generational model structure

We consider three generations, denoted by the subscripts G, E and F, for gasoline, ethanol

and flex respectively. At time t, the fleet sizes or numbers of cumulative adopters in each generation are

YGt, YEt and YFt. An increase in fleet size may occur due to a fresh adoption, the numbers of fresh adoption

in each generation are AGt, AEt and AFt, alternatively a change will occur due to an adopter upgrading from

one generation to another. Upgraders from gas to ethanol are denoted by UGEt, from gas to flex, UGFt, and

from ethanol to flex UEft. Thus, the gasoline vehicle fleet size at time t is

Y ¿=A¿−U ¿ ,t−UGF , t (2)

14

we denote the period to period change using lower case, e.g. y¿=Y ¿−Y ¿−1, thus it follows that

y¿=a¿−u¿ ,t−uGF ,t (2a)

The ethanol vehicle fleet size at time t is

Y Et=AEt+U ¿ ,t−U EF , t (3)

and the period to period change is

y Et=aEt+u¿, t−uEF , t (3a)

Thirdly, the flex vehicle fleet size at time t is

Y Ft=AFt+UGF ,t+U EF ,t (4)

and the period to period change is

y Ft=aFt+uGF ,t+uEF ,t (4a)

We hypothesis that there are mG potential adopters of the gasoline car, those who would eventually adopt

a gasoline vehicle if no alternatives became available. There are mE additional potential adopters of

ethanol vehicle and mF incremental adopters of a flex vehicle. Figure 3 illustrates the dynamics of this

multi-generation adoption process.

15

Figure 3. Flow Diagram of multi-generation adoption process

In our study, we were able to observe only the cumulative number of current adopters of

each generation, YGt, YEt and YFt, the adopters of gasoline, ethanol and flex cars respectively. In equations

(1), (2) and (3), we hypothesise that these observable variables aggregate six underlying, but unobserved,

adoption processes: AGt, AEt , AFt, UGEt, UGFt, UEft. The total number of potential adopters for gasoline

vehicles mG; for ethanol vehicles is mG + mE of ethanol cars, and for flex vehicles is mG + mE + mF.

The approach of the discrete generalised Bass model (Bass et al., 1994) is that the number of

adopters in a time interval is a random variable determined by the conditional probability of adoption and

the number of possible adopters, with an additive noise term as shown for the gasoline generation in (5).

a¿={( pG+qG( Y G(t−1)

mG)) x¿}[mG−Y G (t−1)]+ε¿ (5)

16

mFmEmG

Gasoline Car Adopters, AG

Upgraders, UGE

New Ethanol Adoptors, AE

New Flex Adoptors, AF

Upgraders, UEF

Upgraders, UGF

Considering gasoline cars before the introduction of ethanol cars, the number of adoptions of

gasoline cars per period at time t is τG ≤ t<τ E. The Bass coefficients p and q are the coefficient of

innovation and the coefficient of imitation, respectively, we interpret q as the effect of word of mouth.

The marketing factor xt is used to reflect the tax environment, ε ¿ is a noise term representing the random

variation in this adoption process. The coefficients p, q and x are estimated for each generation and are

identified by a subscript G, E or F. Note that before ethanol and flex cars are introduced,Y ¿≡ A¿, after

the introduction of the ethanol and flex generations, YGt is modified by (2), in recognition of the erosion of

potential adoptions due to upgrading. The observability of gasoline adoptions is now decreased, but we

can still discern the relative importance of gasoline adoptions and upgrading. For example, when the

cumulative number of gasoline users decreases, we can tell that upgrading is dominating gasoline

adoption.

When the cumulative number of gasoline adoptions decreases, the magnitude of this

decrease forms a lower bound on the number of ethanol cars bought by upgraders. The upper bound on

the number of ethanol cars purchased by upgraders is the number of gasoline car users in the previous

period. For upgraders, the market factor is the relative benefit offered by the ethanol generation compared

to the gasoline cars, which these upgraders currently enjoy, see (6).

uGEt={(q¿( Y E (t−1)

mG+mE)) xEt}[Y G (t−1)]+ε GEt (6)

The remaining new ethanol car adopters are relatively observable in aggregate coming from the new

potential created by ethanol cars, see (7).

17

aEt={( pE+qE( Y E (t−1)

mG+mE))x Et}[mE−AE (t−1)−U EF ( t−1 ) ]+εEt (7)

Upon the introduction of the flex generation, its cumulative number of adoptions comes from three

different sources: flex new adoptions, upgraders from gasoline cars and ethanol cars, see (4) and (4a).

The same assumptions underlying (6) are used to model the adoptions of flex cars originating from

upgraders from gasoline and ethanol cars, see (8) and (9) respectively.

uGFt={(qGF( Y F (t−1)

mG+mE+mF)) xFt}[Y G (t −1)]+εGFt (8)

uEFt={(qEF( Y F (t−1)

mG+mE+mF)) xFt}[Y E (t−1)]+εEFt (9)

The remaining new flex car users are relatively observable in aggregate; these are adoptions from the new

potential created by flex cars (Equation 10).

aFt={( pF +qF( Y F (t−1)

mG+mE+mF))xFt }[mF−AF ( t−1 )]+εFt (10)

The marketing effort factor is used to capture the impact of government tax rates described

in Section 3, see (11a), (11b) and (11c).

x¿=exp ( βGTAX ¿) (11a)

xEt=exp ( βE TAX Et ) (11b)

18

xFt=exp ( βF TAX ¿) (11c)

5. Results

Following Stremersch et al. (2010), we estimated the parameters of the multigeneration model

described in Section 4 using SAS 9.4 by using the seemingly unrelated regression option on the PROC

MODEL procedure. The parameter estimates are given in Table 1, while the overall performance can be

seen in Figure 4, which shows the data alongside the model predictions for gasoline, ethanol, and flex

vehicles. The model has captured both the move away from gasoline technology, represented by negative

adoptions during the first half of the 1980 and since 2005, and the relative failure of ethanol technology.

The model fit (R²) for the three technologies is 68%, 33%, and 94%, respectively. The reasons for the

relatively poor fit for the ethanol generation become clear as we go through the implications of the

parameter estimates.

Figure 4 Model and Data Comparison for Vehicle Technologies in Brazil (1957-2017)

19

Looking first at the values of the estimated market potentials in Table 1, the potential for gasoline

vehicles is 33.53 million; for ethanol cars, the potential is not significantly different from zero; for flex

vehicles, the increased potential is 25.93 million. Thus, in total, the model predicts about 60 million

vehicles in the long run, in December 2017, the market penetration is 35 million, of which flex vehicles

represent 74% of the fleet, gasoline 25% and ethanol 1%. The model captures the failure of ethanol car

diffusion as it failed to bring new people into the market.

The market for gasoline vehicles behaves like that of a single technology model, with

significant coefficients of innovation (pG = 0.029, p= 0.008) and imitation (qG = 0.876, p < 0.001).

Likewise, the market for flex vehicles has parameters of similar sign and magnitude, (pF = 0.033, p =

0.024, qF = 1.022, p < 0.001). The exception is the model for ethanol vehicles which presents a high value

for innovation coefficient (pE = 1.06, p < 0.001), and a negative imitation coefficient (qE = -5.32, p =

0.06). According to Meade and Islam (2006), this corresponds to a pure innovation scenario, where the

adoption curve follows a modified exponential. As discussed by Mahajan et al. (1984), imitation (word of

mouth) can be favourable, unfavourable or indifferent towards a product. Here, it has acted against the

diffusion of ethanol vehicles. These findings show that the model’s parameters are consistent with the

loss of trust faced by this technology during the ethanol shock period, as described in our literature

review1. For flex vehicles, we -have a low estimate of coefficient of innovation (pG = 0.003, p = 0.024)

for flex-fuel as it is modified version of ethanol cars but flex-fuel received strong word of mouth impact

(qF = 1.02, p < 0.001).

We investigated whether innovation acceleration occurred across successive generations of

technologies. Acceleration happens when the difference between the two parameters is positive and

significant (Pae and Lehmann, 2003); thus, our model indicates a significant acceleration when

comparing ethanol and gasoline cars (pE – pG = 1.028, p < 0.001), which confirms that the diffusion of

1 A negative word-of-mouth parameter for nuclear energy was also found by Guidolin and Guseo (2016) when studying the possible effects of the energy transition that is going on in Germany.

20

ethanol vehicles was entirely driven by innovation. The flex vehicle market has a higher diffusion speed

than gasoline, but the difference is not statistically significant (qF – qG = 0.15, p = 0.47).

21

Table 1. Parameter Estimates of the multigeneration model

Parameters Estimate Std Error t Value Pr > |t|

Gasoline

Adopters

Innovation pG 0.029 0.011 2.74 0.008

Word-of-Mouth qG 0.876 0.154 5.70 <.001

Market Potential mG 33.531 4.613 7.27 <.001

Tax Coefficient βG -0.042 0.004 -9.98 <.001

Ethanol

Adopters

Innovation pE 1.057 0.228 4.63 <.001

Word-of-Mouth qE -5.316 2.768 -1.92 0.060

Market Potential mE -0.023 0.036 -0.63 0.530

Tax Coefficient βE -0.048 0.013 -3.77 0.004

Flex

Adopters

Innovation pF 0.003 0.001 2.32 0.024

Word-of-Mouth qF 1.022 0.154 6.64 <.001

Market Potential mF 25.927 0.585 44.33 <.001

Tax Coefficient βF -0.012 0.016 -0.77 0.443

Upgraders Gasoline to Ethanol q¿ 0.541 0.313 1.73 0.090

Gasoline to Flex qGF 0.075 0.032 2.33 0.024

Ethanol to Flex qEF -0.071 0.097 -0.74 0.464

Considering the effect of taxation, represented by the marketing effort parameter, βs, we see that

the effect of the tax rate is significantly negative for gasoline (βG = -0.042, p < 0.001) and ethanol

vehicles (βE = -0.048, p = 0.004). Thus, for these two generations, the effect of taxation is to stifle

adoption. However, for flex vehicles, the effect of taxation is not significant, indicating that the stifling

effect of taxation has been reduced or virtually removed. Finally, considering the parameters of the

22

upgrading processes, we see a positive and significant (leapfrogging) upgrading from gasoline to flex

vehicles (qGF = 0.075, p = 0.024), a marginally significant upgrading from gasoline to ethanol cars (qGE =

0.541, p = 0.09). There is no significant evidence of upgrading from ethanol to flex vehicles. The

estimated values of the 15 parameters in this multi-generational model are not independent of each other -

in the appendix in Panel A, Table A.1, we show the correlations between parameter estimates. Although

the majority of correlations are low, showing a lack of dependence between parameter estimates, there are

some exceptions. For the adoption of ethanol vehicles, we see a high negative correlation between the

coefficients of innovation and imitation. This indicates that it is difficult to separate the effects of these

two processes. The estimate of the potential for gasoline cars is highly correlated with the adoption

parameters of ethanol, pE, qE, βE and the upgrading parameter from gasoline to ethanol, qGE. The difficulty

experienced by the model in separating the potential of the gasoline vehicle from the adoption of ethanol

vehicles is reflected in the negative correlation between the residuals of the model for gasoline and

ethanol vehicles, shown in the appendix in Panel B, Table A.1.

The estimates of the coefficients of imitation in our study, qG = 0.88, qF = 1.02, are very high

compared to the average of Sultan et al.'s (1990) meta-analysis result of q = 0.38. This accords with the

extant literature ((Tellis et al., 2003; Van den Bulte and Stremersch, 2004) which suggests that a country

like Brazil (collective society, high on power distance and uncertainty avoidance) will have above

average imitation coefficients. Nardon and Aten's (2008) study provides an insightful interpretation of

the high innovation coefficient for ethanol ( pE), arguing that the development of the Alcohol Program is a

manifestation of the Brazilian cultural characteristic of being able to quickly adapt to adversity and

having the flexibility to incorporate sudden change. The impact of institutions on diffusion is captured by

the negative tax coefficients on the probability of adoptions, i.e. lowering tax increases the probability of

adoption.

6. Conclusions and Policy Recommendations

23

This article has adapted a multi-generation innovation diffusion model to evaluate the impact

of governmental incentives on the diffusion of three vehicular technologies in Brazil. Two modifications

were made. ‘Leapfrogging’ was introduced to capture the frequently observed phenomenon of an adopter

of gasoline switching to flex technology, skipping the ethanol technology. A marketing factor was

introduced to the multigenerational model to capture the effect of the tax/incentive scheme, the main

instrument used by the Brazilian government’s Alcohol Program. The literature-based assumption that tax

exemptions can increase the diffusion rates of technologies is tested. Our results reinforce this narrative

by showing the negative effect of taxes on adoptions of gasoline and ethanol vehicles but not for flex

vehicles in Brazil. The markets for the first generation of vehicle fuel technology, gasoline, and the third

generation, flex vehicles, behave as smooth, well behaved, S-shaped diffusion models.

The ethanol vehicle’s experience demonstrates that even though incentives are necessary for

the beginning, in the long run, some level of self-sustainability must be achieved. Our results showed that

ethanol vehicle adoptions were mainly due to innovation by new adopters from a pool of insignificant

magnitude, with some upgrading from gasoline. The model captured the consumers' loss of confidence in

ethanol vehicles, evidenced by a negative word-of-mouth (Moutinho dos Santos and Parente, 2006).

The flex vehicle technology represents, as both data and model suggest, the mainstream

technology in the country. Although still consuming some gasoline, flex vehicles are the main current

strategy for Brazil to achieve its CO2 reduction goals in transportation. Since the diffusion of flex vehicles

occurred predominantly by positive word-of-mouth, it seems that, regardless of any setback in the ethanol

industry, flex vehicles will remain the dominant mainstream technology.

Our results have policy implications for countries wishing to introduce a new generation of

vehicle technology. Tax reduction is necessary for the first steps to the diffusion process; however,

policymakers should not only rely on these incentives. Since cars are durable goods, word-of-mouth plays

an important role in the long run. For flex vehicles, the versatility of being able to choose which fuel is

preferred can be considered the main relative advantage.

24

Fuel price stability is a marketing variable that is likely to have an impact on the continued

adoption or discontinuance of ethanol technology, and it deserves further studies. In addition, as potential

future research, there is the possibility of a 4th technological generation of hybrid-electric flex cars in the

Brazilian vehicle market. Toyota indicates that its first model will start being produced in Brazil by the

end of 2019 (Muniz and Gasques, 2018). Such technological innovation that allows the use of electricity

for vehicle traction would represent an efficiency improvement over both flex and current hybrid electric

vehicles.

References

Aguirre, M., Ibikunle, G., 2014. Determinants of renewable energy growth : A global sample analysis.

Energy Policy 69, 374–384. https://doi.org/10.1016/j.enpol.2014.02.036

Al-Alawi, B.M., Bradley, T.H., 2013. Review of hybrid, plug-in hybrid, and electric vehicle market

modeling Studies. Renew. Sustain. Energy Rev. 21, 190–203.

https://doi.org/10.1016/j.rser.2012.12.048

Andersen, F.M., Larsen, H. V., Skovgaard, M., 2007. A European model for the number of End-of-Life

Vehicles. Int. J. Automot. Technol. Manag. 7, 343–355.

https://doi.org/10.1504/IJATM.2007.017065

ANFAVEA, 2018. Brazilian Automotive Industry Yearbook. Brazilian Automotive Industry Association,

São Paulo and Brasilia.

ANFAVEA, 1994. Brazilian Automotive Industry Yearbook. Brazilian Automotive Industry Association,

São Paulo and Brasilia.

Baran, R., Legey, L.F.L., 2013. The introduction of electric vehicles in Brazil: Impacts on oil and

electricity consumption. Technol. Forecast. Soc. Change 80, 907–917.

https://doi.org/10.1016/j.techfore.2012.10.024

Bass, F., 1969. A New Product Growth for Model Consumer Durables. Manage. Sci. 15, 215–227.

25

https://doi.org/10.1287/mnsc.15.5.215

Bass, F.M., Krishnan, T. V, Jain, D.C., 1994. Why the Bass Model Fits without Decision Variables.

Mark. Sci. 33, 203–223. https://doi.org/10.1287/mksc.13.3.203

Bastian-Pinto, C., Brandão, L., de Lemos Alves, M., 2010. Valuing the switching flexibility of the

ethanol-gas flex fuel car. Ann. Oper. Res. 176, 333–348. https://doi.org/10.1007/s10479-009-0514-7

Bastin, C., Szklo, A., Rosa, L.P., 2010. Diffusion of new automotive technologies for improving energy

efficiency in Brazil’s light vehicle fleet. Energy Policy 38, 3586–3597.

https://doi.org/10.1016/j.enpol.2010.02.036

Benvenutti, L.M.M., Ribeiro, A.B., Forcellini, F.A., Maldonado, M.U., 2016. The Effectiveness of Tax

Incentive Policies in the Diffusion of Electric and Hybrid Cars in Brazil, in: 41o Congresso

Latinoamericano de Dinamica de Sistemas. São Paulo.

Benvenutti, L.M.M., Ribeiro, A.B., Uriona, M., 2017. Long term diffusion dynamics of alternative fuel

vehicles in Brazil. J. Clean. Prod. 164, 1571–1585. https://doi.org/10.1016/j.jclepro.2017.07.051

Brito, T., Moutinho dos Santos, E., Rousseau, I., Nava, P.C., 2012. The Dialectics of Energy Security and

Nations Interdependence: Reflections Focused on the Role of Oil and the Brazilian Dimension, in:

Monié, F., Binsztok, J. (Eds.), Petroleum Geography and Geopolitics. Mauad X, Rio de Janeiro, pp.

15–50.

Brito, T.L.F., Moutinho dos Santos, E., Galbieri, R., Costa, H.K.D.M., 2017. Qualitative Comparative

Analysis of cities that introduced compressed natural gas to their urban bus fleet. Renew. Sustain.

Energy Rev. 71. https://doi.org/10.1016/j.rser.2016.12.077

Cao, X., Mokhtarian, P.L., 2004. The Future Demand for Alternative Fuel Passenger Vehicles: A

Diffusion of Innovation Approach. The California Department of Transportation, Sacramento.

CETESB, 2017. Vehicle Emissions in São Paulo State 2016. São Paulo State Environmental Company,

São Paulo.

Coelho, S.T., Goldemberg, J., Lucon, O., Guardabassi, P., 2006. Brazilian sugarcane ethanol: lessons

learned. Energy Sustain. Dev. 10, 26–39. https://doi.org/10.1016/S0973-0826(08)60529-3

26

Collantes, G.O., 2007. Incorporating stakeholders ’ perspectives into models of new technology

diffusion : The case of fuel-cell vehicles 74, 267–280. https://doi.org/10.1016/j.techfore.2006.02.001

Cordellini, D.F., 2018. Ethanol on biofuels structure in Brazil. FAE J. 21, 19–35.

Danaher, P.J., Hardie, B.G.S., Putsis, W.P., 2001. Marketing-Mix Variables and the Diffusion of

Successive Generations of a Technological Innovation. J. Mark. Res. 38, 501–514.

https://doi.org/10.1509/jmkr.38.4.501.18907

DiMaggio, P.J., Powell, W.W., 1983. The Iron Cage Revisited : Institutional Isomorphism and Collective

Rationality in Organizational Fields. Am. Sociol. Rev. 48, 147–160.

Duarte, R.G., Rodrigues, S.B., 2017. Co-evolution of industry strategies and government policies: The

case of the brazilian automotive industry. BAR - Brazilian Adm. Rev. 14, 1–28.

https://doi.org/10.1590/1807-7692bar2017160100

Egbue, O., Long, S., 2012. Barriers to widespread adoption of electric vehicles: An analysis of consumer

attitudes and perceptions. Energy Policy 48, 717–729. https://doi.org/10.1016/j.enpol.2012.06.009

Ferreira, A., Kunh, S.S., Fagnani, K.C., De Souza, T.A., Tonezer, C., Dos Santos, G.R., Coimbra-Araújo,

C.H., 2018. Economic overview of the use and production of photovoltaic solar energy in Brazil.

Renew. Sustain. Energy Rev. 81, 181–191. https://doi.org/10.1016/j.rser.2017.06.102

Freitas, I.M.B., Dantas, E., Iizuka, M., 2012. The Kyoto mechanisms and the diffusion of renewable

energy technologies in the BRICS. Energy Policy 42, 118–128.


Furtado, A.T., Scandiffio, M.I.G., Cortez, L.A.B., 2011. The Brazilian sugarcane innovation system.


Gnann, T., Plötz, P., 2015. A review of combined models for market diffusion of alternative fuel vehicles

and their refueling infrastructure. Renew. Sustain. Energy Rev. 47, 783–793.


Goldemberg, J., 2007. Ethanol for a Sustainable Energy Future. Science (80-. ). 315, 808–810.

https://doi.org/10.1126/science.1137013

27

Goldemberg, J., Coelho, S.T., Guardabassi, P., 2008. The sustainability of ethanol production from

sugarcane. Energy Policy 36, 2086–2097. https://doi.org/10.1016/j.enpol.2008.02.028

Goldemberg, J., Coelho, S.T., Nastari, P.M., Lucon, O., 2004. Ethanol learning curve - The Brazilian

experience. Biomass and Bioenergy 26, 301–304. https://doi.org/10.1016/S0961-9534(03)00125-9

Goldemberg, J., Schaeffer, R., Szklo, A., Lucchesi, R., 2014. Oil and natural gas prospects in South

America: Can the petroleum industry pave the way for renewables in Brazil? Energy Policy 64, 58–

70. https://doi.org/10.1016/j.enpol.2013.05.064

Guidolin, M., Guseo, R., 2016. The German energy transition : Modeling competition and substitution

between nuclear power and Renewable Energy Technologies. Renew. Sustain. Energy Rev. 60,

1498–1504. https://doi.org/10.1016/j.rser.2016.03.022

Guidolin, M., Mortarino, C., 2010. Cross-country diffusion of photovoltaic systems: Modelling choices

and forecasts for national adoption patterns. Technol. Forecast. Soc. Change 77, 279–296.


Guler, I., Guillén, M.F., Macpherson, J.M., Guillen, M.F., 2002. Global Competition, Institutions, and the

Diffusion of Organizational Practices: The International Spread of ISO 9000 Quality Certificates.

Adm. Sci. Q. 47, 207–232. https://doi.org/10.2307/3094804

Hofstede’s Insights, 2019. Country Comparison [WWW Document]. URL https://www.hofstede-

insights.com/country-comparison/brazil/ (accessed 7.10.19).

Hofstede, G., 2001. Culture’s Consequences: Comparing Values, Behaviors, Institutions and

Organizations Across Nations, 2nd ed. Sage, Thousand Oaks, CA.

Ibenholt, K., 2002. Explaining learning curves for wind power. Energy Policy 30, 1181–1189.

IBGE, 1988. National Survey by Household Sample (PNAD). Rio de Janeiro.

IPCC, 2014. Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to

the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, Working Group III

Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change.

Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

28

Isabella, G., Yu, A.S.O., Silva, A.M. Da, Pegetti, A.L., 2017. Another driver of the Brazilian fuel ethanol

supply chain: the consumers’ preferences. Rev. Adm. 52, 304–316.

https://doi.org/10.1016/j.rausp.2017.05.003

Islam, T., Meade, N., 1997. The Diffusion of Successive Generations of a Technology: A More General

Model. Technol. Forecast. Soc. Change 56, 49–60.

Islam, T., Meade, N., 1996. Forecasting the development of the market for business telephones in the UK.

J. Oper. Res. Soc. 47, 906–918. https://doi.org/10.1057/jors.1996.116

Jenn, A., Springel, K., Gopal, A.R., 2018. Effectiveness of electric vehicle incentives in the United States.


Jeon, S.Y., 2010. Hybrid & Electric vehicle technology and its market feasibility. Massachusetts Institute

of Technology.

Jiang, Z., Jain, D.C., 2012. A Generalized Norton–Bass Model for Multigeneration Diffusion. Manage.

Sci. 58, 1887–1897. https://doi.org/10.1287/mnsc.1120.1529

Jun, D. BIN, Park, Y.S., 1999. A Choice-Based Diffusion Model for Multiple Generations of Products.

Technol. Forecast. Soc. Chang. 58, 45–58.

Kim, K.H., Shon, Z.H., 2011. Long-term changes in PM10 levels in urban air in relation with air quality

control efforts. Atmos. Environ. 45, 3309–3317. https://doi.org/10.1016/j.atmosenv.2011.03.033

King, J.L., Gurbaxani, V., Kraemer, K.L., McFarlan, F.W., Raman, K.S., Yap, C.S., 1994. Institutional

Factors in Information Technology Innovation. Inf. Syst. Res. 5, 139–169.

Kloess, M., Müller, A., 2011. Simulating the impact of policy, energy prices and technological progress

on the passenger car fleet in Austria-A model based analysis 2010-2050. Energy Policy 39, 5045–

5062. https://doi.org/10.1016/j.enpol.2011.06.008

Langbroek, J.H.M., Franklin, J.P., Susilo, Y.O., 2016. The effect of policy incentives on electric vehicle

adoption. Energy Policy 94, 94–103. https://doi.org/10.1016/j.enpol.2016.03.050

Leibowicz, B.D., 2018. Policy recommendations for a transition to sustainable mobility based on

historical diffusion dynamics of transport systems. Energy Policy 119, 357–366.

29


Lieven, T., Rietmann, N., 2018. Do policy measures in fact promote electric mobility? A study across 20

countries. Int. J. Transp. Dev. Integr. 2, 39–48. https://doi.org/10.2495/TDI-V2-N1-39-48

Mahajan, V., Muller, E., Kerin, R.A., 1984. Introduction Strategy for New Products with Positive and

Negative Word-of-Mouth. Manag. Sci. 30, 1389–1404.

Masini, A., Frankl, P., 2002. Forecasting the diffusion of photovoltaic systems in southern Europe : A

learning curve approach. Technol. Forecast. Soc. Chang. 70, 39–65.

Massiani, J., Gohs, A., 2015. The choice of Bass model coefficients to forecast diffusion for innovative

products: An empirical investigation for new automotive technologies. Res. Transp. Econ. 50, 17–

28. https://doi.org/10.1016/j.retrec.2015.06.003

Meade, N., Islam, T., 2015. Modelling European usage of renewable energy technologies for electricity

generation. Technol. Forecast. Soc. Change 90, 497–509.


Meade, N., Islam, T., 2006. Modelling and forecasting the diffusion of innovation – A 25-year review.

Int. J. Fore 22, 519–545. https://doi.org/10.1016/j.ijforecast.2006.01.005

Mesquita, D.L., Borges, A.F., Sugano, J.Y., Santos, A.C. dos, 2013. O desenvolvimento de processos de

inovação sob a ótica da teoria dos custos de transação: o caso da tecnologia flex-fuel. Rev. Adm. e

Inovação 10, 119–140. https://doi.org/10.5773/rai.v10i1.1081

Meyer, P.E., Winebrake, J.J., 2009. Modeling technology diffusion of complementary goods: The case of

hydrogen vehicles and refueling infrastructure. Technovation 29, 77–91.

https://doi.org/10.1016/j.technovation.2008.05.004

MF, 2018. Table of Incidence of the Tax over Product Industrialized (TIPI). Brazilian Ministry of

Finance, Brasilia,.

MMA, 2014. 2nd National Inventory of Atmospheric Emissions by Motor Road Vehicles. Ministry of

Environment, Brasilia, Brazil.

Moreira, J.R., Goldemberg, J., 1999. The alcohol program. Energy Policy 27, 229–245.

30

https://doi.org/10.1016/S0301-4215(99)00005-1

Moreira, J.R., Nogueira, L.A.H., Parente, V., 2005. Biofuels for Transport, Development, and Climate

Change: Lessons from Brazil, in: Bradley, R., Baumert, K.A. (Eds.), Growing in the Greenhouse:

Protecting the Climate by Putting Development First. World Resources Institute, p. 132.

Mouette, D., Machado, P.G., Fraga, D., Peyerl, D., Borges, R.R., Brito, T.L.F., Shimomaebara, L.A.,

Moutinho dos Santos, E., 2019. Costs and emissions assessment of a Blue Corridor in a Brazilian

reality: The use of liquefied natural gas in the transport sector. Sci. Total Environ. 668, 1104–1116.

https://doi.org/10.1016/j.scitotenv.2019.02.255

Moutinho dos Santos, E., Parente, V., 2006. The Second Alcohol Shock in Brazil - Looking for More

Flexibility, in: International Association of Energy Economics. Potsdam.

Muniz, G.B., Gasques, M.V., 2018. É do Brasil: Toyota Confirma Produção do Primeiro Carro Híbrido

Flex do Mundo. Revista Autoesporte.

Muñoz, M., Oschmann, V., David Tàbara, J., 2007. Harmonization of renewable electricity feed-in laws

in the European Union. Energy Policy 35, 3104–3114. https://doi.org/10.1016/j.enpol.2006.11.006

Nardon, L., Aten, K., 2008. Beyond a better mousetrap: A cultural analysis of the adoption of ethanol in

Brazil. J. World Bus. 43, 261–273. https://doi.org/10.1016/j.jwb.2008.03.008

Nastari, P.M., 1983. The role of sugar cane in Brazil’s history and economy. Iowa State University.

Negro, S.O., Alkemade, F., Hekkert, M.P., 2012. Why does renewable energy diffuse so slowly? A

review of innovation system problems. Renew. Sustain. Energy Rev. 16, 3836–3846.


Norton, J.A., Bass, F.M., 1987. A Diffusion Theory Model of Adoption and Substitution for Successive

Generations of High-Technology Products. Manage. Sci. 33, 1069–1086.

Pae, J.H., Lehmann, D.R., 2003. Multigeneration innovation diffusion: The impact of intergeneration

time. J. Acad. Mark. Sci. 31, 36–45. https://doi.org/10.1177/0092070302238601

Pefley, R.K., Pullman, J.B., Suga, T.P., Espinola, S., 1980. Feedback controlled fuel injection system can

accommodate any alcohol-gasoline blend, in: International Symposium on Alcohol Fuels

31

Technology. São Paulo, pp. 247–253.

Peter, R., 2002. Conceptual model for marketing solar based technology to developing countries. Renew.

Energy 25, 511–524.

Pfeiffer, B., Mulder, P., 2013. Explaining the diffusion of renewable energy technology in developing

countries. Energy Econ. 40, 285–296. https://doi.org/10.1016/j.eneco.2013.07.005

Pinto, J.T.M., Amaral, K.J., Janissek, P.R., 2016. Deployment of photovoltaics in Brazil: Scenarios,

perspectives and policies for low-income housing. Sol. Energy 133, 73–84.

https://doi.org/10.1016/j.solener.2016.03.048

Puglieri, R., 2013. Analysis of the Sugar-Alcohol Industry Under the Optics of the Competition Defense:

How to Define Relevant Markets. University of São Paulo.

Rao, K.U., Kishore, V.V.N., 2010. A review of technology diffusion models with special reference to

renewable energy technologies. Renew. Sustain. Energy Rev. 14, 1070–1078.


Rao, K.U., Kishore, V.V.N., 2009. Wind power technology diffusion analysis in selected states of India.

Renew. Energy 34, 983–988. https://doi.org/10.1016/j.renene.2008.08.013

Rogers, E.M., 2003. Diffusion of Innovations, 5th ed. Free Press, New York.

Rosenzweig, P.M., Singh, J. V., 1991. Organizational Environments and the Multinational Enterprise.

Acadamy Manag. Rev. 16, 340–361.

Shi, X., Fernandes, K., Chumnumpan, P., 2014. Diffusion of multi-generational high-technology

products. Technovation 34, 162–176. https://doi.org/10.1016/j.technovation.2013.11.008

Sierzchula, W., Bakker, S., Maat, K., Van Wee, B., 2014. The influence of financial incentives and other

socio-economic factors on electric vehicle adoption. Energy Policy 68, 183–194.


Speece, M.W., Maclachlan, D.L., 1992. Forecasting Fluid Milk Package Type with a Diffusion Model.

IEEE Trans. Eng. Manag. 39.

Stremersch, S., Muller, E., Peres, R., 2010. Does new product growth accelerate across technology

32

generations? Mark. Lett. 21, 103–120. https://doi.org/10.1007/s11002-009-9095-0

Struben, J., Sterman, J.D., 2008. Transition challenges for alternative fuel vehicle and transportation

systems. Environ. Plan. B Plan. Des. 35, 1070–1098. https://doi.org/10.1068/b33022t

Sultan, F., Farley, J.U., Lehmann, D.R., 1990. A Meta-Analysis of Applications of Diffusion Models. J.

Mark. Res. 27, 70. https://doi.org/10.2307/3172552

Teixeira, E.C., 2005. The Development of Flex-Fuel Technology in Brazil. Instituto DNA Brasil, São

Paulo.

Tellis, G.J., Stremersch, S., Yin, E., 2003. The International Takeoff of New Products: The Role of

Economics, Culture, and Country Innovativeness. Mark. Sci. 22, 188–208.

https://doi.org/10.1287/mksc.22.2.188.16041

UNFCCC, 2015. Adoption of the Paris Agreement - Conference of the Parties COP 21. Proposal by the

President. United Nations Framework Convention on Climate Change.

https://doi.org/FCCC/CP/2015/L.9/Rev.1

Van den Bulte, C., Stremersch, S., 2004. Social Contagion and Income Heterogeneity in New Product

Diffusion: A Meta-Analytic Test. Mark. Sci. 23, 530–544. https://doi.org/10.1287/mksc.1040.0054

Versluis, C., 2002. DRAMs, fiber and energy compared with three models of market penetration.

Technol. Forecast. Soc. Change 69, 263–286.

Wansart, J., Schnieder, E., 2010. Modeling market development of electric vehicles. 2010 IEEE Int. Syst.

Conf. 371–376. https://doi.org/10.1109/SYSTEMS.2010.5482453

Yu, A.S.O., Nascimento, P.T.D.S., Nigro, F.E.B., Frederick, B.W.B., Varandas, J.A., Vieira, S.F.A.,

Rocha, R.L., 2010. The evolution of flex-fuel technology in Brazil: The Bosch case. Technol.

Manag. Glob. Econ. Growth (PICMET), 2010 Proc. PICMET ’10 1250–1260.

Zachariadis, T., Samaras, Z., Zierock, K.H., 1995. Dynamic modeling of vehicle populations: An

engineering approach for emissions calculations. Technol. Forecast. Soc. Change 50, 135–149.

https://doi.org/10.1016/0040-1625(95)00057-H

Zhu, Y., Tokimatsu, K., Matsumoto, M., 2015. A Diffusion model for Natural Gas Vehicle : A case study

33

in Japan. Energy Procedia 75, 2987–2992. https://doi.org/10.1016/j.egypro.2015.07.607

34

Acknowledgements

We gratefully acknowledge the support of the RCGI – Research Centre for Gas Innovation, hosted by the

University of São Paulo (USP) and sponsored by FAPESP – São Paulo Research Foundation

(2014/50279-4) and Shell Brasil. This study was financed in part by the Coordenação de

Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. This research was

conducted when Thiago was a visiting doctoral student at the Department of Marketing and Consumer

Studies, University of Guelph, Canada.

35

Appendix

Table A1. Correlation matrix of parameter estimates and residuals

Panel A: Correlation Matrix of Parameter Estimates

pG qG pE qE pF qF mG mE mF βG βE βF q¿ qGF qEF

pG 1.00

qG -0.19 1.00

pE 0.40 0.21 1.00

qE -0.42 -0.11 -0.90 1.00

pF -0.03 -0.07 -0.29 0.18 1.00

qF 0.22 0.13 0.42 -0.45 0.20 1.00

mG 0.32 0.21 0.80 -0.85 0.02 0.57 1.00

mE -0.47 0.27 -0.11 0.21 -0.09 -0.18 -0.28 1.00

mF 0.32 0.15 0.11 -0.14 0.15 0.17 0.20 -0.17 1.00

βG -0.18 -0.74 -0.17 0.03 0.08 -0.01 0.02 -0.24 -0.17 1.00

βE 0.04 0.17 0.31 -0.65 -0.02 0.33 0.63 -0.19 0.12 0.17 1.00

βF -0.01 0.04 0.10 -0.07 -0.37 -0.74 -0.01 0.04 -0.31 -0.04 0.04 1.00

q¿ -0.26 0.41 0.58 -0.61 -0.04 0.44 0.82 0.01 0.04 0.07 0.62 0.02 1.00

qGF -0.57 -0.28 -0.11 0.12 -0.01 -0.15 -0.14 0.23 -0.50 0.37 -0.08 0.00 0.12 1.00

36

qEF 0.29 -0.18 -0.39 0.46 -0.26 -0.29 -0.47 0.00 0.12 -0.13 -0.41 0.08 -0.61 -0.44 1.00

Panel B: Correlation Matrix of Residuals

Gasoline Ethanol Flex

Gasoline 1.00

Ethanol -0.68 1.00

Flex 0.00 -0.12 1.00

37

Documents

Modelling and forecasting multi-generation, multi-country ... · Web viewIslam and Meade (1997) argue that after the radical innovation of the first generation of a technology, in